"The Macroprudential Toolkit: Measurement and Analysis" - 12/2/2011 Afternoon

Uploaded by USTreasGov on 19.12.2011

>>> It's now time to move to our next event.
I have the high privilege to introduce our lunch speaker
To most people in this audience Don doesn't need introduction,
so I'll highlight a few really incredible things in his career.
Don had a very distinguished career at the fed.
First as an economist on the staff, then as a director, the
first director of the division of monetary affairs, and then he
became a member of the board of governors.
He was appointed to be vice-chairman of the board of
governors in 2006.
He became an invaluable adviser and confidant to many chairs of
the fed, most recently chairman Greenspan and chairman Bernanke.
Don is scholar and policymaker, and he has written many payments
as vice chair, you can find papers in his VITAE.
He has remarkable insight in the financial markets and has a
passion from learning from all those around him.
So just a little on a personal note, Don has been a remarkable
mentor and friend to many of us over the years at the board.
In some of the most intense and perhaps scariest moments of the
financial crisis, Don was the steady anchor and we turned to
him and we turned to him often.
Not just for his technical expertise, but often for his
personal guidance.
He became the fortunate recipient of many phone calls
and e-mails from people just saying, I can't take it anymore.
I'm leaving.
He would just talk us down and sometimes step in and say, I'll
take it from here.
So since leaving the fed, Don is a senior fellow at the brookings
institute and he's serving on the bank of England's policy
committee to implement macroPrudential policies in
We're extremely fortunate to hear from him today.
>> Thank you.
Thank, Nelly and Dick for inviting me here.
It's great to see many, many old friends and colleagues here.
I can't possibly live up to Nelly's introduction, but I will
I'll talk some about the financial policy committee.
This is an important and interesting aspect of my
post-federal reserve life.
My membership as an external member on the interim financial
policy committee at the bank of England.
I thought you might find it interesting -- this audience
might find it interesting to hear a little about the
implementation of macroPrudential policy in the
United Kingdom.
We issued our third set of recommendations just yesterday.
This seems like a good time to discuss our efforts to identify
and deal with threats to financial and economic stability
in the United Kingdom.
I was kind of tempted to entitle this a Pennsylvania Yankee and
Reuben's court.
That's sort of what it feels like.
Let me begin with some background on the establishment
of the interim financial policy committee.
It's an interim committee because the law hasn't been
passed yet to officially establish it but it's up and
running in many respects.
official in the UK saw some room for improvement, as my
elementary school report cards used to say, in the performance
of various authorities responsible for financial
Both in the buildup of vulnerabilities and they saw it
partly reflecting structural issues, in particular the
separation of bank supervision from the central bank, which
occurred in 1997 when the bank of England was given
independence to pursue that inflation target seemed to elm
bead the efficacy of supervisor and end of crisis response.
The bank retained some generalized responsibility and
published a much admired financial stability report, but
it did not directly participate in formulating and implementing
supervisory policies to establish and protect financial
So it did not itself have the levers to carry out its role
The most obvious failing was no single institution had the
responsibility, authority or powers to monitor the system as
a whole, identify potentially destabilizing trends and respond
to them with concerted action.
Shortly after the government took office in the spring of
2010, the new government did, it published a discussion draft of
a new structure that moved the regulation supervision of
individual firms back in to the bank of England.
Meanwhile, the regulation of conduct within the financial
system includes the conduct of firms toward retail customers
and participants and wholesale markets is to be KAERTed out by
ate special body to focus responsibility and
accountability, the government has proposed a new subsidiary of
the bank, the Prudential regulatory authority, the PRA,
that will concentrate on microPrudential oversight while
a new financial policy committee will provide for the dedicated
macroPrudential overlay.
The law to implement this new structure is still under
discussion, but the in the meantime the government has
established an interim financial policy committee, and it's given
us two charges.
First, identify risk to financial stability in the
United Kingdom, and make recommendations to deal with
those risks.
Most of our recommendations are to the financial institutions
and to the FSA, the final services authority, which still
has the responsibility for oversight of financial
Both risk identification and recommendations are included in
the financial stability report, which is now the responsibility
of the FPC and was published yesterday.
Second, we are to make recommendations to the
government about what tools we need in the new law to do or job
popely and properly.
Ultimately we're to have two kinds of powers with respect to
tools first, to make recommendations to the
authorities where the FPC believes that specific
regulatory actions are required in order to protect financial
stability, and these recommendations are subject to
consideration by the relevant authority, then to the usual
rule-making process if the authority decides to go forward.
Second, if we see the threat to financial stability as being
immediate and serious, we will be able to issue directives to
the microPrudential authorities on regulatory tools that should
be deployed in the pursuit of macroPrudential policy.
The interim FTC has 11 members including four externals serving
on a part-time basis.
It's chaired by the governor.
There's an observer from the treasury who also attends and
sits in on the discussions.
The discussions have been vigorous and good.
A wide range of views are solicited and expressed about
the risks of the financial system and what can be done to
get rid of them.
Meetings tend to focus on narrowing the number of issues
and identifying the most effective responses.
We greatly benefit from the range of backgrounds that
members bring to the task.
Two members come from the FSA.
Several are from the financial stability side of the bank of
England with their Macco financial PERGS active.
Importantly, two of my fellow externals have quite extensive
experience in the London financial markets.
They have utilized this to bring issues to our attention and to
inform many of rest of us without such experience how our
concerns are playing out in the real world.
I'm going to concentrate on the leg of our mandate that asks us
to identify risks and make recommendations.
Under risk identification now, unfortunately in one sense this
hasn't been difficult.
The UK is part of Europe, even if it's not part of the euro
Its banks and businesses are exposed to developments in the
ongoing saga of the fiscal banking and business stresses so
evident there.
we've met and our first meeting was in June, we've identified
exposure to these developments as the primary systemic risk to
the UK financial system.
We are also building on a risk identification structure already
in place at the Bank of England.
Staff provide extensive briefing to FPC members in advance of
They include information and analysis on the macrofinancial
environment and short and medium-term risks to financial
Both in the UK and abroad in those markets in which UK banks
are very active.
A key aspect of that analysis is the ability of UK banks to
withstand potential, adverse developments.
We review many different met Fricks and model outputs within
this process.
In addition, the bank has an active market intelligence
on what's going on in markets and the views the market's
participants about emerging risk.
Of particular value is that members bring their own
knowledge and understanding of the UK financial economic system
to the meetings.
For myself I make it a practice each time I go to London to meet
with people in the financial sector there so I can learn
first hand what stability issues are on their minds.
With very few exceptions or recommendations have fallen FWU
one of two broad categories.
One category EMbodies getting additional information for
ourselves and for the public that we believe is necessary for
the FPC and the markets to monitor and take actions to
contain risks to financial stability.
The second category has been a series of recommendations that
build additional resilience into impairing its willingness and
ability to perform key intermediary functions and in
particular its ability and LGSness to supply credit to UK
households and small and medium enterprises.
On the information side, we've identified several practices
that raise questions in our mind about stability but for which we
didn't know enough to make a judgment.
We ask the FSA to gather additional information.
One such practice involved complex funding structures
including those associated with synthetic ETFs and collateral
We were concerned about a he potential build-up in
interconnections that were difficult to understand, opaque
to investors and counterparties and therefore could contribute
to mispriced risk and instability in the stress
Now, the FSA looking into it found that British banks were
not active in this area, but it will be important to monitor
this sort of complex interconnection going forward,
to assure that the kinds of complex opaque funding
structures that contributed to the buildup of risks and proved
so fragile in the crisis DOONTS re-emerge.
We also asked FSA to look into loan forebearance and associated
I thought of this as the extend and pretend information
By allowing both borrowers and lenders additional flexibility
to work through debt obligations during temporary periods of
stress, forebearance can be a stabilizing thing for markets.
Loans subject the to forebearance are inevitably
subject to greater risk than loans that did not need special
If they're not adequately provisioned for, banks are in
might appear from the published capital ratios and balance
The FSA had already looked into forbearance in residential real
estate, but we and it be extended to UK bank, households
and corporate sectors on a global results.
Initial results for the UK commercial real estate sector
indicated some underprovisioning but not enough to stuff a
systemic risk.
Under a risk-based approach more work will be done on global
exposures in the future.
We also ask that banks publish additional information so athat
private parties can better judge the safety and soundness of UK
bank counterparts increasing the role of market discipline and
maintaining financial stabilization.
UK banks are not as transparent as those in the United States.
For example, complete reports are published semi-annually SXNT
quarterly, and the published reports are just for one day,
want end of the quarter, the end of the half-year, with very
little information about intraperiod balance sheets.
Very subject to window dressing.
One recommendation to the FSA is to work with the British bankers
association to increase the trains parent see of UK banks
along a number of dimensions.
This is an ongoing project we'll be returning to periodically to
Last June was focused on the European sovereign exposures.
Under certainty among those extending credit to banks about
such exposures could undermine financial stability by inducing
a generalized reluctance to fund UK banks in response to the
threatening situation to the euro area.
the expiece you'res of individual UK banks were
released to public, and we asked the FSA to look at and public
LISH aggregate information on the he can pose you're of
BARNLGs below the largest tier, the very top tier.
The publication of the requested information did occur, and it
showed that euro area exposures are limited.
Just yesterday we recommended that the publication of leverage
ratios be accelerated from 2015, when they're required under
basil 3, to no later than the beginning of 2013.
Widely varying applications of risk weights distort comparisons
across banks and have the potential to reduce public
confidence in the information content of capital ratios at
that use risk-weighted assets and the denomindenominator.
Leverage ratios are imperfect met Ricks for resilience, but
they can be valuable supplements to supposedly more sophisticated
The sharp rise in such raise yos before a crisis at a number of
institutions both here in the United States, the United
Kingdom and continental Europe, were a warning sign of
We expect it will help parties evaluate risk, bolster market
discipline and providing greater incentives for UK banks.
The second category of recommendations has focused on
building bank resilience.
It's important that banks not try to bolster their own
individual strength by reducing the availability of credit to UK
businesses and households even in periods of stress.
Although a definitive parsing of weakness of credit growth
between supply and demand is not possible, credit availability is
seen as somewhat constrained in the United Kingdom, especially
for small, medium enterprises and further restraint would tend
to weaken an already slow recovery and feedback on the
banking system.
The best way to build resilience while maintaining lending is to
increase the level of capital and that's been the a main
objective of the financial policy committee.
In June and September we advised the banks to take advantage of
any opportunities to increase capital in periods of strong
profitability by retaining the earnings, not distributing them
in dividends and and bonuses and to reduce distributions when
earnings fell short.
We asked the FSA to work with banks on the follow-through this
When we met a week ago, we recognized that the threat from
the PROBS in the euro area ha intensified and so had the need
for an ample capital cushion.
At the same time a variety of forces weakened the growth of
bank earnings, which consequently had been inadequate
to build any additional capital cushion over the past couple
One of the effects of the increased threat relative to the
level of capital has been a rise in funding costs for UK banks
and a reduction in availability, especially at longer tenors.
They could feed through to the cost and availability of credit
to households and businesses.
Consequently, we strengthened our recommendation to the banks
to say that if earnings are insufficient to build capital
levels further, banks should limit distributions and give
serious consideration to raising external capital in coming
I was told serious consideration was British for we mean it.
That was reassuring.
The FSA will carry this message to the banks in coming weeks as
it reviews bank plans for compensation dividends and
assessing the adequacy of their capital plan.
In addition to being capital levels, banks can take steps to
manage their balance sheets to increase resilience without
reducing critical lending and other activities.
Yesterday they advised the FSA to work with the banks to
identify such opportunities, to strengthen balance sheet
resilience see without exacerbating market FRA Jill lit
or reducing lending to the economy.
On the liability side of of the balance sheet this is obtaining
funding wherever possibly and reducing reliance on short-term
sources or counterparties to become concerned about the
credit worthiness of the borrower and therefore begin to
hoard liquidity.
As we saw in 2008, even secured short-term funding is vulnerable
when confidence is undermined.
Among assets, financial sector claims could be a channel for
contagion when counterparties get in trouble for reductions
and some of the claims might free up capital to support the
At the same time we need to recognize that borrowing and
lending among financial institutions can perform
valuable economic functions in supporting the real economy --
in supporting real economy lending by smoothing the
availability of funds to lenders and redistributing from lenders
with a surplus to those facing stronger demands and more
productive goals for the funds.
They've identified for further examination the role of
different kinds of intrafinancial sector
transactionses and financial and economic stability.
Among other things the financial stability perspective and the
externalITIES may have different risk weights.
Let me say a few words about countercyclical macropolicy,
having experienced a little of it.
Much has been written SXAN liesed about the challenges of
KUKing macroPrudential in good times.
Taking away the punch bowl.
The earnings of capital of institutions will look strong
and the borrowers are in good financial shape even after they
increased their use of credit.
Asset mispricing will be difficult to identify and
overriding judgments and vested interests in the private sector
and at the intersection of private and public sectors will
be resistant to chain.
It's difficult to convince even dedicated and dispassionate
elected representatives that the good times can't go on forever,
or at least until after the next election.
It won't be easy in my view the widespread awareness of these
problems along with the memory of the very tough times we're
now experiencing because of past credit excesses will make it
reasonably likely that the authorities will do the right
thing under these circumstances.
That is the boom circumstances.
The other side of countercyclical policy easing
when times become tough confronts a very different and
perhaps more difficult set of challenges.
When the economy is weakening and bad debts are building, easy
requirements may reduce the effect of financial sector
problems on the real economy.
How do you know those problems won't get much worse?
Perhaps for reasons entirely external to the economy in
question and out of control of the authorities.
If conditions do continue to deteriorate substantially,
releasing capital and liquidity buffers, lowers he requirements
could come back to haunt the economy and the authorities if
it results in widespread failures, unemployment as credit
tightens and it requires fiscal intervention to stem the
downward slide.
The FTC has faced just such an issue as the situation in the
euro area worsened last summer and fall and the discussion of
the conflicting pressures is reflected in the record of our
September meeting, which we've already published.
The committee concluded at that time that lowering buffers would
not be appropriate out of concern of what might be coming
The cost of leaning unnecessarily hard against an
expanding credit and rising asset prices would be growth and
innovation foregone.
Very hard to see and measure.
The cost of adequate capital and liquidity is large and
A loss of confidence in an unstable financial system.
A risk neutral macroPRU dents shall regulator will find it
harder to take it away in good times than to spike it in bad
We need research in tough times.
Although the form -- although forms of macroPrudential
policies have been practiced for many years in a number of
countries, the consistent, systemic application of this
perspective to highly markets now undertaken in the UK, U.S.,
and other advanced economies is in its infancy.
I've drawn a couple of can conclusions from my experience
in this new endeavor.
First, it's no accident that many of the FTC's
recommendations have centered on information.
The authorities can't conduct macroPrudential policy, the
markets can't reinforce the efforts of the authorities
without detailed knowledge of key elements of the financial
markets and their interrelationships.
In that regard the work of the office of financial research and
the researchers right here at this room at this conference is
he essential to establish and preserve financial stability.
It is unnecessary but not sufficient as you know.
The information and analysis much feed into an open and
inclusive process that brings a wide variety of experience and
judgment to bear.
Second, building additional resilience into the system in
difficult times without impeding the flow of credit to households
and businesses is very hard.
It will be at just those times that resilience is most needed
but also when earnings are likely to be low and capital
markets expensive to access.
Surely the solution is to require high levels of capital
and liquidity -- excuse me.
Before trouble hits, so bad times are much less likely to
call the viability of financial institutions into question.
