A special report on financial risk l February 13th 2010

The gods strike back

RISk.indd 1 2/2/10 13:08:02

The Economist February 13th 2010 A special report on �nancial risk 1

Financial risk got ahead of the world’s ability to manage it. Matthew Valencia asks if it can be tamed again

ed by larger balance sheets and greater le­verage (borrowing), risk was being cappedby a technological shift.

There was something self­servingabout this. The more that risk could be cali­brated, the greater the opportunity to turndebt into securities that could be sold orheld in trading books, with lower capitalcharges than regular loans. Regulators ac­cepted this, arguing that the �great moder­ation� had subdued macroeconomic dan­gers and that securitisation had choppedup individual �rms’ risks into manageablelumps. This faith in the new, technology­driven order was re�ected in the Basel 2bank­capital rules, which relied heavily onthe banks’ internal models.

There were bumps along the way, suchas the near­collapse of Long­Term CapitalManagement (LTCM), a hedge fund, andthe dotcom bust, but each time markets re­covered relatively quickly. Banks grewcocky. But that sense of security was de­stroyed by the meltdown of 2007­09,which as much as anything was a crisis ofmodern metrics­based risk management.The idea that markets can be left to policethemselves turned out to be the world’smost expensive mistake, requiring $15 tril­lion in capital injections and other formsof support. �It has cost a lot to learn how lit­tle we really knew,� says a senior centralbanker. Another lesson was that managingrisk is as much about judgment as aboutnumbers. Trying ever harder to capture

The gods strike back

�THE revolutionary idea that de�nesthe boundary between modern

times and the past is the mastery of risk:the notion that the future is more than awhim of the gods and that men and wom­en are not passive before nature.� So wrotePeter Bernstein in his seminal history ofrisk, �Against the Gods�, published in 1996.And so it seemed, to all but a few Cassan­dras, for much of the decade that followed.Finance enjoyed a golden period, with lowinterest rates, low volatility and high re­turns. Risk seemed to have been reducedto a permanently lower level.

This purported new paradigm hinged,in large part, on three closely linked devel­opments: the huge growth of derivatives;the decomposition and distribution ofcredit risk through securitisation; and theformidable combination of mathematicsand computing power in risk managementthat had its roots in academic work of themid­20th century. It blossomed in the1990s at �rms such as Bankers Trust andJPMorgan, which developed �value­at­risk� (VAR), a way for banks to calculatehow much they could expect to lose whenthings got really rough.

Suddenly it seemed possible for any �­nancial risk to be measured to �ve decimalplaces, and for expected returns to be ad­justed accordingly. Banks hired hordes ofPhD­wielding �quants� to �ne­tune evermore complex risk models. The belief tookhold that, even as pro�ts were being boost­

An audio interview with the author is at

Economist.com/audiovideo

A list of sources is at

Economist.com/specialreports

Number­crunchers crunchedThe uses and abuses of mathematical models. Page 3

Cinderella’s momentRisk managers to the fore. Page 6

A matter of principleWhy some banks did much better than others. Page 7

When the river runs dryThe perils of a sudden evaporation of liquidity.Page 8

Fingers in the dikeWhat regulators should do now. Page 10

Blocking out the sirens’ songMoneymen need saving from themselves.Page 13

Also in this section

AcknowledgmentsIn addition to those mentioned in the text, the authorwould like to thank the following for their help inpreparing this report: Madelyn Antoncic, Scott Baret,Richard Bookstaber, Kevin Buehler, Jan Brockmeijer,Stephen Cecchetti, Mark Chauvin, John Cochrane, JoséCorral, Wilson Ervin, Dan Fields, Chris Finger, BennettGolub, John Hogan, Henry Hu, Simon Johnson, RobertKaplan, Steven Kaplan, Anil Kashyap, James Lam, BrianLeach, Robert Le Blanc, Mark Levonian, Tim Long, BlytheMasters, Michael Mendelson, Robert Merton, Jorge Mina,Mary Frances Monroe, Lubos Pastor, Henry Ristuccia,Brian Robertson, Daniel Sigrist, Pietro Veronesi, JimWiener, Paul Wright and Luigi Zingales.

1

2 A special report on �nancial risk The Economist February 13th 2010

2

1

risk in mathematical formulae can becounterproductive if such a degree of ac­curacy is intrinsically unattainable.