Unfortunately, the current threats to financial stability
are occurring while the transition to this stronger
regime is under way but not yet complete.
Third, countercyclical macroPrudential policy is
This is true in good times when it appears the system is strong
and there will be resistance to dampen the upswing, but it may
be even more difficult to allow or even encourage drawing down
capital and liquidity buffers in bad times to to reduce the
potential for tightening credit conditions to feed weakening
economic trends.
This it problem illustrates nicely the different
perspectives of macro and micro regulation.
From a micro perspective preserving capital and liquidity
strengthens the bank and helps to keep it from failing.
From a macro perspective bill ultimately weeken about.
Con grewance between macro and micro situations require
financial institutions have high levels of capital and liquidity
before real trouble hits.
Those are my prepared remarks.
Thank you.
>> We'll take some questions after I get some water.
>> Any questions?
>> Charles, he said with some hesitancy and fear.
>> Don, the most important regulator and proposals that
have come in the UK fora long time.
There's the independent commission on banking.
You didn't say a word about that.
Shouldn't the FTC make some public statement about what it
thinks will be the effect on the UK's financial stability of the
IBC proposals?
you were gone.
>> What did it say?
>> It said we are supportive of those, and we thought they
would -- and a variety -- a little background here.
There are a couple, I guess, two -- correct me if I'm missing
something here.
Two key pieces of this.
One is a the retail operations from the wholesale operations.
I've said that too quickly.
It's not clear what's inside the ring fence and outside the ring
But the idea here so it's not strictly glass or VOLGer or
something in between that.
But to keep a lot of the systemically important payments
stuff inside the ring fence and keep that extremely well
capitalized and separate from some of the riskier activities
would happen outside the ring fence.
The activities inside the ring fence, they say, would have very
The activities outside the ring fence would also have high
required capital levels but would build more on
international standards.
Those standards would be less of a constraint inside the ring
Putting them -- separating them not only, I think, will give the
authorities a better chance to resolve a troubled institution
in ways that won't result in financial instability -- well, I
think that's the main thing.
And it does enable the authorities to impose tighter
restrictions on the stuff inside the ring fence.
The second point is the one I've kind of been mentioning.
They're separated but they're very high levels of capital.
Higher than would be -- it's apparently higher than what will
The financial policy committee endorsed both of those things
and said that we thought they would contribute to financial
We hope that the parliament doesn't just leave it to the
regulator to figure out what is supposed to be inside the ring
fence and outside the ring fence.
It's important that the parliament, the elected
representatives set out some rules and principles for guiding
this division.
At the same time, giving the regulators some discretion about
how it's to be applied over time.
Yes, we did did endorse it.
Orr other questions?
>> We spent some time yesterday and today talking about
You covered quite a lot about that in your own remarks.
Of course, in the stress test one of the issues that comes up
is trying to ask the institution being stressed how they would
react to various kinds of shock.
We haven't really talked about the official sector and how
transparent in advance the official sector should and/or
can be about how it will respond during crises.
That's also an issue we had to grapple with on monetary policy,
Sort of rules and discretion and commitment and so on.
Do you have any advice for the official sector on how to think
about its role in all of this?
>> So I think the official sector certainly, A, needs to
have plans in place, contingency plans in place.
B, I think it needs to set out as clearly as it thinks it can
without creating the problem it's trying to avoid.
How it might react it to various situations.
A real problem is you don't know how the thing is going to come
down or how it's going to act.
There's not that much I think you can say in advance if, A,
then we'll do B.
I think it's somewhat limited.
You've raised a broader issue really, which is transparency
about the macroPrudential process itself.
In the united kingdom there's a great deal of concern in the
parliament about accountability of these new committees and of
the bank with all this new authority.
I think it's very hard -- at least in monetary policy, you've
got some measurable outcomes.
Inflation, unemployment, that you can judge policy by with a
couple-year lag, and you've got most of the time one instrument
to work with, a short-term interest rate.
Sometimes the size or the composition of the balance sheet
as well.
It's much more complex what you're doing in maximum row
Prudential regulation.
You won't know for years, maybe decades, whether you've been
successful or not because success will be judged whether
or not there's a crisis in the financial system.
It's hard to get measurable outcomes where you can say you
guys can did a good job or bad job until its way too late to
have that be a productive conversation.
There are a lot of dimensions upon which macroPrudential will
be operating, so tieing one instance to one outcome is
One of the things we need to do and are doing is being
transparent about our process.
So we are publishing a record what we in the U.S. call
minutes, like minutes of the open market committee.
A record of our deliberations so that the public can see all the
topics that were discussed and the fact that the
recommendations reflect the outcome of several different
Viewpoints being discussed.
We have the financial stability report, the governor gives a
press conference when the financial stability report is
Of course, there are testimonies beforeliamentparliament.
I think in the macrostability area, it's more important than
in the monetary policy area to be transparent about the
processes that you're using to assure people that you're
looking at various possibilities.
You've considered potential different tools to use because
it's so hard to measure the output.
>>> This may I noticed some continental regulators have said
we want you to pull back on your lending assets, funding,
whatever, you know, outside the home country, because we think
that's going to help maintain GDP here.
So, you know, there's a implicitly macroPrudential in
some jurisdictions has a national limitation right.
To what extent do you think about those feedback effects and
how do you think the community should deal with this issue in
the long run?
>> It's very hard.
We do think about it.
We certainly think about feedback of other
recommendations on what UK banks might do in terms of lending.
I think our first -- and in the UK first.
That's our responsibility, but elsewhere as well.
So we have been very careful to couch these recommendations
about balance sheet management and balance sheet resilience in
ways to say you should be looking at what you're doing and
cutting back where you can without exacerbating financial
fragility or cutting back loans to households.
I think there's a natural tendency for the banks and
authorities, the country, to want their banks to be as safe
as possible.
And there is a risk here.
It's the macromacro in some sense.
The macroprudential risk is as each country protects its banks
and lending in its geographic area, the banks cut back on
lending elsewhere.
That's one reason why Wii been stressing the level of capital
in our recommendations.
That's the one way around all this, isn't it?
If you can raise the level of capital, not the capital ratio
per se, the level will raise it but not by reducing the
denominator, you can make your institutions more resilient
without the adverse feedbacks, but it's very hard under current
We're pushing hard:.
>> You mentioned one of the resolve troubled institutions
more easily, and that's logical from the way you described it.
Prior to that there's a divide across the pond about the
philosophy of liquidation authority and the ability to
resolve institutions.
The perception is from our side that British authorities didn't
have a lot of confidence in their ability it to resolve
institutions, especially by comparison to their U.S.
Do you think that's going to change?
>> I think that's one of the points here.
To make them to -- to enable them to resolve these
institutions in a more orderly way while imposing losses on
some of the creditors.
I think you have to appreciate that the size of the banking
system relative to the country is much bigger in the UK.
I think banks are like four and a half times GDP in the UK, for
So it's very difficult for the government to step in and do it.
It's a huge fiscal risk here, but I think what you're seeing
is an effort to build the capital requirements so the
government won't be the tax -- the taxpayer won't be on the
hook, and to make the resolution easier.
Certainly the goal I can tell you -- I've certainly had
conversations like this.
The goal is to get where you want them to be and where they
want to be, where we want to be.
The ability to resolve institutions no matter how big,
and this is a step in the right direction.
>> We might have time for one more question.
I guess not.
>> Thank you very much.
>> Sure thing.
>> Good afternoonment we're going to take battle of time now
to go through the conclusions Approximate if I can ask our
workshop presenters to come up here.
We have Greg Berman, Larry tab, Dell gray, robin Doyle, Laura
and Bob Jerrold.
>> I guess so.
We can sit here.
This is set up for the next -- set up for -- in order, I guess.
They'll be set up there as well.
Hi, robrobin.
We're in order.
1, 2, 3, 4.
>> If we were we remember back to yesterday morning in those
panels, our discussion leaders sort of built the foundation for
further discussion later in the day on three focused themes for
the analysis of threats to financial stability.
The first is the actual data itself that makes the analysis
The second is the use of standards as a means to
facilitate risk management among other things such as operational
The third is the use and implementation of data --
diligent data management practices as a means for
operation risk.
In the afternoon sessions our workshop leaders led spirited
discussions based on the themes.
They brought them down to a greater level of detail.
We asked a representative from each workshop to take a few
minutes today to review the conclusions drawn out of the
Let's start with 1-A, systemic risk math 1.
>> Here are the slides we have prepared.
Thank you.
Our workshop was focused on data, data particularly around
systemic risk measurement and focused narrowly on markets and
It was with the world bank of Canada and Don from university
of Maryland.
In order to focus the discussions on a very, very big
topic, we have 90 minutes.
We focused the discussion.
Pretend you're the risk manager of the financial system, and
with that in mind, what kind of information, what kind of data
would you like if you were in fact a risk manager to the
financial system?
What would you like to see?
risks to capital?
What are the risks to liquidity and with those risks in mind and
perhaps measured some of them, what might be the contagion that
spread through the system?
We started the conversation really with a supervised firm,
so we start -- we said let's contain ourselves for now to the
realm of supervised firms that interacts with the rest of the
financial system.
Think about collecting international risk information
from the firms to construct not a risk report for the financial
system within that realm covering credit counterparty
market risks and what was left out my apologies is also
liquidity risks.
As an example we talked for a while about the data that was
collected through the stress tests.
Things are here for what the templates liked like.
It's a fairly remarkable amount of data there.
This is the ten largest firms and ten biggest counterparties
and risk factors.
We go to the second slide, please.
The risk report may look like this.
This is startly narrowly from the market risk report you might
think about.
Down the column to the risk factors.
For example, the collection across interest rates spreads.
Currencies and equities and so on.
You have the aggregate sensitivities aggregate up to
the system.
On the right, gosh, you might not want something not just for
locals but more stress to get a sense for how resilient the
system is.
We skipped -- can we skip that slide and go to the next?
So we had indeed a very spirited discussion in our session.
Here's a couple of issues that came up in the course of this.
First, this came up a number of times.
How should we think about risk?
Should we think about it in measuring the loss in fair value
terms or accounting terms?
The reality is it's done in both terms, depending on which part
of the bank you're looking at if you look just at banks.
This mixed regime of accounting presents its own set of
Second, number of folks said we should think about collecting
data that enables modeling the impact of all the behavioral
feedback loops.
It's not just enough to model the impact, but think hard about
the data you'd like to collect to get that second and third
order effect.
Third, what about disclosure, selective disclosure to inform
market discipline?
Not necessarily a complete open KIMONO but some disclosure to
help the market discipline aspect of this.
Last, avoid complexity.
A couple of folks said, look, this is all incredibly complex.
Let's not forget simple metrics like the leverage ratio.
Finally, he led this discussion about essentially suggesting and
calling for a systemic collection of options data by
the OFR.
Not just use direct, if you will, cash price information,
but options and derivatives by nervous as a way of getting a
sense for forward-looking risk measures that the market has
embedded inside it.
I think that probably does it.
I hope I captured it.
>> Thank you.
>> Thank you.
Okay, workshop 1-B.
>> Thanks.
The purpose of our workshop was discussion about data standards,
and I think most people would say data standards should be
pretty standard.
Trust me, it was an extremely heated conversation at times.
Something as simple as the format of a date becomes a
matter of opinion.
So we basically just have one slide.
These are some of the theme that is we hit upon.
I think that these are theme that is most folks agreed on,
but there was a variety of opinions even on these themes.
One of the things that we discussed was a puzzle about why
in many other industries do you tend towards a national
You know, you can download 10 trillion videos from YouTube or
an unGodly number in the same format.
There are many standards that seem to have not been mandated
but evolved organically.
What is it about finance where that does not seem to be the
There's lots of standardization in finance, and we went through
some examples of where that happens but not to the extent it
happens in other industries.
We opined on that a bit, and I think the phrase that came up is
it's easier to cheat in finance than other industries.
If you want to upload your video to YouTube, you have to meet
that format.
If you don't, it just doesn't come out anywhere at all.
You really have notice choice.
In manufacturing, if you want your part to fit in the car, it
has to meet the spec otherwise it gets thrown out.
In finance especially when you deal with regulatory data with
the transmission of DAT DA, often the person on the
receiving end has to do all the work.
That led to some to conclude it might be an area where
regulators need to do more pushing rather than less pushing
than they would have done in other types of industries.
We talked a little bit about the has been needed for as long as
there's been legal entities, and yet there's still no standard
that evolved out of the industry by itself.
In addition to thinking about the fact that there might be
requirements of more regulatory push, even the LEI, which I
think we have a number of opinions that this is the most
basic and the most simple thing that one can do.
As soon as you sought to discuss it in 10 minutes of discussion,
even something Na simple is not simple.
It's extremely complicated.
You start to bring up hierarchy issued and cross-border issues
and privacy concerns.
We talked about that.
We spent time talking about the need to consider different types
of data during a crisis and a non-crisis.
Not just the data but even how the data is collected.
If you are in a pre-crisis mode and no one is defaulting, then
you might not even have a format to consider recovery rates and
default rates in a post-crisis mode.
Interest rates may be less interesting for a particular
bond as opposed to the recovery rates.
The way you think about data may change between precrisis and
post crisis.
We spent time talking about what type of data are we supposed to
I think most of the everybody jumps right into the well we
need to standardize data for and how do you we communicate a swap
and equity holdings and financial transactions and
We realize that especially during a crisis you need a lot
more data than just the financial data.
Sometimes you needy tailed information on the legal
contracts, and that's actually extremely difficult because
there are no standards for how legal contracts are written.
We somewhat lamented about the fact you can't turn that illegal
contract into a cash flow very, very easily.
Finally we talked a little bit about the need for developing
common terms and definitions.
That is really underscored by the fact that the last line says
government fixed income management, and I have
absolutely no idea what that means.
I think we need to have standardizations of what we put
on our own slides.
I think this was really speaking to the fact that when we talked
about maybe we could get a standard on date and figure out
what a date really means so we all understand that, but what
does "settlement date" mean?
It means something completely different for different
instruments and means something different in the U.S. or Europe
and Asia.
When you get down to the level of being able to have GRAEZ
formats where it's very, very clear, you need some overlays to
explain what the data is.
Weapon spent a little bit of time talking about semantics.
We didn't talk about taxONIMIES, about if we did we'd have ten in
our group.
>> Systemic operational risk 1.
>> Thanks, Andrea.
We had a great panel.
We had Walt Luken and we had a very spirited debate really
about data standards and really looking at data material models
and how we get all the stuff.
We had a practical group of people, which meant that
everybody started talking about LEIs and universal product IDs,
everybody said how are we going to do?
They said there's no way to do it quickly.
What quickly came to the floor was that, you know, if we're
looking at macro systemic risk, needles.
We need to look at elephants and look at the big picture things
we can gather that quickly.
That was a good, interesting debate.
We also need to take a look at developing conceptual basis for
what matters.
So are we looking at leverage or concentration or
collateralization levels?
Let's figure out what we try to measure before we start asking
for the universe and start boiling the ocean.
Also issues came up with was that, yeah, we can ask for all
this data and you can give us reams of data, but what does it
mean when it gets to the regulator?
They understand it way better than the regulators will.
It will keep them involved the whole way.
Also the systemic risk reporting doesn't require real time data
NM crisis mode so you can do it on a periodic basis.
When it gets into the crisis mode, you need more, you know,
realtime information.
So you can't just put this off gathering the granular data
You need to move towards that.
In the intermediate term you can look at higher level numbers.
On the next page?
There are also other ways to understand risk besides just
crunching lots of numbers and we talked a lot about surveys and
more outreach oriented mechanisms to really talk with
the firms and have them try to get you -- what are your top
What are you really worried about?
What are the things that make you -- that will blow you up and
really work on a somewhat soft basis to try and figure out
where to focus.
Another topic we talked about was outreach in collaboration
are much better than enforcement action so that the OFR should
think about not necessarily doing it with a stick, but
really working with the firms to they need.
If they start turning it into more of you didn't get me this,
you're getting fined, it shuts down the lines of communication.
Also in terms of publishing the data, that becomes a difficult
and tricky issue.
When you look at risks and what blows you up, this needs to be a
level of confidentiality so that all of a sudden you don't create
runs on the organizations.
In the last topic we really talked about was governance and
ensuring if we're going to go into this soft outreach mode and
try to find out where your problems are and how do I work
understanding the macro risks you're not just dealing with
someone not connected at the board level or CEor or top
levels of the organization.
We really need to build in a governance structure so that the
board understands these issues Just make sure the information
the to the OFR on a high level basis is actually the right
level of -- the right information and not -- and
not -- things are being hidden.
I think that was pretty much it.
>> Thanks, Larry.
Workshop 2-A, systemic risk measurement it 2.
>> Okay.
This was on a sovereign risk, systemic risk, and national
accounts, and I was the leader along with Carmen REINHart and
Whenever Carmen is it in a worm shop, it's an active discussion.
What we talked about was many aspects of sovereign risk.
balance sheets are important for macro Prudential analysis for a
couple of reasons.
First the banking sector is tightly linked to the sovereign.
There are a number of feedback loops between the banking sector
and the sovereign in terms of flows, con sin gent liabilities
and both the banking system affects GDP which affects the
fiscal situation and then fiscal consolidation affects GDP as
well, which can affect economic activity, the banking system.
The sovereign balance sheet is often an inevitable channel of
risk and loss absorption in a crisis, as we've seen.
So there's a -- there's a need and we had a discussion on
understanding sovereign credit risk and the distinction in some
cases, whether it's a willingness to pay or whether
it's an ability to pay.
And discuss some different country cases.
There's clearly a need to have more information on sovereign
balance sheets.
Inventory and valuation of contractual liabilities and
explicit guarantees.
We discuss that some countries like New Zealand and Australia
have good frameworks for putting together the economic balance
sheet of the government with exposures, assets and
Then this last bullet point on the first light on hidden debt
and contingent liabilities, we discussed many different angles
on this.
The government hidden unfunded liabilities.
We discussed the blurred distinction between federal,
state and local responsibilities in certain areas and unfunded
liabilities and clearly a need in that area to get better data
and get a better understanding.
We also discussed a need for more information on recognition
and being very clear about central bank liabilities and
central bank debt and this is very important in the context of
certain foreign exchange crises in particularly emerging market
countries or countries with floating exchange rates.
Carmen mentioned some interesting work on below the
line of arears and looking at when the blown arears occur in
the fiscal accounts and these are an important early warning
indicator in some of the two countries at that came OUT worst
in this early on were Greece and Portugal.
So if we can go to the next slide.
We discussed the need for better measures of sovereign risk.
It's a concern that the main barometer for sovereign risk is
debt to GDP.
We have all kinds of good risk measures until the financial
sector, in other sectors.
We have value of risk and we're relying on this accounting ratio
for sovereigns which lead to certain problems.
We need better information on modeling the feedback and spill
overs between the BAJing sector and the sovereign.
Going forward we talked about the need and the desirability of
carrying out joint stress testing the banking sectors
today with the sovereign both with growth feedback and then
relate this to macroprudential policies and CLOOERly related to
too big to fail policies which are large contingent liabilities
from the banks.
This is work we're doing at the IMF, and we have a framework
we've done on several countries where we do stress tests
sovereign banking and link it to growth.
I have one more slide.
This is from the global financial stability report
published last year.
I think it's a good way to capture the complexity of what's
going on especially in Europe, different channels.
The first channel is if you go to market to market fall in the
value of government bonds that affects the banks.
You get an increase in contingent liabilities back to
the sovereign, and then you get an increase in bank funding
costs from the higher sovereign spreads back to the banks, can
make their situation worse.
As we have seen in some cases, the erosion of official support
goes down.
When the sovereign is bad enough, you're not going to be
able to support the banking system.
Then we get these contagion between banks and then as weave
seen, when one sovereign lets say in Europe comes under
pressure, then similar sovereigns also come under
It's an enormously complicated system, one of the frameworks
that I'm using and doing a project with the ECB right now
uses contingent claims models for banks and sovereigns and to
look at the contingent liabilities to spillover and
linking it to the growth models.
I think that's it.
>> Thank you, daily.
Workshop 2-B, the systemic risk data free market 2.
>> Okay.
So our workshop focused on the data framework for systemic
We focus to three key theme.
We be looked at aggregation, briefly at standards, since it
had been covered pretty extensively in panel discussions
and then we talked a little bit about technology.
A key theme that came across for sure in the session was that
In many, many cases you need to have the core data elements that
comprise the information that you're using for risk
So there is needs to capture data at a very granular level.
However, you have to be careful with using data only at that
There's a need to aggregate information and look at it
depending on who your audience is what purposes you use it for.
Granulator is good in the right cases and needed for analysis
but can't always be used for communication.
A point well-made is AK integration is oftentimes done
from the purpose -- the perspective of the person who is
trying to deliver a message.
That will naturally have bias in it.
An interesting point made was that it will good in cases where
data can be shared among different parties with different
interests and can be looked at from different perspectives.
Then there can be a challenge to the outcomes and conclusions
made from that data, so if you think private verse public
sector having the information to the fable and STHARing views on
what that information might be conveying.
It was a very interesting point.
They're intimating linked.
Classification, we think of that as reference data.
It's important to group information into certain
We talked about things like industries and, of course,
There was a warning, of course, that went with that.
That is that you need to pain taken your classification
You need to maintain your reference data to do that very
So it does take a regular and ongoing maintenance to do this.
Next slide.
>> A critical point came up.
Greg touched on this a little bit.
You need -- you really need both numeric and non-numeric data.
You end up -- you definitely end up when you talk about data
frameworks migrating to trade information.
Trade information by itself won't give you always the big
Oftentimes you need to understand the terms and
conditions in the trades.
So you need to collect numeric data and think about the
non-numeric data when you think about systemic risk.
Terms and conditions, legal conditions and reputational
risks and operational risks and looking at processes and whether
they work properly.
Somehow this has to be brought into your data system.
We did, as I said, mentioned standardization.
We did very briefly.
We took a quick poll in the room.
We said who believes standardization is important?
And it was fairly unanimous that some level of standardization is
required to do the things we've talk about.
The last item was aggregation and confidentiality, and so the
point being that aggregating data is a good way to keep data
So it is, again, a any part of the data system in order to
protect the confidentiality of information.
There's a couple other ways to do that as well.
Making the data anonymous and waiting until its stale, which
I'm not sure how effective it is but the point was made.
It was a good session, and hopefully some of these points
will be helpful to the OFR.
>> Thanks, robin.
Last but not lease, work SHOB 2-C, systemic operational risks
>> This will be made available here.
Since we're two people, it's too much pain to walk to the podium.
Our particular workshop covered tracking relationships and the
challenges of counterparties and collateral in complex and
interrelated and connected markets.
We started at the same place robin left off, which is talking
about transparency and relationship flow.
You need to know who is connected to whom, but there is
a question about how much transparency you want to
participate in certainly in term it is of the public transparency
issue and discldisclosures.
When could one disclose and how much information could be
Whether there would be privacy concerns, and then lastly
whether you need to disclose it in a period more stable on a
regular basis.
If you disclose relationships weighted in one direction or
another during a period of instability that you may, in
fact, cause the panic you're trying to avoid.
We talked about collateral.
As you know, in many connectors about counterparties, collateral
is key.
So we discussed what it is that we could collect on collateral
as was mentioned in numerous cases.
It's not so easy it to collect collateral information sitting
in all sorts of different situations for banks and
non-banks and exactly what it is and how it's haircuted is a
relevant piece of information.
We encourage the OFR to try and get a handle on that because we
think it's an important part of monitoring systemic risk.
We need to follow the chain of collateralization in terms of
Very difficult issue, but one that has to be on the table.
We also discussed how reporting burdens could be lightened.
Clearly, there's a lot of discussion in the last two days
about the amount of information that would be nice to have.
There is a lot of it, and a lot of it is probably already
produced inside firms themselves.
So we talked a lot about how we could pull that data that
already exists, and if for no other reason to look at what it
is that firms actually produce in their own risk reports as a
first-base measure.
That would gave comparing KRON TRAST, and tell it what it is a
firm thinks is the most important risks to start with.
That information should be available within banking
supervisors, so that's already in some sense readily available
by those that do on-sight supervision, so there's no need
to push that by asking for those kinds of things.
That can be done internally.
level of the types of things we discussed.
I want to turn it over to Bob because he brings a different
perspective to the table that might be helpful to the
>> Thank you.
I thought I left Ithaca to get out of the classroom, and
yesterday afternoon I found myself back in.
It was quite fun.
It was a nice, heated discussion.
I just wanted to share a few sights of mine from sitting
there and listening to the group.
In one ways with respect to the first point.
There was a view expressed that continuous release of
information about systemic risks by the OFR could be a financial
crisis dampening mechanism, and so I short of share that view.
Of course, there was a counterview that perhaps in a
financial crisis they were releasing too much information
could make things worse.
Of course, we didn't resolve that but had a heated discussion
on that tradeoff.
On the second point when I asked the question to the participants
in the room as to whether they had machine rateable data on
counterparty risk exposure on the collateral position.
It seemed the consensus opinion was, yes, that data was
available in machine-readable form, so our conclusion was that
the OFR should get that data.
That would be very beneficial.
Now, on the last two points, we get a sense for the regulatory
burden or the negative asking for additional information.
There was a positive side, too, expressed in our session.
I think the industry participants recognize that the
OFR was in a unique position to collect and aggregate
counterparty risk exposure data.
Not just get data on the first link in the chain but to get
data on the entire linkages of the network.
And that that data would be beneficial for policy-making
under consideration.
They were very pleased at that possibility.
That happened between the first point.
This is the type of information that should be shared back with
the industry.
Of course, in a SanITIZED form.
I think they would definitely value it, and I view that as the
carrot that the OFR can provide to the industry.
Those are my three additional insights.
>> Great.
Thank you, Laura and Bob.
I want to thank all of the work shop leaders and up participants
for the insights.
I moved around to two or three of them, and they were all
really great and spirited.
We need to move into the next item on our agenda, which is
session 3, risk management, what's the frontier?
Thank you.
>>> All right.
I thought we'd go ahead and get the session started.
The afternoon session is forging best practices in risk
management, and we're going to be presenting a paper that is a
three-part piece co-authored.
The three pieces are to a large extent independent.
What we're going to do is follow the standard presentation
format, except for going to have each speaker present their
CHAPTer and immediately following that is a discussion.
We'll then provide some commentary, and we will try to
leave as much time available for commentary as possible.
The first piece of the paper is going to be presented by Paul
Wasserman and it's firm level risk management.
This is taking a look at managing countercyclical LILT in
a world of regimes.
Paul is the Jack Anderson professor at the Columbia
business school.
After that Craig Broderick will be discussing this.
Craig is from Goldman Sachs, and he's the chief risk officer.
The second paper is risk governance and incentives.
They're going to talk about risk adjusted compensation, and it
will focus on behavioral biases.
That will be presented by cliff Rossi.
Cliff is the executive in residence and tieser teaching
fellow at the university of Maryland business school.
That will be discussed by Andy.
Andy is the executive vice president and chief risk officer
for State Street Corporation.
The final paper will be systemic effects and it's going
to recognize contingent obligations and the limits of
hedging as sources of systemic risk.
That paper will be presented by -- that piece will be
presented by David MORTICAH, who is the president and co-founder
of risk economics and the discuss ant is Mike Alex who is
senior vice president and co--head of the risk and policy
functions at the federal reserve bank of New York.
I'm Craig Lewis, and I'll be the moderator.
I'm the chief economist at the securities and exchange
commission and director of the division for strategy and
financial innovation.
Paul, come on up.
>> Craig introduced me as a professor, which is accurate,
but this year I'm working at the office of financial research.
It's a great place to work.
Join us.
Send your students and colleagues.
We need to keep building the team.
I think mentioned this morning, this paper and a couple of
others today are commissioned works.
We were given the task on writing on forging best
practices on best practices, and we divided up the broad question
of best practices in risk management in the way
illustrated in the figure into these three categories with
David taking the large surrounding golf course, cliff
taking the penthouse suite and I got the kitchen, firm level of
risk measurement.
There are plenty of interesting plumbing and wiring issues to
talk about in risk measurement, but for this setting it seemed
more appropriate rather than get into details of best practices
on specific measures of operational risk, market risk,
and so on, to try to identify some broad topics and also
issues that are relevant to the OFR.
What I do in the paper is we can go to -- we'll go to the slide
in just a moment.
Before we getting to sort of the main focus of the paper, the
first thing I do is highlight what I see as some of the main
changes and unmet challenges in firm level risk measurement in
the wake of if not crisis.
There's a fairly long list in the paper, so I'm not going to
go through all of them.
Actually, quite a few of them can be broadly put under the
umbrella of radically heightened attention to counterparty risk.
In fact, I would say that I think that a complete change in
perspective on both the measurement and pricing of
counterparty risk is one of the lasting legacy of the financial
We can see that we have quite a few indications of that
discussed in the paper.
One is decoupling of funding rates, so LIBOR versus OIS and
three-month LIBOR versus six-month LIBOR.
The general move toward central cleaving the derivatives,
reduced credibility and reliance on credit ratings with no object
just substitute have been emerged yet.
A completely different perspective on sovereigns as
counterparty versus counterparty risk.
Lastly, the institutionalization of credit value adjustment both
in regulation and in market practice.
There's time later, and I would say that's perhaps one area that
deserves further attention, possible cause for concern going
It's not the main focus of what I'm talking about today.
I think counterparty risk is particularly relevant to the OFR
since it deals with the interconnections with the
various parts of the financial system.
Let me go to the main focus, if we could go to -- we gave
ourselves a budget of one slide each, so there's one slide that
summarizes the key question.
I guess we could go ahead to the next slide.
This is my one slide.
Let's go through the bullets and explain what I mean by them.
This is the importance of taking a long-term view in historical
data looking forward.
I suggest it in looking back but also over the horizon of the
measuring risk going forward.
When you do that, a key feet tour of the historical record is
the presence of volatility regimes.
Firm level risk management that anticipates a change in regime
is a source of risk amplification and many firms
respond the same way to elevated volatility.
A macroprudential view needs to encourage risk measurement sense
I have to shifts volatility while creates the amplifying
effect of this very sensitivity.
through those points and what I'm getting at.
On the first two imagine it's back sometime in the first half
of 2007.
You're calibrating a risk model based on historical data.
You're looking back maybe four years of historical data, which
would be a respectable amount of time to look back.
If you're doing that, you're calibrating your model to what
is historically one of the lowest periods of volatility
that we've experienced perhaps ever.
I've got quite a few figures in the paper that the point to
Relative to that period, the period that we're currently of
in of high volatility which is inconceivable and hard to
capture in a well calibrated risk model.
If at that point in time looked back three to four years but 20
pattern of periods of high volatility and low volatility
and it came up several times yesterday and today.
Again, there's several pictures -- several notions of
volatility that you could take by which the -- that respect
that pattern.
Several figures in the paper that illustrate that simple observation.
>> Presumably, you want it to start investing in making a lot
of effort to get their risk measurement very accurate.
However, what are the consequences having everybody
measure accurately.
We have a pattern that's illustrated the several example,
starting with a bank run.
The crash of 1987 and portfolio insurance, the crisis of August
2007 which Andrew Lowe spoke about this morning and has
written about as well.
Collateral calls on aig.
Widening hair cuts in the repo market.
All of these are illustrations of what happens when lots of
firms take action, which, if followed by one firm alone in
isolation is a perfectly sensible prudent risk strategy.
When adopted by many firms becomes a source of risk
The consequence of having sensitive risk measurement
combined with a change of volatility regime.
What then is the solution to that?
We saw bank runs indifferent to risk.
It basically make us indifferent to the riskiness of your bank.
Clearly that's not the approach that we want for sophisticated
financial institutions.
We want them, as you said, to continue to invest in making
risk measurement sensitive.
But I guess the way I summarized it.
We went over that slide a little bit quickly, I think the goal is
measure sick likely but manage counter sick likely.
We want firms sensitive to how they're measuring.
This relates to short-term volatility.
You want terms getting short-term volatility right but
you want them thinking about long-term volatility.
The point I'm making it's important to separate the risk
management function.
One example of the measure that doesn't do that is the new bar
standard which combined a normal var into a stressed V.A.R. into
a single numbers gets into the single measure.
In doing so, you back test it and observe whether you're doing
a good job measuring V.A.R. there are mechanisms that
explicitly separate the piece that you're adding on to a
counter cyclical measure and still allows you to precisely
measure your sensitivity to the current risk environment.
That's a useful organizing principle for firm level of risk
measurement and for regulation to encourage.
>>> Thank you.
I'm still reflecting on living in the kitchen.
Sometimes it's warm and coast in.
Sometimes it's the latter period.
Thanks for the opportunity to read these papers and comment.
And discuss it.
The first part of the paper, and I hope you get a chance to read
it in detail, all three sections are very interesting, but the
first part addresses what has changed in the field of risk
management over the last period of time, unkey fined
specifically, but we have come a long way in terms of
sophistication and tools and awareness that we have.
I think as a practitioner, we agree with that.
There's no comparison where we are in our sophistication and
awareness of down sides and what would have been much too
technical to contemplate kinds much issues that we refer to and
manage on an ongoing basis.
The paper does a good job of identifying and running through
some of those.
My observation, I've seen this lesson before.
And I know I speak to many of you.
In this case, because both the significance of the losses, and
the dislocation and so forth that have occurred, I have no
doubt we will remember it, and even if we weren't so inclined,
in and of ourselves, many people in this room will make sure we
I think that's a profound change that will clearly affect our
behavior going forward.
That's point number one.
The professor in this first section of paper turns to actual
market behavior, didn't go through this in great deal of
detail, but the general points were in periods of shifting
volatility reseems, there are tools that you can use to
enhance the effectiveness of other risk measures.
You can impose additionality.
I find that to be quite interesting.
One thing made in the paper, that V.A.R. is useful in a
kicking off tool in risk management.
Because it require in many cases firms to pull together in a
single form.
Apples to apples basis, risks across the firm.
We went through that ourselves, one thing I think I would note,
this goes back to comments that you made in one of the earlier
presentations, if you want to manage risk evocatively, you
have to measure effectively.
One thing I would mention it's clearly helpful in that regard,
I would go further and say that in order to measure it
effectively, that's an area where certainly the last crisis
in the industry, all participants had difficulty
doing, when I say price perspective, you come up with
not what the position is tomorrow.
But where it is based on your models but where you sell it
right now.
If you don't get the prices right on a going in basis, it's
the proper volatilities and other characteristics and that
leads to correct management decisions.
I think the emphasis on that would be great.
But the basic point about regime shifting being an important
state of mind, I like that characterization used in the
paper, is entirely right.
We've tried to do regime shifting, using other approaches
in the past, sometimes more successful.
Sometimes others, we do it but rate, so quickly give affect to
volatilities and therefore avoid the looking back at times when
it's really low and present measure, and it's quite
effective in that regard.
It obviously violates another, which is, you need have a long
look back and in some cases a long forward looking period.
We use stress tests to do that.
That brings me to a third point.
Which is it is great to have more effective, more
sophisticated risk metrix and we always try and improve the
accuracy and so forth of the measures that we use to manage.
By experience, I think our firm's experience in general is
that it is sometimes more effective to have a simple risk
More clear about the specific expose the firm to large
downside risks.
So we have tended to use that letter approach, rather than
vote huge resources to really sophisticated single metrix, we
take credit sensitive widening, it shocks your credit sensitive
portfolio by credit spreads, we have the ability to, when
spreads are high, where periods are low, we P look at examples
of those sorts of very simplistic risk measure, when
used with judgment and an ability to toggle and otherwise
adjust, it can be remarkably effective where the risks lie.
It is unbalanced and a potent combination.
None of that is consistent with one of the professor's earlier
comments which is sensitivity to the circumstances under which
that regime change may or has already taken place.
The third section is how to practice effective risk
management in a firm level without practicing across the
industry and that's the $100,000 question that we spent a lot of
this current seminar on.
It's very complicated.
All I can say, we do our best to adhere and accomplish that
objective on practicing good risk management on a firm basis,
when I say good risk management, I mean going into periods of
potential curation with risk mitt Gants and overall risk
profile and capital profile and liquidity profile which gives us
an ability to be resilient ourselves and maintain our
standard, maintain our credit terms, I should say, with
counter parties on an unchanged basis across that cycle.
So, nothing is more detrimental, and in some cases, less
actionable, than going into a crisis with risk mitt Gants that
were barely adequate for favorable periods and needing at
that point to change to tightened credit standards.
We, I think have been more -- we've been pretty successful in
doing that in this period.
There's lots of other components that we can talk about in more
detail, including trying to keep the overall risk exposures to
reasonable levels, not only giving effect to, you know,
current exposure, but the directionality and relative size
of positions compared to market liquidity, and therefore your
ability to unwind and collateral designs you might make on a
client and their ability to meet them and lots of other things we
can talk about in more detail.
But those are my critiques.
Good, interesting paper.
Thank you.
>> Okay, I'm going to shift gears a little bit.
I'm probably the loan WHOFL in the whole two-day session, I'm
not going to be very analytical.
In the interest of full disclosure, I'm going to tell
you I'm a refugee from the banking industry and recovering
cro these days.
I hope that puts things in a little perspective and take
professor lo's analogy and twist it around a bit.
You look at my bio.
You need to know what the trash looks like to figure out what a
good salad smells like and tastes like.
Hopefully that's the gist of what my paper is about.
But effectively, what I was looking at, if you go back to --
and everybody, I'm sure could look at all of the different
press information and what not, information, fdic's office.
Inspector general's information with regard to a variety of
different institutions that exhibited all sorts of
breakdowns in corporate and risk governance the last year, and
come away with a variety of an EK dotes.
They aren't empirical and this paper is not empirical.
But what I got to reflect on, certainly my experience in the
last 20 year, a little of those a little more hotly discussed
than years leading up to the 2007 and up to 2009 when I left
to go into academia.
The idea is this.
There's two strands of work out there.
Some coming from the areas of optimal contracting and design.
And others coming from the field of behave yoral economics.
I had to admit I was never an adhere ant to the area of behave
yoral economics until I started experiencing it for myself.
So from the standpoint of someone who lived through this
A couple examples where, how would it have been if you could
rewind back to 2005 and tell the board or tell senior management
that you might be able to tell them that there's a significant
decline in home prices coming.
I can tell you from first-hand knowledge at several different
board meetings that that is a very difficult conversation to
Even telling them there's a possibility in 2005 of a small
deterioration of home price is a difficult problem.
What I do effectively, I guess I call it my primary contribution
if you will.
Is basically merge together these areas.
That is looking that the idea that under weak corporate
government structure, where awards can be held more captive
to strong managerial power hypothesis, if you will, where
human managers have more incentive to design there and
have more to add in control and design of their compensation
structures, those incentives in turn laid into their utility
maximization function where their incentive structure is
based on a whole series of targets and biases, which we can
spend the rest of the day discussing but I only have five
more minute, I think.
Things like confirmation bias, and the driving in the rear-view
mirror, if you will.
The herd mentality.
If I heard it once, I heard it several times in my career, why
can you do it and why you can't.
These are all issues that I bake into this structural model if
you will, along with this thing that I certainly have come to
believe more in, which is the notion of ambiguity.
That those of us risk managers, are in the business of trying to
quantify uncertainty and realize when we present that information
to senior management.
Often times everybody else has their view of the world as well.
In looking at all of these pieces, basically I was trying
to distill basic things of what I've seen the cost of the
industry has seen within my own situation, and over the last
several years, and come away with at least the structural
model, realizing that I'm always a little skeptical of putting
together a theoretical model but tried to describe a real
behavior, but at the end of the day, I think it actually paints
a fairly strong picture of what would best be described as some
of the bad behavior that was exhibited by a number of
So in this paper, I actually present a series of potential
policy prescriptions of a variety of things, including
financial incentive, including management, to strengthen and
understand the risk management better.
These are things like requiring dno insurers to be more vigilant
how they go about assessing the United States lying risk
The application of sba and ray rock and other things whoafully
needed at several institutions that I had the pleasure of
working at and importantly, very importantly, it all has to start
from within.
An institution that doesn't already get it.
It's very difficult to at some point embrace risk management.
From that standpoint, enhance the stature of the organization
this gets back to the discussion we've had in the last couple
>> That sometimes they can be self-defeating.
If you have poor data.
What you can experience.
It may help facilitate the lack of investment in data and
modeling and technology, and it all starts from that standpoint
on, from having a good statute built within in organization.
The data and analytics can be self-fulfilling.
I said you can be sitting in board meetings where you have
conversations with ceo or what not about a particular view on
home prices and interest rates.
At the end of the day.
The data might not be as robust as you might have that.
And win the conversation that day or not.
That's effectively.
Treatment, if you will, what this paper was about.
And I look further to Andy.
>> If Craig is in the kitchen with ball.
I'm happy to join Paul in the attic.
I'm happy to be invited and in particular, giving me an
opportunity to comment on cliff's paper on risk
Risk governance is something I focus on in my prior life where
I was consulted on risk management.
I'm learning that it's easier to critique the organization's risk
practices from the outside and management from the inside.
This is a sign that I could become a victim of the cogfy TIV
bias described in cliff's paper.
I'm going divide my comments in brief.
Three parts.
First I want to address motivation, what to cure about
risk governance.
Second, wasn't to comment on cliff, the diagnosis of the
problem, what's the underlying cause of poor risk governance.
And I'll reserve most remarks to potential solutions, what did we
do to include risk governance.
Let me start with the motivation for the paper.
Which is the easiest part.
There's no motivation for the crisis to me the crisis was a
failing of risk governance and risk management financial
institution, essential fact of the crisis was that it exposed
glaring, unexpected weaknesses at several of the world's
largest and most sophisticated financial institution.
You can think AIG, Merril, UB KRFRMTS and there's a whole host
of others.
There's a host of agreeance on this point.
Cliff mentioned the report, congressional testimony,
international supervisor's report.
If you prefer the dramatized version, you can watch the
recent movie, margin call.
To me, the focus on governance is a natural result of the
magnitude of the problem.
When things fail spectacularly, look for blame at the top rather
than at the bottom.
So, let me turn to diagnosis.
What's novel, I think about cliff's paper, is that he used
beyond anecdote to seek out the root causes of governance
failure, in particular he draws on literature of behavioral
economics to show how poor risk governance, can be caused by the
incentive structures in banking and can be compounded by
cognitive bias.
In a nutshell, cliff describes how a firm living through a bull
market, that's experienced low recent losses, that's a follower
of competitive practices and prefers certainty to uncertainty
or ambiguity, operates a set of biases that erose the acceptance
of risk governance.
This is a description of every financial firm going into the
crisis before at 2007.
So, I think he makes a valuable distribution formalizing the
link between effect and risk governance and reporting this is
a problem in banking.
The question is, what do be do about it?
Cliff, said this falls in three category, look at insurance,
regulators to create financial incentives for BAFRNGSs to
improve governance performance.
Second use of risk adjustment performance metrix, particularly
in compensation design to help altar incentive and there's two
mechanisms that can be adopted to strengthen the role of risk
Number three, I'm cautious about the effectiveness of the first
Very briefly, rating agencies and supervisors had incentives
to monitor risk governance, that's not new.
While ramping up efforts to do so.
The track record before the crisis shows how difficult it is
to observe the effectiveness of risk governance from the
For this reason I'm not optimistic that DNO insurers
will have the ability to monitor risk practice and the level of
LDN premiums on the level of $10 million a year for a large bank
is too small to assert meaningful pressure on the board
of senior management.
Respect to risk adjustment performance metrix.
I similarly think these kinds of dark these metrix are not a
silver bullet.
Ray rock has been around for a long time.
For all of the reasons that were stated in Paul glasserman's
The measurement of what ray rock is all about.
Was difficult during the crisis.
So I think the producers on ray rock results may be a
directionality improvement but create illusion of both decision
and lead to new forms of gaming.
I think we have to be very careful of that.
Finally I'm more optimistic.
And in click, I spoke to the board and recognized the board
is only a part of governance framework.
I'm a strong believer that changes in governance starts at
the top and boards can have an influence on the independence
within a firm and more importantly boards need to raise
their game in risk oversight.
They need to put in place dedicated board risk committee,
that are comprised of directors in the background experience to
add the right question about firm, risk profile, challenge
management, make sure downside scenarios, and views have been
It's too much to expect that boards master all of the
complexities of risks, it's not their job.
The job of business managers and risk managers.
And I think risk oversight can send a clear message to the CEO
about the importance of risk that cascades throughout the
The bottom line for me is that we need to reform risk
governance from the inside, starting at the top of the
house, rather than relying on outside agents and bottom of
metrix to get it right for us.
>> I consider it a privilege to be able to spent today and
submit a paper along with colleagues of this caliber on
this panel, and to share in this with my colleagues.
In response to the request to prepare a distribution, I tried
to use my experience as a highly leveraged transaction lender as
senior rating agency official, being involved in two large
insurance company, both AIG at senior management levels and
then a large hedge fund.
I tried to think about the things I saw across coastal
Maine and what the commonalities were.
And we can move to the slide before that.
So, I think my theme is, very much in spirit with Dale gray
and Robert's view.
We need to start to think more and more about contention claims
or contingent claims landscape, or dynamic involving around it
systemwide between financial institutions and how those
financial institutions support other activities in the economy.
It's got' close link in terms of thinking about advance rates in
the mechanism.
It also borrow, I think a bit from Nigel Jenkinson's early
work with his colleagues, looking at interconnections
between financial institutions in the system how continued upon
a shock.
How small a shock can be.
How much severity of those shocks can grow.
All of the financial intermediation is systemwide
borrowing and lending, conditioning upon changing
Even hedging requires use of leverage and requires borrowing.
We created a set of mechanics in order to collateralize, clear
and settle these obligations.
What that often leads to, as emerging network.
They've been referred to as change of obligations.
We've seen similar language, and some other papers in this area,
and so forth.
What's interesting, though, is when you start to look at option
pricing literature and a lot of language used in common practice
to talk about hedging, it's typically one period, single
trader language, you know.
The truth is.
Post derivatives pass dependent.
The irreversibility that shows up in the standard Delta neutral
Ignores what takes place when you have to make large changes
to your position, rebalance your swaps to get them back or you
have to make a large purchase or sale around a big deviation in
order to rebalance your hedging positions, that often leads to
when there's a significant regime change after a long
period of sort of the same patterns.
Leads to polarization.
This is something that, in the commodities world, the world of
sort of -- the worries carry, were in common practice and use
for 60, 70 years, has kind of, in my opinion been lost in the
main treatment finance world.
So people talk about having positive or negative carry and
they talk about it like it's free honey.
When it's that.
If you're positive carry, you're probably short options that you
should go looking for.
It's usually some option that you should completely -- it's
not time to ring the bell.
It's okay, what did I just sell that I didn't know about.
Because I got paid something.
And I'm not sure what more.
So that -- my big concern -- my big concern, that's the thing
I've seen continue to pop up, at various institution, that's
where a lot of this comes from.
When residual risks, think of it as remainders, I have to discuss
this in a large bankruptcy recently.
Think about it this way.
I talked to accountants and lawyers.
Imagine if you're rounding errors in your check point.
Depending on points.
You got it wrong or added up wrong.
That's the difference between the one model and the multi
period model.
Something as simple as Delta one, right, which is using the
simplest possible incidents.
True to approximately hedge that thing and you end up on the
other side.
When that happens directively, you'll find that you don't have
enough capital, and you don't have enough collateral to make
good on your obligations.
That's really the essence of this.
What's called cheapest fund to deliver option is one of the
more useful tools that I have found to try to understand at a
book level.
How risk accumulates on a book level.
How a cross book risks, aggregates and accumulates
across portfolios.
By using a system of forwards, where we look at a group of
large hedge funds refer to on a risk-based leverage in industry
You start to use conditional leverage and think about what
if, what fors, elasticity around systems that represent buckets
of risk, I'm am BIV slept to whether that risk bucket is
coming from a particular asset or particular position.
What I want to identify is sensitivity in my balance sheet
to change of continue conditions.
I want to translate where that's comes from.
I want to know what sensitivity is.
Look how they cancel in different environments.
And I think, one of the big concerns I've had, and I've
observed is when people treat the matrix as static.
I think a good example of that was in the trade environment
where people would model trades, and very few people figured that
out in '06 and put on very different kinds of transactions
as a result.
Implied correlation used to construct those trades,
correlation was independent.
You have a jump in correlation, you have no change in
As a result, you end up with the mezzanine tranch being a factor.
Some guys, not even derivative guys went, how can that be
possible, right?
As a result.
They put on trades and said, I don't think thisth is
correlation variant and it turned out not no-to-be.
We've seen this over and over again in the past markets.
Where folks took a on-market instrument.
We've seen it repeatedly, where all of a sudden your crowded
hedge, instead of offsetting your risk position ends up
moving in the same direction as risk position.
We saw that in '05 with gm.
When I say korkorian became Kevorkian.
So your long GM credit, in every CDO, everybody long GM credit
and some folks hedged themselves by going short on stock.
What you assume happen, credit falls out of bed.
Stock drops, right and you're hashed.
What happened, credit fell out of bed, when GM defaulted and
they start bankruptcy.
Kokorian said that looks like a good buy, ill ebuy GM, stock
rallied, entire market rallies for three months, nothing could
GM was everywhere.
And so, I think we need to do more what, if and but for
analysis but we need to do it on an apples to apples basis.
We need to have our tradeoff be measure on an apples to apples
basis, and the simplest measurement to do that is the
When I was a lender, you say you always start with revenue,
people can't play games on inventory, they can't play
games, if you start with revenue and work your way from interest.
All right?
I think if you start with the forward it's imbedded in the
option, whether it's a put or call.
I'm not just talking about forward funding rates but I'm
talking about the forward that applies to every asset and risk
We already do that.
We use those things to think in a rigorous manner about all of
the sensitives in the folder.
I understand there's a cost and complexity to that.
But we do have technology nowadays that deal with those
I've seen it in the other sciences, shame on us.
A lost things were developed originally for economics and
have been applied elsewhere and we haven't used them.
And if we reSICHTed, shame on us.
That is a key thing.
I think -- I'm going to leave you with a practical example,
long bond credit, CDS spread option is really what it is.
I also have on a position that's basically an interest rate swap,
with funding on that.
A Spratt funding trade.
I tease out the difference between the bond, and the CDF,
then I make assumptions these things effectively correlated.
Well, they're not.
And a big adverse move on my funding leg and on credit full
swap, you know, premium is going to cause a big problem for me.
But I often say, no, I'm fine.
It's all set, and it really is.
So the paper covers a lot of tradeoffs.
The paper gets into some very real, empirical, real-world
examples of deviations from dealt it one, return
distributions and some of the more path dependent things like
ratchets, shout options, these features sound exotic but they
are imbedded in insurance portfolios all over the world in
a variety of ways, show show up in financial instruments many
which we thought were mundane.
You accumulate them on a balance sheet even certain kinds of
resolving credit lines or stand by letters of credit type of
You start to see rapid feature, and shout features start to show
up and when those things compound, it can cause problems.
Thank you, Dick.
For including me in this distinguished group.
I appreciate that.
You know, I would also, as with my said colleagues, like to say
that the comments that I make here today, are my own, they do
not reflect the views of my employer, the federal reserve
bank of New York, nor the board of governor, federal reserve
system, or anybody else I may have looked for in my career.
The -- before commenting on the very paper that David has
described, I wanted to tech coa comment that was made earlier,
and look at something that cliff mentioned a few moments ago,
that the attempt to mention the riders versus systemic risk, is
not something we started in 2009 after we realized systemic risk
was out there.
For a long, long time I started this after -- I became involved
in it after long-term capital in 1988.
And then throughout the early part of the -- this century, not
And we've long understood that some of the tensions that both
Paul and David have pointed out in their chapters in the paper,
That there are short comings, there's a need to stress test,
there's a need to measure collateral levels, through
periods of stress.
And set your terms of trading with an idea that thing mace not
always be so Rosy.
We've identified the notion of crowded trades, looked at
operational issues and pluming.
In fact, somebody alluded to earlier to the effort that under
Tim Geithner in 2005 to fix the back log of confirmations that
had grown so large with the exponential growth in the credit
derivatives market.
These issues have been out there.
I would agree with Craig that the progress has been pretty
Though I wonder how much of that progress is directly the result
of the fact of the crisis, and seems to be relevant, given that
so many of the fixes are addressing the approximate cause
of the crisis.
So, this isn't new.
We identified weeks between leverage, funding liquidity,
funding liquidity and risks, post long term capital and
understood this notion of leverage, increasingly rapidly
with losses in volatility.
The crisis, we still experienced the crisis, and I would argue
that so much of that is a function of just a simple notion
that there's regime shifts and uncertainty and things that had
thought to be riskless or very low risk, became quite risky.
Seeing through through the -- all of the possibilities of the
way spillovers can happen, is a very difficult thing to do and
keep us challenged, I think for some time.
>> So, with that I will make a couple comes with David's
chapter, it's a thing to comment comprehensively.
I encourage folks when available to everybody to take a careful
read on it.
I did a GRAED great job of providing a lot of examples from
the paper that will say, a couple of basic arguments that I
took for it.
Were from financial intermediaries must or do -- I
think it should be must, take on basis risk or reSid YALG risk
that in the aggregate can expose the system to large
underappreciated risks.
Another thought is that borrowing, lending and hedging
can lead to continued -- continued claims that exacerbate
leverage in extremists and notion of risk-based leverage
and how to measure what that sort of conditional, elasticity
of that risk is important challenge.
As for basis risk, I think it's important to point out that not
all market participants intend to hedge every position.
So the purpose is always take residual risks that are --
either reflect a view of economic fundamentals, or market
fundamentals, reflect some particular inside that they may
think they have about the mechanics of the market and the
settlement process, et cetera.
And I think it's important to make sure that those risks are
understood and properly priced.
There are also risks that are the inevitable consequence of
incomplete markets.
Some folks who desire to hedge, that using David's example, may
not hedge the exact position they have.
One of the features that run up to the crisis was, as I once
learned, the best hedge is the sale.
Second best hedge is putting on a liquid position that offsets
something that has become a liquid and cross your fingers
that they work together.
So there's a lost that that happen pre crisis, and creates a
basis risk that may be underestimated or underpriced in
And then, of course, there are the risks that people take on on
>> So, this notion of basis risk, there's a lot of basis
risk out there, leading to potentially unacceptable risk,
and the aggregate is good.
I think we ought to focus, though, on those risks that are
not desired, and that are a function of either incomplete
markets or the -- the clearing of settlement practices that are
out there.
David does point out in the paper that some of the
innovations have reduced some of those latter sorts of risks,
including the trend towards multi lateral letdowns of
derivative positions, tear-ups, and the movement which obviously
is spurred by Dodd Frank, to move from bye lateral margining,
to central counter parties.
Those trends are important for reducing that lateral risk,
there's a question of how to reduce that risk.
It's pretty clear that it does.
The -- we'll all start to mitigate that kind of residual
risk that builds up in the system.
So, a couple of concluding remarks.
This is good work.
There is a lot of ideas that have -- that have been brought
I do think he would benefit from sort of crystallizing his
version of pricing forwards and using forwards ever as an
indicator of systemic risk, the link to practices and
commodities, derivatives is an interesting one, I, for one,
would benefit from seeing more clarity, on that particular
It would be also interesting to see if it's testable.
To see how this approach might have elicited information pre
It would have sent signals to behave differently than folks
As far as risk-based leverage, the one thing that's
mathematically true, was experiencing losses.
We'll have an increase in leverage, all other things being
equal, I think differentiating leverage that increases as a
result of losses from leverages that increase as a result of
growing claim, would also be a very useful addition.
But having said all of that, interesting work, I would like
to see something actionable as a supervisor now.
I'm very keen to see actionable good ideas.
That's the sum and substance of my comments.
>> What we do now is thank the authors for a very wonderful
paper and discussing their insights.
We've decided that the format seems to work quite well.
We now open this up for questions.
We thought we month WO maybe take several questions all at
once and begin to process that way.
Are there any questions from the floor?
>> Are we using a Mike, or --
>> The earlier session talked about stress testing done by
supervisors, and regulators, I'm just wondering if the panel can
discuss the connection that might be between the stress
testing that's done by the regulators and supervisors,
stress testing that's done by firms.
What should be the dialogue that happens between the supervisor
and the person -- the people who are being supervised?
You know, what should be the nature of that dialogue?
Related question, related to risk governances, should be part
of the process.
Broadly and with respect to specifics of the risk management
>>> Very interesting.
And I have a question that I would like to pull up.
Really along the lines of what Dick asked -- I think I would
point to coming up here, I would be curious to hear from officers
what theythy of the potential feedback given back to the par
tis pants, who participated in the stress test, for example.
One of the things that became clear talking about a stress
test and actual activities around that, when they come to a
stress test.
They classify or characterize assumptions about how feedback
might work in the economy, under different scenarios.
I think you can make a credible case, that a stress test could
give a sense of where you might be in terms of the first order
moved and how you might be stressed.
But it doesn't tell how he might have responded.
And other participants might group in a similar fashion,
similar stresses which makes or firms distressed and have to
sell, might have balance sheets to absorb, that's not really
something, unless I've missing how you do your stress test, you
put in a lot of details in your models, I would be interested to
know if you can't do that.
But if you can't, and the federal reserve, where you're
doing the stress test has adequate market information,
what would be valuable to you, have been back you to.
Could they be valuable about congestion in holdings and
short-term funding, kind of goes to the question that's been
raised here, how much should the banking position bekept, out of
the public view.
If it's more in the view, could that strengthen the view that
might be the case?
That might actually help people have incentives to not take
certain type of positions on?
I would be very curious to know what you think might be a value
from today's environment, brought back you to.
Obviously, has a vision of getting more and more views
about transactions, there's additional information to be fed
Future views on either one of those values now.
>> There's a lot to process there.
Why don't we take those two sets of questions?
Talk a little bit about stress testing and regulators.
Advisory process, governance issues.
How you might consider modeling feedback effects from the stress
>> I'll start with the second question first.
I want to have a basis of understanding the risk
And ensuring that there is a clear communication up and down,
and the strategy that firms are pursuing is first of all owned
by the board of directors and second of all, implemented in a
sound way through the organization, the organization
should have a good idea what they are trying to do and what
risk taking is associated with those activities, and a good
process for checking themselves, when the business begins to
deviate from that path.
So that's -- that's the governance around the risk
management processes.
It's far more than making sure that the numbers add up, it's
about ensuring that the concepts that drive the strategies are
appropriately linked to the positions that the firms are
taking on.
The stress testing done by supervisor, I think that this
was dealt with by Kevin on an earlier panel.
Lots of advances, and the manner in which we have -- the federal
reserve has developed the stress test under way as we speak.
Communication with the firm, make sure we're dealing with the
same information, stress the same positions, but the great
value of supervisory -- the supervisory stress test as
opposed to the specific stress test is the ability to apply
comparable stresses and models across all of the institutions
and see what the results will be.
I will note that interestingly, at least for the training
stresses that we've used, that the firms are on a regular
basis, applying similar stresses so there's not that much
different between the stresses that are employed by the
supervisors and by the firm.
However, I would say, supervisory perspective, very
useful to see firms to perform a variety of stress test, not just
the ones deemed relevant in any particular annual supervisory
>> On Mike's first point, I could not agree more.
The requirement for supervisor's is manifest.
When we were asked to come up with a risk appetite statement
fairly recently, that was the first time we put in place a
formal risk appetite statement, it was in direct response to
request from the primary regular U later and incorporated in that
risk appetite statement was a supervisory stress test which
was designed to provide our board in this case with comfort
that we had resilience against some pretty adverse
circumstances, and that was presented to the board and the
fed made sure we got the message, they came around and
talked to our board.
The feedback loop was pretty explicit in that regard.
Similarly when it comes to the stress test and bifurcation or
the interrelations between firm specific and supervisory, we
have no pride of authorship at all, and our -- the first icap
processor that we went through, we thought, gee, that's pretty
exhaustive and tortious, first of all, thanks a lot and it was
also very effective, and we knew we would have to do it on a
repeat basis, probably not as frequent went, or soon
thereafter, but probably the latter not the former, that we
needed to put the process in place to be able to replicate,
what we had, the successor to that as an ongoing tool in our
own risk management.
>> We Vey variety but feed them through the same process.
It's all part and parcel of the same.
>> If I can specific, cliff went down the path bravely of
cognitive bias, I have intuition that there's some kind of bias
that makes stress testing easier to digest.
Reverse stress testing being, instead of saying, here's the
scenario, what happens with the scenario, take the opposite
view, what's the scenario which you go bust.
Whatever the cognitive bias is that makes people overestimate
airplane crashes, I'm sure there are names for those thing, may
apply to the reverse stress test.
It's difficult to attach credibility, this was discussed
And is completely and plausible.
And becomes much more vivid and scenario described is the one
which you go bust, focusing on what sort of factors would lead
you to that situation, might be for cognitive as well as other
reasons, and a useful component of stress test.
>> I might jump in here, as for governance, the first thing that
comes to mind, you never do things for your regulator, you
do it because it's the right risk decision to make first.
A lot of things I talk about first these days is hair cover.
Some of the places I was at, I refer to them as my sabbatical,
since I was only there for like seven or eight months.
After having informed folks if things like what they should put
up for the third quarter 2007 loan loss profession and they
didn't like to hear will about those things.
And I used to talk about the old examiner in charge.
One of the institutions who is still my friend these days,
about this issue, where are you guys, providing us with air
Look, we're looking at the bank as being able to make their own
decisions about the risk management that we have.
But, as I described to him, that sometimes having that air cover
from the supervisory standpoint to be able to look at major
trends in terms of the direction of the risk infrastructure,
quality of the risk infrastructure, I can go on and
on and on about stories where actually sometimes my
organization was referred to as a cost center, in fact I would
have to lobby for my budget from the business unit which seems
inherently perverse.
So, from that standpoint, having a supervisory role that is
strongly connected to that, and I think we need to do things
like strengthen our risk-based deposit system to account for
better risk management infrastructure, I think we need
to look at that.
And look at other supervisory systems out there.
Among other things and I'll shut up and turn it over to the rest
of the panel.
>> I want to reinforce the point that Mike made, I think it's
entirely legitimate for supervisors to look at the
governance of risk management processes of stress testing, of
the whole tool kit that's used to manage the firm from a risk
oversight perspective and I'm reminded, I think the bay back
to that kind of qualitative assessment of governance
processes is much greater of additional regulatory effort,
validating our quantitative models and reminded something I
heard the governor speak about.
He described the flame work has having a slanted roof.
The tall pillar is the first pillar which has the
quantitative models and then the supervisory pillar, the second
pillar was shorter and the shortest of the three was the
market exposure pillar, pillar three, I think we need to right
the roof.
That then it out and there's much more benefit to putting
more emphasis on pillar two, which is the spirit of what Mike
described in Dick's question.
Briefly, on feedback to market participants, I think echoing
the theme Gary started with this morning, there's a tremendous
amount of value in disclosure, to the marketplace to allow the
market, but other banks to understand what the results
looked like, and get at some of the issues you were describing,
and I was mystified last year that the "C" car had the secret.
The "C" car results were not a relief.
Having been on the receiving end of that.
We all would have benefited by having the results put out in
the public domain and I'm glad to see that this year, that is
the way the "C" car result was handled.
There's going to be pretty extensive disclosure and we all
benefitted from that.
>> I'm going wrap up this segment of the responses and I
would like to start out by saying, there's always going to
be attention between disclosure and transparency, right?
>> Full position disclosure versus the amount of position
disclosure and what actually translates into functional
transparency for both the institutions, as well as for the
regulators, and others, responsible for oversight.
I think there needs to be enough disclosure, to provide
regulators with the forensic detail needed to challenge
conventional wisdom, and as a challenge, consensually
challenge made by market participants during particular
regimes, and I think that serves the risk manager in the firm and
senior management in the position.
The ultimate long-term viability and profitability of your
institution while be what dependent on shared conditions
are thing us can't observe directly.
And therefore challenge.
I'll give a example examples and end with that.
There was a very popular scenario, as early as 2001.
And it was something that came across my attention during post
9/11, a scenario with regard to a large counter party, large
firm that they saw, largest counter party and they said look
if that firm were to go under, what would that do to aaa asset
repos from a liquidity standpoint and balance and
stress standpoint.
I thought it was a clever thing to do.
Turned out the institution they were not concerned about was not
the institution they should have been concerned about.
The other thing, disconcerting, they didn't apply that same
stress to the super senior positions which is at least at
large if not larger than aaa repos.
They said it's zero risk rate and we don't have to fund them
in cash so it doesn't matter.
Therefore unfunded liquidity as an institution.
So, there's an example where, more forensic detail to the
regulator could have said, guy, is that really enough?
Should you maybe think about this?
Should you maybe think about that?
Another area is cash equivalent, I've been testifying on auction
rate securities and bid support.
And one of the things, with an unanticipated spike in funding
rates, the pressure to support the arf resulted in a collapse.
Those were sitting in money market funds all over the place.
That's one of those situations where I think the firms and the
regulator, could have had a productive dialogue around when
do these things start becoming ill liquid, not because you
can't support them, but because you don't want to.
We SA that with canable repo.
We saw it with CDOs.
What happens when you have those in place.
There's an argument, whether they are eligible anymore.
To the extent that's the $80 billion sieve.
We start to challenge these thing, there's a nice
partnership between risk management and senior management
and firms, without having to disclose everybody's positions
in every other counter part.
>> About ten minutes left.
We'll shift gears a little bit.
And see what questions that we have with the time we have
available to us.
>>> So in 1998, did this come up as a -- property of importance?
In all of risk measures that you looked at?
For example the value of risk does not satisfy some in
particular, of a good risk measure.
On the other hand, the conditional V.A.R. SAES such a
Did you address that particular one in terms of extreme value
theory, the difference between all of these short forms?
All of the different theories?
That's it.
The risk measurement itself, if that property is not important,
did it show up?
Is there a practice of example where this is irrelevant or was
it a source of accumulated risk as David was perhaps saying
about the accumulation of risk due to errors in measurement?
That would be a model risk problem and is it
underestimating or overestimating which leads to
the question, what is an appropriate level of risk
tolerance that you would like to have in order to not to leave
inefficiencies in the market.
>> I can say that I do cite that paper, and I mentioned in
passing, that spawned large literature.
That's something I chose not to go into.
It's important theoretically.
The thing is, practical matter.
The main concern is it's not so additive.
You can get a larger V.A.R., by combining two that you would get
I think it's rare in practice, even in theory, it's known that
there are many situations where it's known that that, in fact,
cannot happen.
I think it's important but not a first order of concern as a
practical matter and reason to abandon V.A.R.
>>> Thank you.
The gentlemen next to the camera and on the left side.
>> So, I heard a mixed message from the panel as a whole about
whether it's feasible for risk management to deliver what we
need it to deliver.
Continuing the salad analogy, the nonfunction sector throws a
lot of credit risk, mismatch, interest rate risk and spirits
or risk premium into the squeeze Nauert and it gets all mixed up
into the financial system and risk management play ace key
role whether garbage or salad comes out.
So, Paul, I think, and Craig, were fairly optimistic.
And stepping back, what David and Michael said, if I was a
normal person, I would say there's no hope of reliably
getting salad out of the system.
So there are historically different ways to do it.
It used to be that you knew where the interest rate was,
where the credit risk was.
It was all in lumps.
That biggers on steroids.
Imagine your grandchildren running a restaurant in the
Would you rather have them be in the vicars world or in a world
where performance or risk management is going to be key to
the stability of their business?
>> I know that's a nasty question.
First of all.
The market is already complete.
If you don't have financial intermediary, they can't
Part of that is taking risks.
So, and they are the challenge that you can't put the Jeanie
back in the bottle.
All right?
Now that we know there's a concept of fair value and
dynamic price genius, I don't think we can shove the Jeanie in
the bottle and say we'll ignore it.
>> You agree, he says we're never going to get there, but we
keep trying.
I think there's that constant tension in my own professional
Which I heard senior management say I think we've got it.
I've been concerned.
The role we have to take, a number of institutions is be
that guy.
Constantly looking for sort of a weird hitch in a structure, dual
trigger, triple trigger, whatever it happens to be, where
it doesn't quite function under certain scenario, so, I think I
wouldn't say I'm completely pessimistic about our ability to
get our arms around stuff.
We fly planes much more safely today than we did, 25, 30 years
What happened our technology has gotten a lot more complex and
level of expertise you need to have goes way up.
And they went a lot more times firing chickens through GE
engines and we still affect people's marketplace.
>> Since David and I are tag teaming, I would jump in and
say, I do believe it's challenging.
I do not believe it is hopeless.
And I would say, that a lost things we've been talking about
both on this panel and at other times in the conference about
better governance, about having more resilience in better
capital liquidity level, better data to understand where the
risks lie and how they made change under stress.
All of that is critical, as well, managing a system that is
no more complex than it needs to be.
It has been at times more complex than it needs to be
because of a variety of incentive problems that are out
there, a variety of structural issue, to simplify, to provide
that intermediate, necessary intermediate YAGS to the
economy, get better risk management, get better
governance, get better data, you have a better result in the long
>>> What I don't understand, maybe you can help me with this.
Why the marketplace itself doesn't demand better risk
I can think of three possibilities is one, they don't
really value better risk million management.
They've done cost analysis.
They think it's not worth it for us.
They don't know what good risk management is.
Or more likely they don't know how poor it is.
If it's this last one, maybe more frequent, more
transparency, about mismanagement and the structure,
is what is required.
Not simply for the regulators to know it.
For that to get back into the marketplace.
Perhaps regulators play a role getting back into the
marketplace and quality of risk management of various banks.
It's easy to blame the market.
A lot of us didn't know how bad risk governance was a few years
I think it took the prices to really demonstrate what the gaps
in governance were.
So there's a full sense of security that percent VADed the
whole industry, you know, the it was true of inside firm, true of
supervisor, I'm sure the market was lulled by that same sense of
I suspect the market will focus much more on governance going
If you look at returns of bank stock over the last ten years
they haven't been particularly heroic.
And I think, you know, the boards of directors -- I comment
on the boards earlier and their role, I don't think the
boards -- the companies that bailed, or became distressed are
proud of their record.
They demade they do a better job going forward.
That's another way the might do it about the role going forward.
>> If you look at where banks trading today, you can say that.
One of the discounting anning over the industry is risk
governance discount.
>> That might speak to your third point.
I think it's hard for -- you know, a lost things obviously
drive that P.E. and it's not just notion of risk governance.
It's really transparent in market.
It's not that transparent to market.
It's difficult to judge markets in governance.
>> Get asked about risk governance all of the time now.
About other constituents who had not previously expressed
specific interest in that field CH direct answer to your
question, they're greatly interested.
It's fairly difficult for them to get clear messages, and
clear -- form reliable opinions in many case, just given the
complexities and given the fact that firms quite adept at
presenting effective story, whether they actually have
effective risk management processes or not.
That's the reality.
My last comment would be, there are motivations and consequences
that we have to address when we are transferring or repackaging
risk Bice trying to recharacterize risks or split
them up and change the nature of the instrument.
The question you ask are you still tracking the risk?
A way that reflects the reality of the risk?
It's still a duck.
I don't care if you dress it up as a two headed turkey, or a
bald eagle.
It's still a duck.
You say if I move it from "A" to "B" and I'm not marking it
anymore, because I call it something else.
Insurance swap is an insurance contract.
Did it change the nature of the instrument?
I don't think it did.
>>> Good.
We come to the last session of the day, which we look at the
modeling of financial directions.
We will have presentations and an opportunity to discussion
I give a one-minute introduction to talk about the discussion, so
Charles goodhart can meet with no introduction.
He's done it all.
Has been a PRE fesser at LSE for a long time.
He was bank of finland, including as a member of the
policy committee, and he's always great to listen to.
Dimitri works at the bank of England and worked with Charles
on a number of papers, and I'll divide the presentation between
the two of them.
Then two discussions, the first one is Tobias Adrian, who works
at New York fed.
And has written a number of important papers, during the
crisis, explaining the importance of how this affects
events and made things worse, then the other discussion is
John Geanakoplos, he also has worked on wall street and has
talked for a very long time, but it turned out it was one or two
decades or something like this, right?
So he's very well positioned to discuss the paper.
With no further adieu, Charles.
>> Thank you very much, thank you.
Even though you renegaded us to the get toe from which all
practitioners have fled.
I need to delineate that you and I are in the list of presenters,
are the only people whose pictures in black and white,
which just goes to show that compared with the rest of you,
macro economists are colorless.
What I am going to do, I am going to say just a few words
about the general approach and theory that I think we should be
taking to modeling, while I'm going to leave it to Dimitri to
discuss in detail, and in power point where I have not.
The particular form of the model, to involve, include,
financial frictions, I see particularly, with my
encouragement, are being put together.
As I'm sure that all of you know, the standard macro
economic DSG models played zero role and no financial in that.
Effectively, the models were real business models, with the
inclusion of wage and price fictions.
I said wage and price fictions are evasive and continuous.
It's least arguable.
Financial frictions are those that have driven the economies
further away on occasions, if you think of 1981, '92, '97,
'98, 2007, 2009 and what is going on in Europe.
All these suction that it's been the major, probably the most
important ingredient causing our economies to deviate from
They are able to model such fictions, with the view of
preventing such crises in the future and when they occur
easily resolving them.
No doubt the main driving force behind the more serious
financial frictions occurs because the potentialality and
at times the actuality of default particularly major.
Models and there are some include corporations but don't
allow default are therefore inadequate.
You think of the key default over history.
Lehman brothers, some remember Knickerbocker, and few of us
remember Germany.
Most of the crisis have been related to financial failures,
not to failure of nonfunction corporations.
If you don't have default, there's no function fictions or
no need for financial mediation.
Indeed the logic behind a world without financial frictions,
would be that there is no need for money either, because
anybody I.O.U., would be immediately and perfectly
acceptable in exchange for any purchase of any good or service.
Now, it's actually this assumption of no default that
has been assumed in most previous standard DSG models and
it require ace number of totally plausible behave yoral
assumptions, first existence of complete financial markets,
including the ability to hedge against unknown unknowns.
Second is the assumption that all agents are so ethical, that
TWHOE in all cases repay their debts in full, even when there
is no constraint or sanction of any kind that might prevent them
from doing so.
The appreciation of the facts are stated in the world and
history, make it's sense to incorporate at the heart of any
general Librium analysis.
Notably and especially the default.
That has the ability to have the financial fall of money, and
financial liquidation.
Both of our discussion emphasize swings in leverage, swings in
leverage are intimately closely related to changes in the
appreciation of the probability of default.
Now it's true that there have been a lost studies in a basic
partial equilibrium framework which avoids dealing head-on
with default.
But it's almost always explicit in certain situation.
Studies of financial acceleration, there would be no
need for collateral at all nor any risk premium depending on
the equity value of the borrower, if it were not for the
possibility that the borrower might default and the exact same
argument holds.
It is again true that the easiest way to incorporate
financial frictions my a DSGE model is add a credit risk
premium for the kind of variable that Phillip Hartman was
discussing, describing, yesterday afternoon.
And add that to the expenditure function and perhaps also to the
reaction function, but the very that approach, useful and
appropriate in certain circumstances such as in
forecasting, cannot provide an appropriate model that factors
cause financial crisis and cannot therefore be used as a
basis for analysis how to handle and improve financial stability.
So, the core of our analysis resides in making default by all
or any agent, central to the construction of our model.
It's not easy to do.
And the resulting model construction is perhaps more
complicated than would be desirable, in principle.
Nevertheless, our modeling approach is also flexible and
can be adjusted to take into account differing issues and
questions, such complexity arrives in two main sources,
first concerns the question of how to model default isself.
Here we follow the father of this in this field and very
cohorts whom John has been one.
And it's measure by the proportionate repayment rate.
That characteristic of the state and nature.
The other main problem the potentialality of default
requires a number of separate agents in the model increases
very sharply.
Once default happened, it is no longer possible to use the
representative agent paradigm.
When one tries to do so.
That whole sector defaults and disappears which doesn't seem
plausible or there is no default at all which takes us back to
the standard DSGE model.
Equally, the existence of default provides a rationale,
for including at least one, and depending the question to be
answered, possibly several types of financial problems.
Dmitry and I, published many things in a book.
I'll turn to my colleague, Dimitri to give you a power
point related idea of what we're trying to do.
>> Thank you very much.
For the wonderful, inspiring words that you put together.
I hope the home stretch of this will be equally inspire, what I
will try to do in the few minutes that I have, try to
bring more information on what is set out.
Without giving the long day, without being too technical
about it.
Trying to communicate the building blocks of what we're
trying to do and hopefully inspire the remaining audience
to read the paper that I will talk about a little bit at the
end on their way back home.
Next slide, please.
So, as Charles was mentioning, basically the key idea of her
modeling process is the frictions that necessitate and
make necessary, and make mandatory, so to speak, active
and indigenous default.
The first friction, they the inability to commit.
Inability to commit, the sing the trust and emergence in the
economy, therefore, they impose penalties for default.
See we follow the way they've modeled default, so as to
increase the efficiency of the orderly sanctioning of the
Second friction that Charles already talked about is the
distinction between complete and financial markets.
If there were no missing financial markets, in loan terms
were comprehensive.
That means any penalty for default could be forcibly
applied, and severity would be P included on demand and result
>> In other words.
When markets however, are incomplete, in other words when
share is not possible.
Then allow for positive fault in equilibrium can be well for
improving and this is a well-established instruction,
whereby if you have one slight possibility, one slight and rare
state of nature, whereby somebody defaults, if you
preclude the default all together by making sufficiently
severe penalties, jail or decapitation.
Then you would kill, as well, decapitate as well, risk-taking
behavior all together.
Therefore, there is a trade-off.
Fundamental trade-off that perm YAETs all of our work before the
behavior and dead weight loss of default.
Finally, when you have institutions and depart from the
very convenient financial intermediary to do things that
do not accept passively.
Whatever they do.
Then you improve opportunities, and risk sharing, and then you
breed different types of lending and boring opportunities.
Therefore the action of friction, and default in
equilibrium is the integral part of the function of the market.
>> Next slide, please.
>> Now, since we can justify and argue about the necessity of
default, and frictions that generate the possibility of
default, then we have to underline -- to pinpoint the
reasons that are caused by default.
First one is the default can lead to an efficient level of
default and I can get more hazard.
In other words, no we have margin of benefit of defaulting,
versus margin and cause of defaulting, that makes necessary
the introduction of a full-fledged monetary sector in
the economy, to be able to balance out and depict and
delineate the margin of benefit cost of default and margin of
benefit is what you can consume.
Because you've defaulted and margin of course is associated
directly or indirectly with the bankruptcy and penalties imposed
on the investors and apropos here, we have to mention,
liquidity and default is something that Charles already
underscored in many of his -- much of his work in our modeling
approach, that liquidity and default are an integral part of
the two sides of the same point.
It's like trying to monitor a policy, without having a sector,
trying to perform an operation without a patient being present.
So, the financial system, the second friction is the financial
system is an amplifier.
So therefore the drop in supplier credit, due to their
losses, further suppresses prices and income making.
There is reminiscent of the usual mechanism.
That a long time ago, described in his SECHL Senator paper in
And financial institutions, resulted in being transferred
throughout the economy and being in other sections of the
Therefore, one can start talking about contagious or systemic
To some extent.
We talk about general liquidity.
The last -- this is what we've already mentioned and emphasized
the interplay of liquidity of default.
One won't exist without the other.
Without the frictions, would render as a monetary sector.
Likewise, default, a feeling of obligation, contractual
obligation, would be always a modeling device out of
equilibrium that never materialized in equilibrium.
The key thing in models is not to allow for the possibility of
default, but to have active default in equilibrium in our
And, more importantly, what is generated from the effects
justify and underscore the importance of regulation, what
does this mean?
If you have default, then the liquidity you reach is not
simply inefficient due to the fiction you've introduced but
most importantly, constraint.
All of the work has been done by John based on the idea.
Therefore you reach the surplus and therefore given the
frictions, regulation, or government policy, or policy in
general, may improve final allegations.
And finally, it becomes a reality, and more importantly,
the identification, of the specific channels through which
con stage onis transported through the economy.
Next slide.
To make it more informal.
I talk briefly on the general characteristics, that one
benchmark model of ours possesses.
That is basically our work.
So this model, whereby the financial system emanates, from
he default.
We have credit crunches.
Therefore our financial stability regime, and we have
financial FRAJ IT, the multiplier, then you can also
talk about a financial stability.
Multiplier or accelerator.
Our system allows since we have financial intermediate years
that are active.
Therefore you impose upon them regulatory measures.
We are in a position, since we have ingredients to start
talking about how it can be imbedded into such a model and
we can assess and pique their impact.
Finally, since we have, as I said before, liquidity, and
default, we can talk about liquidity in a very precise
It underlines liquidity, underscores leverage.
Our model ingredients, our model that can flash out, namely we
have it two goods, and we move away from potatoes and tomatoes,
so we have now potatoes and houses.
So, we have two source of values, and the important thing,
since we have collateralized loans, we have nation.
It was the key ingredient when we talk about default.
One, the mortgage-backed security and our timing would
consist of the simplest possible model that escalates to
something like 70 equations with two periods.
Period one is the present.
And we have two states of natures in the future and we
have three types of hassles which basically defer from each
other with respect to their preferences and indocument, we
have houses endowed with the consumer.
And the first in the future and finally we have first-time home
buyers that emerge in period two in the future, and so they
sustain the house demand.
And the key innovation with respect to the past war, is that
we have two types financial institutions.
We have standard bank, which has a high risk applaud version,
which is very risk capitalized.
Very big balance sheet.
Then we have an unbanked financial institution which we
take as proxy, representing the shadow banking system that
has -- that has the lowest conversion, and we've seen the
modern -- how important the presence of the banking system
in the commencement of fire safes.
>> Finally, we have central bank that have short-term loans.
But our central bank is made of character and can fight off
Next slide.
Modern characteristic, basically we investigate uncertainty with
respect to relative quantity.
Of potatoes versus houses.
Not only are we there.
Also we have some initial.
Given we have -- given that we have a well-defined transaction,
madder in their are quite important.
Second, what they tried to do, all of the standards.
The consumer will end.
I cross future states of nature.
What are intermediate years, basically improve smoothing, but
they have a cost they can fire, very add first financially.
Actually, this kind of trade-off of improving TEM pear smoothing,
versus am PLIfying the cost, is admission of what's happening,
in fact what's happening, the debt crisis in Europe, let me.
Finally, the regular U LAGS that we have introduced, serves the
role of restricting -- next side.
In a very well defined way.
You have adverse productivity, you have some sort of adverse
affect on the price of collateral, or for houses, then
you have -- you generate the process of default mechanism.
And thus instigate the banks, that in turn generate and this
gives you potential.
And this is reminiscent of downward spiral finally, we have
present the distortion.
On a penalized list and you can support your wealth.
This is basically present.
First, we have -- first we examed, then investigate
marginal requirements.
Capital requirements, you can think of them to the shadow
banking system.
And finally, they are nonprovisioning, what the system
had introduced a long time ago.
Then we start investigating, or trying to address the question,
wetter this regulatory are complementary.
Next slide.
Basically, it can be summarized.
Crisis prevention, crisis management.
In practice, what we investigate, in other words, in
other words, they make it less extreme.
If and when it occurs, in other words more differently to reduce
the variance of the site.
In this category, one can think of marginal requirements.
Loan value requirements and potentially, capital liquidity
requirements on commercial banks.
Second challenge, thought as crisis management will show up
the banks in the event of a bust.
So, incest on having capital, and liquidity requirement, are
part of this sort of regulatory intervention, however, in my
case, since we tie one of your hands behind your back may
exacerbate fire safe and may make them much worse.
Finally, SWREET regulatory organization to our regulatory
interventions and this is basically the regulation
practice, leaning against the loan.
And this regulatory macro tombs refers to us building
Next slide.
Some of the conclusions, some of the conclusions that we've
already obtained, from our model is basically, modeling the
frictions carefully, and being precise, in order to idea the
presence in the system, which -- why, argument fates.
The second kind of robust conclusion, is something that
was mentioned, that was mentioned yesterday by Phillip
Hoffman, is that we focus on the challenge, on the transmission
challenge of financial fragility.
Through which they operate, and we conclude it's probably more
important than focusing on institutions and margins.
So what happens is important to regulator.
Third, the policy affects the short end, while the regular
policy interdevelops at a different stage.
Then we see the different distributionary affects that
regulatory tombs have when compared and contrasted with
monetary polls.
Therefore, in the end we can conclude.
These have to be revisited and reintroduced.
In other words, since we have stability, we require not just
capital seems they are not likely to be sufficient.
This is the point that very adequately has made by Phillip
yesterday, but also in the recent paper by Dick and Charles
And, next slide.
Overall, overall, basically, our approach, what it does, it
proposes, they bring together liquidity and indulges before in
the center of analysis.
Institutions, and institution is important and institutions makes
a difference in the results that one gets, and the foundation's
regulatory intervention.
The object is to propose framework to monitor a
regulatory policy, in an integrated model of financial
A model of financial fragility and financial instability.
Our work in the next slide can be summarized, all of these
papers, for your perusal.
Of course I won't talk about them.
There's earlier work.
But the work on liquidity and default has been done with the
usual suspect.
Then, official legislation theory hasn't done again to
other wonderful colleagues of ours.
And we also have the opposite kind of argument.
The moment of the type of argument.
This is work they've done on learning and default with the
usual use SPEKTs and finally, the next project that we're
working on is to take the finalized models and open them
Juan Martinez of the bank of Chile and myself will have given
an initial shot in this project.
Thank you.
>>> Thanks for having me here, and thanks for writing the
It's a pleasure to read and a pleasure to review.
I'm from the bank of New York, what I'm saying are my own views
and not necessarily the views of the bank of New York.
When I e-mailed Charles last week.
He said, you know, we've written four papers recently.
Why not review these four papers.
Then, on Monday, Dimitri sent me an e-mail with another four
So out of the eight papers I picked one which is the one
Dimitri focused on which is by Charles, Dimitri and several
others, we'll call it the DKGV model.
We'll go a little deeper in the model.
Putting a different perspective on the model than Dimitri did in
the model.
The structure you need to think about these four different
tools, okay, so it ends up being quite complicated.
But I think it's the minimal ingredients.
So the basic structure, you have some -- the home owner.
You have first-time buyers, and essentially, essentially what is
said, putting deposits in the bank, and there are -- then the
farmers are borrowing from the bank, so that buying houses are
funded by mortgages.
So the bank can secure advertise some of these models.
Which is like a repo conduit.
But they're funned by a repo, and repo is is on the bank's
balance sheets.
No refund of money here.
So you go to the next slide.
Then you can talk about these four different tools.
So we have a look at the bank balance sheet now.
On the other side.
They're cash, they're mortgages, and then they are like reverse
repos that are collateralized.
That are on the balance sheet.
And on the liability side.
You have deposits, central bank loans and equity.
And the tools basically regulate every single -- not every, but
every single part of the balance sheet of the bank.
So there's a liquidity ratio that is defined as the cash plus
loans, divided by the cash plus loans, plus the MBS reverse
So it's sort of like a translation.
Into this framework.
Then there's a ratio, which is a constraint on the borrower,
which is basically constraining the value of the mortgages, not
to exceed.
A certain amount.
Then the constraint on the hair cut of the shadow banks, so
there's constraint that they can take out.
And then the constraint proportionate to the equity
divided by the reverse repo, because, all the other assets on
the balance sheet, have zero risk rates.
The tools are attempts to discuss not in the paper.
The funding ratio, debt to income ratio, stress test and
various central bank policies.
That's like already enough going on here.
We'll go over each of the four tools in details.
They don't really have provisioning and tools.
Go to the next slide.
How do policies mitigate risk.
They also structure policies such as the creation of CCFs.
The improvement in the quality of capital.
And other financial structural policies, they're more about
circumstancely Cal policies, this is a dynamic model.
And policies over the cycle.
So how policies were.
There are two channels.
So the first channel is tightenening tools, against the
bobble and credit booms, so, the racial ale for that.
The financial sector, has this.
And they may want to lean to that.
The market potential policies is very much tied to the financial
Then there's a certain racial ale for market potential policy,
which is that by tightening market potential policies, there
is a potential tool that's increaing the resilience of the
system as a whole.
So the cyclical tool cannot increase it entirely.
By increasing capital retirements, you increase the
resilience of the system against future leveraging.
So, what we're going see, for each of these four tools, the
set of risk return tradeoff is going to play a role.
The risk of the tradeoff, leaning against the wind, by
tightening market potential policy, you may credit growth
more costly and slowdown credit growth and by doing that, you
are -- you are increasing the resilience of the system.
You get less risk at the cost of less growth.
Let's move to the next slide.
So, the motivation that is usually given, one that creates
above, those Brazilian, and, two, it leans potentially
against greater risks by raising the cost of capital.
So what is happening in the paper?
The contribution of the paper is really too sort of rigorously
underline the realistic arguments that policymakers do.
To look at the general.
So indeed, the type of capital requirements reduce the total
amount of lending and generate regulatory arbitrage.
There's a movement into the system.
As a result, they consume this and facelss risk.
This is the risk of market potential policy.
Now, one of the points that the papers is making and this is
pointing out, capital requirements might not be a good
tool to lean against.
Because when you're in the boom.
Capital is also -- in other words to lean against a boom.
You have to raise capital.
And -- so we're going to look at the assignment but this is not
going to be the case.
Lastly, there's a differential impact on different types of
agents, so by the capital requirements, it's generally
good Fort people, for the home owner, because they are
protected against default, but it's bad for the borrower.
Now, what about evidence and incrementation challenges?
There's very little evidence on cyclical market potential
variations in capital.
Risk weights have been used in some countries such as Israel,
but for like movements of capital over the cycle has not
been systematically used.
There's indirect evidence from Spain, from the dynamic
provisioning, similar to capital.
There's some indirect evidence from the U.K. and both of them
are sort of like somewhat favorable.
In particular, the evidence from the U.K. suggesting that capital
is indeed reducing credit growth.
Three provides a framework so, you know, the framework is there
in the U.S.
So, let's move to the next slide.
These are the variations.
I think it's comparable to capital policies.
So, as you've seen in the previous slides, there is a
institution from the banking sector into the banking sector.
So you might want to complement capital requirements.
>> And so what the capital requirements, you're really on
the core institution, in particular the banking
institutions, while hair cuts and margin, kept to a much
larger universe of financial institutions.
In particular, you know, trade-ups in future derivatives.
What does that mean?
So when there's their is an adverse shock, right, you have a
movement back from the shadow banks on to the banks, and this
is in turn affecting the banks, as the shock generate ace fire
The mortgages on the bank's balance sheets.
And so the insistence of the shadow banks are reinforcing the
fire sale.
So, hair cuts are reducing these fire sales.
At the cost of less lending.
So there's again a risk-return tradeoff and it's recent to the
The impact on welfare, in moving around hair cuts, is small and
vigorous in calibration of theed model.
There's no real evidence on systematic movements, and hair
cuts, in order to address micropotential policy goals.
There's some simulation evidence.
The basis for you, using hair cuts, there's a market potential
tool, the regulatory basis is not entirely clear.
It's not part of the framework.
And, so on the one hand, moving one's hair cut might minimize,
might take away some false regulatory advertising.
But might generate additional regulatory arbitrage.
You go from the bangs to the really shaved bangs.
There's yet another form of regulatory arbitrage.
So the ratios protect us against market price fluctuations.
They protect borrowers against default and price.
What is happening in the paper.
So what they do is reduce higher sales.
And shadow banking instability.
They have an effect against the borrower, borrowers decrease
wild lending increases.
Again, problematic.
I'm surprised at the boom, so the ratio constraint is left in
the boom.
The challenge is regulatory advertising, again.
It's very difficult to conceive a system where you couldn't get
around regulation, in particular, conforming
mortgages, there are many ways to get around the element of
conforming mortgages.
There's some evidence from Korea that does point to the
usefulness of the policies.
In particular, one of the interesting results from Korea,
tightening ratio, tends to have an impact of expectations about
the policy.
Expectations shifting.
Let me finish with the liquidity coverage ratio.
Now, what is in the model is not what you find in the model three
So the policy is they're protecting it from shock.
It's a cash buffer held against the stability side of bangs.
Run interpretation.
It might slow down the liquidity crisis.
It wouldn't run out in one day but would be at least 30 days.
So the way the liquidity the way it's used is not such a tool.
It's more used as constraint on leverage of the financial
One of the things we're finding out.
Which I have not seen elsewhere.
It would be a good preemptive fool.
Because you remember, what I said earlier, when you're an
Enterprise possibly, they tend to be inflated.
Eventually very easy to fulfill capital requirement.
Can liquidity ratio, it's the opposite.
At least liquidity ratio as defined.
The assets are denominator.
They're actually making it harder to fulfill liquidity
So, another thing that the paper is pointing out, having a tight
liquidity requirement in the bust is emphasizing fire sale.
So you want to loosen liquidity requirements in the bust.
This is also something they want to think about.
In terms of evidence, there's not -- there's not much
empirical evidence, in terms of implementing a, there's a
proposal to have the coverage ratio, and funding ratio,
there's no element to those ratios in the current proposal.
So, I have a couple more slides, but I think I'm running out of
>>> I'm so old, I knee two pair of two kinds of glasses.
It's a pleasure to be here.
Thank you for inviting me too.
I want to talk about the paper and methodology.
Wasn't to first talk about Charles improving us to think
more about default.
I'm going talk about the details of the model.
In the slides that we got we had all of the equations.
I thought you would see all of the equations.
Dmitry did all of the equations, I'm going to talk about the
history of default to give you an idea where the methodology
comes from.
>> Next slide, please.
To emphasize what Charles says, may be P slightly different
>> What is really missing from macro economics.
We have to confess we didn't predict the crisis.
Didn't predict the stimulus.
It was not even before but after the crisis they didn't work well
most of them traditionally.
They were based on shocks.
The current technology.
Changes in technology.
And, even after the fact, when you have to fit the data the
model has to fit the data.
The crisis.
I'm not sure until recently those models did a good job
capturing what happened.
Modelers never feel the obligations of saying what
really happened with the economy.
It's not this thing failed therefore that thing failed.
It's all abstract shots which is bad.
There's no default on the models.
There's no changes in lending standards as a result of change
in default perceptions.
For example, no change in leverage standards.
No allowance of underwater agents or affect of them being
under water.
Of course all of this is changing.
How would you model default?
First thing allow for default itself.
The possibility some people might lose a lost money because
they wouldn't be repaid.
They wouldn't be able to do what they have before.
>> You now the second thing is perception of default.
What does that do?
>> First thing, people sense there's a higher risk of default
they charge a higher interest rate to make a loan.
There's an effecten crisis.
There's a second effect on lending standards which is
I think default is very important to introduce into
macro models because of the change in lending standards.
Not that the prices and interest rates change but standards
People ask for more collateral, for example.
This part, I come to this in the Dmitri model.
Changes in asset prices have to be taken into account.
>> If there are change in lending standards.
People can't buy as much, and therefore active might change.
More default means higher lending standards and less
If you want to pull all of this together.
Fine a way to keep the model simple and come up with a
solution to the question, how can supply equal demand for a
loan determine two variables.
The price and lending standards.
That's the model that you have to work out.
I want to go through a little history since we talked about
>> I'm not paying attention.
I'm in historical mode.
>> Default in corporate finance is recorded as important for a
long time.
Charles alluded to that.
In indirect effects and so on, who was thinking about that?
What did he do?
He said I have a model.
He said you can't buy anything unless you have cash in advance
which you can borrow from the bank.
If you always have to repay the bank, you're now using money.
Except you put money into the model.
Got the interest RAFT zero and got a great determine Nancy and
everyone expects it to be low.
Borrow a little money, everyone expects it to be my.
>> He then introduced default penalty.
If you don't pay it back.
You pay penalty per dollar that you don't pay back.
If it goes high enough.
With Chuck Wilson, he built the model which he said suppose the
penalty is low.
>> You get people defaulting.
It will be in default and you have to take that in account.
That's the first model that has money too.
And that's the starting point for Dmitry's model.
What is missing from that model.
The lender, the central bank is forced to lend what the policy
maker is given, he lens to anybody.
The interest rate is determined the demand equaling supply.
It's lenders without think IG about the chances of default.
>> In 1986, next slide.
Sorry, this model.
>> Is about inside and outside money.
SPURD by Martin.
Ethan and I worked with a story about money.
With that, you have the price level determined.
You have money with a positive value that would exist.
Then you get hyper inflation and liquidity.
Anyway, the point is you have a model in money and general
You can do it all in one or two periods, that's the building
block in which these models rest.
Then the three of us went back to their default models and we
said to ourself, you know, the lender isn't just the bank,
central bank that doesn't care whether it gets paid back or
The lender is -- people are lending.
>> We decided -- we imagined the mortgage market where all of
these guys pulled together.
The lender effect is someone who buys a share in the pool.
In equilibrium, you have to calculate not just the price of
the commodity, but the mortgage pool's pie but the default wait
Every mortgage investor is doing that.
It might be the payment rate or default rate.
K. Is the factor based on force of nature.
Everybody foresees a default rate.
Because we aggregated these into a pool, if you bought more
shares, that wouldn't change the default because your loan is
spread over so many people.
That's the model.
So, now, in this model.
What happens?
Given the default rates, everyone is rationally
That affects the price you're willing to pay for the pool and
the interest rate which affects how many people want to borrow
their money and sell their mortgage into the pool.
All of that is being taken care of.
That's the model Dmitry builds on.
There aren't agents defaulting differently.
Everyone is looking at the average default rate of the
whole pool.
What's macing from that.
Is that there is no constraint on the amount of loans.
There's no lending standards.
Things that adjust when people default is the price that
There's lending standards that ought to adjust.
Three of us couldn't quite figure out what to do.
Then I decided I want to work on collateral instead of on simple
>> So, I -- instead of having a fixed penalty for defaulting on
every dollar, for every dollar you don't deliver, you pay a
I imagine you put up collateral.
So, the difference now is that a loan, is defined as an ordered
pair, not just a promise, "A," but the collateral you have to
put up for it as well.
>> You think of a loan together with DLAT RAL.
They pay attention to the collateral.
Get a price of the loan.
Then the question, how can you decide two things again?
What is the price of the loan and what is the collateral that
you have to put up for the loan.
The answer turned out to be, fix the problems, "A" and allow for
every possible collateral level.
Think of those as separate loans and see which is traded.
Turns out only one traded.
Often few are traded.
I can't explain by.
That's how you figure out what the interest rate and collateral
had to be.
So the three of us went back the same way, and we indogeniZed the
There are other people who worked on it.
Two of those, it's based on the corporate finance view.
I need collateral.
So we don't have skin in the game so we might do something
In the collateral work that we did, that I did.
There's applaud symmetric information playing a role.
Wasn't to make sure it's not the key that when one guy is selling
to another the seller knew more than the buyer.
Typically the buyer was hedge fund and buyer was what screwed
it up.
I don't think it's average selection that changed leverage
all of a sudden.
Anyway, those other papers have leverage.
What do I mean by leverage?
I'm on to definition of assets.
Thank you.
If you put up an asset as collateral for your loan.
A house that's worth $100 then you borrow 80, loan devalues to
80%, down payment is 20%.
Leverage is 5.
$20 of cash got you $100.
That's what I mean by leverage.
That's determined.
Every loan in the market.
It will be traded.
Take the ratio to the collateral and you get the loan amount.
That's the whole point of that theory.
Now, I want to emphasize because it's come up many times, when
you talk about leverage, talk about -- there's two different
kinds of leverage.
A particular house is collateral.
Can you buy 80% of the house?
If you look at the leverage of an individual, look at investor
All of his loans, divided by value of his assets.
Investor leverage is different than asset number.
>> Look at new versus old.
If prices collapse and he still owes the same amount.
Old leverage is skyrocketing.
New leverage is collapsing.
If you want a new loan you can hardly borrow anything on the
So on this, the items I talked about.
I want to come back to Dmitry.
Leverage is determined by uncertainty.
The lender is worried they won't get paid back.
If the price goes down he won't lend as much.
Also determined by innovation.
The longer people have to innovate -- the collateral is
always scarce.
People want to make promises, collateral has an extra value.
Innovators find a way to stretch collateral.
Most of the innovation on wall street you look at it the way
you stretch collateral.
It's like increasing leverage.
If you have a long period of moderation.
You'll stretch the available collateral and if volatility is
low, you have leverage.
>> Both things make leverage go up.
I consider it a two-period model.
Pay a lot.
A little on the down statement.
>> What will the value of the asset be in that delivery room?
Depends how much the optimist can borrow.
By thinking of dual nature, you think what the equilibrium
leverage is and then figure out what the equilibrium price is.
If it turns out -- okay, fine, go ahead.
Turns out the higher the leverage is the higher the asset
price will be.
To say a word about leverage.
There's two different cases.
I hope you're not serious.
I'm going go two more minutes.
There's two different cases that I'm going to consider, I want to
One is --
>> I would rather go very fast.
All right.
So, the higher the leverage, easier for the optimist to get
their hands on money and therefore the higher the price
will be.
What is the leverage cycle next slide?
It's that if you have a long period of low volatility.
There will be high -- high leverage and high asset prices
because of low volatility and because of the innovation.
The price will be very high and that sets the stage for the
crash on the next slide because if something happens that makes
people more pessimistic, it also weigh in on the volatility, next
slide you see that -- next slide again -- that the people left to
buy aren't going to borrow.
You have a lot more borrows.
So marginal buyer is much lower.
Next slide.
This picture, there's a picture on 19.
I have two more after this.
You can see shoulders, picture of housing prices, green line
going way up, way down.
Purple is way up and going way down.
>> I think we have the right regulator, we would have seen
that something crazy was going on.
And we would have been able to do something about it.
I don't think it's impossible to find.
Even though volatility is very low.
Something is wrong.
Next slide.
Back to the end to conclude with Dmitry.
>> What have they done?
It's something I want to emphasize that I agree with.
>> The policy is, many different players, private banks, central
banks, home owners, lenders, you capture all of the indirect
effects that people talked about before.
Marcus talked about.
Analyze many different tools and see what is going to happen.
Markets read clear what will happen.
You didn't do analysis.
Without the approach they take.
Last slide.
But there's more that they can do.
There's more they are doing and they're doing it.
One thing, they don't actually have indogenous records.
The prices go way up.
The per dollar default becomes less onerous.
If the prices are really low, winning default by a whole
house, the penalty changes with the price level.
That's the only way that it's not lenders getting tighter with
They have two periods so uncertainty never changes.
They don't have a legendary capture effect.
>> Something happened in 2007 and 2008.
There's plenty of signs something was wrong in 2007.
Subprime regulators collapsed.
Somehow they were told this, too will pass.
They don't have an underwater agent effect.
>> If they're under water, they won't behave.
They won't fix their house.
That's why forgiveness is important.
There's no role for financial obligation.
There's no father in the paper.
This raises my final point.
They capture indirect effect.
You have to have a model like you have.
I was taught to play both sides of the street.
There is an alternative to what they do, which is have an
agent-based model.
Write down rules, then let the model run.
You can have -- they were worried about 70 equations, you
can have 100 million agents, if you volume an agent-based model.
>> I believe they can theorums.
Sorry I went over.
[ Applause ]
>> I think if you follow the list of suggestions that John
have given you on the last slide you have enough for the next few
I'm going to be very impolite.
Because I have to leave in five minutes.
I am just going to ask a question, hope for an answer
then leave.
This is not a field in which I've done and looking at the
paper the progress made over the last few years.
I thought there was a clear distinction between what I call
the third generation involved in yours.
Which might be the second or third.
The initial paper that I taught.
It was potential default.
Basically you put in place conditions so there is no
default in equilibrium.
>> Ask people to hold collateral and banks have enough capital.
It was a risk.
>> It seems to me that what we have learned, and what you
capture, the nice thing you know, that's why I call it
second generation.
>> It turns out to be just as important.
I thought this is really where looking at it, there has been
I was trying think of thing us were not capturing.
The lift is not quite as long as John's.
I was thinking of maturity mismatch and not going to Libya.
Which strikes me as something, a few years back, we had to
prevent them.
It can happen with government, it can happen -- seems that the
general feature of the financial system, indicates at this point.
I wasn't clear if you had a setup which you can handle these
things or had to add something to this model to get to the
This is my question that I'm wading for and then I'll leave
>> I think we have to add.
Remember, our models are large enough and complex in a sense
And one of the things that I have always heard is that the
kind of model which Dmitry and I tried to develop is actually one
that can be used impercentically and you can actually take
that -- these models to the data and look at them.
And there everybody so in a number of countries.
I believe the central banks have actually used them.
I think it's very important, the models can be used in practice
to look at events in real countries.
We can't yet -- Israel is a real country at this point.
>> Tobias, I think, put it so well that we're going to steal,
if we can, that he was really clear about what he was we were
trying to do, and what we were actually saying, and John and we
have slight differences in emphasis, slight differences in
What unites us is far greater than what divides us.
What unites us is a feeling that we we have to move to is
equilibrium models which default and practices leverage, play a
major role.
Now, since I've been left in charge, can I take questions,
which now Dimitli will answer?
Any questions?
You're all wanting --
>> Ah, over there.
>> I wanted to make a point about leverage, that I guess
John was focusing on.
And the regulatory implications, so I couldn't agree more,
leverage is a critical aspect that should be incorporated in
macro models, but it's not so much leverage per se, and I
think in John's graph of the real estate market.
It's the case in point.
Not all leverage is created equal.
If you look at the futures market.
There's no bailouts of futures traders or ct.
It has to do with underlying instruments you trade and
ability to unwind up and down.
The thing about real estate market that's important to akeep
in mind.
When the leverage went up, it wasn't possible for home owners
to reduce leverage because of the instability of the house.
You can't sell the kitchen in order to reduce the leverage
I think it's the liquidity, as well.
It's difficult to do, and extremely difficult to do by
coming up with therums.
I think that's a simple one to build in to the agent-based
>> Do you want to comment briefly?
We're going to get another question or two.
>> A couple points, there are too many now.
Let me prioritize them.
Thank you very much, Tobias, I like your transparencies, I
promise they will reappear in the future.
Now, we expect to give answers here for Tobias.
We don't have, in our model.
You use for a stable funding ratio.
Now, we do have a credit crunch, where losses to the bank can
mean that the new --
>>> I was just adding more to consider.
Any other questions?
Well, since I am the chairman, in the final session.
I suppose it falls to me to thank Dick and all of his
colleague, very, very much indeed for setting up this
magnificent -- [ Applause ]
>> Well, thanks not only to this panel and everybody to all of
you for coming.
I said in the outset that we want to improve the data.
We have a lot to chew on and I feel makes' percent versions of
Andrew's analogy between salad and garbage, but right now, I
think it's not just salad but maybe a rich five-course meal.
Rich meals often take a lot of time.
>> Good wine, and a strong digestive system to savor and
I think we'll need that time.
I want to let you know that we aim to make all of the materials
available to you in one form or another.
The exact medium we'll use we have to determine, because to
some extent that's a function of our external rules but we will
make them available.
We want to make sure that the work that has gone on over the
past two days becomes the basis for the work that we're doing,
both in the council and in the OFR.
Hopefully this will be the first of many dialogues.
Thanks very much for coming.
And have a safe trip home.
[ Applause ]