For now, the hubris of spurious preci­sion has given way to humility. It turns outthat in �nancial markets �black swans�, orextreme events, occur much more oftenthan the usual probability models suggest.Worse, �nance is becoming more fragile:these days blow­ups are twice as frequentas they were before the �rst world war, ac­cording to Barry Eichengreen of the Uni­versity of California at Berkeley and Mi­chael Bordo of Rutgers University. BenoitMandelbrot, the father of fractal theoryand a pioneer in the study of marketswings, argues that �nance is prone to a�wild� randomness not usually seen in na­ture. In markets, �rare big changes can bemore signi�cant than the sum of manysmall changes,� he says. If �nancial mar­kets followed the normal bell­shaped dis­tribution curve, in which meltdowns arevery rare, the stockmarket crash of 1987, theinterest­rate turmoil of 1992 and the 2008crash would each be expected only once inthe lifetime of the universe.

This is changing the way many �nan­cial �rms think about risk, says Greg Case,chief executive of Aon, an insurance bro­ker. Before the crisis they were looking atthings like pandemics, cyber­security andterrorism as possible causes of blackswans. Now they are turning to risks fromwithin the system, and how they can be­come ampli�ed in combination.

Cheap as chips, and just as bad for youIt would, though, be simplistic to blamethe crisis solely, or even mainly, on sloppyrisk managers or wild­eyed quants. Cheapmoney led to the wholesale underpricingof risk; America ran negative real interestrates in 2002­05, even though consumer­price in�ation was quiescent. Plenty ofeconomists disagree with the recent asser­tion by Ben Bernanke, chairman of theFederal Reserve, that the crisis had more todo with lax regulation of mortgage pro­ducts than loose monetary policy.

Equally damaging were policies to pro­mote home ownership in America usingFannie Mae and Freddie Mac, the coun­try’s two mortgage giants. They led the duoto binge on securities backed by shoddilyunderwritten loans.

In the absence of strict limits, higher le­verage followed naturally from low inter­est rates. The debt of America’s �nancial�rms ballooned relative to the overalleconomy (see chart 1). At the peak of themadness, the median large bank had bor­

rowings of 37 times its equity, meaning itcould be wiped out by a loss of just 2­3% ofits assets. Borrowed money allowed inves­tors to fake �alpha�, or above­market re­turns, says Benn Steil of the Council onForeign Relations.

The agony was compounded by theproliferation of short­term debt to supportilliquid long­term assets, much of it issuedbeneath the regulatory radar in highly le­veraged �shadow� banks, such as struc­tured investment vehicles. When marketsfroze, sponsoring entities, usually banks,felt morally obliged to absorb their losses.�Reputation risk was shown to have a veryreal �nancial price,� says Doug Roeder ofthe O�ce of the Comptroller of the Cur­rency, an American regulator.

Everywhere you looked, moreover, in­centives were misaligned. Firms deemed�too big to fail� nestled under implicit guar­antees. Sensitivity to risk was dulled by the�Greenspan put�, a belief that America’sFederal Reserve would ride to the rescuewith lower rates and liquidity support ifneeded. Scrutiny of borrowers was dele­gated to rating agencies, who were paid bythe debt­issuers. Some products were socomplex, and the chains from borrower toend­investor so long, that thorough due di­ligence was impossible. A proper under­standing of a typical collateralised debt ob­ligation (CDO), a structured bundle of debtsecurities, would have required reading30,000 pages of documentation.

Fees for securitisers were paid largelyupfront, increasing the temptation to origi­nate, �og and forget. The problems withbankers’ pay went much wider, meaningthat it was much better to be an employeethan a shareholder (or, eventually, a tax­payer picking up the bail­out tab). The roleof top executives’ pay has been over­blown. Top brass at Lehman Brothers andAmerican International Group (AIG) suf­

fered massive losses when share pricestumbled. A recent study found that bankswhere chief executives had more of theirwealth tied up in the �rm performedworse, not better, than those with appar­ently less strong incentives. One explana­tion is that they took risks they thoughtwere in shareholders’ best interests, butwere proved wrong. Motives lower downthe chain were more suspect. It was tooeasy for traders to cash in on short­termgains and skirt responsibility for any time­bombs they had set ticking.

Asymmetries wreaked havoc in thevast over­the­counter derivatives market,too, where even large dealing �rms lackedthe information to determine the conse­quences of others failing. Losses on con­tracts linked to Lehman turned out to bemodest, but nobody knew that when itcollapsed in September 2008, causing pan­ic. Likewise, it was hard to gauge the expo­sures to �tail� risks built up by sellers ofswaps on CDOs such as AIG and bond in­surers. These were essentially put options,with limited upside and a low but realprobability of catastrophic losses.

Another factor in the build­up of exces­sive risk was what Andy Haldane, head of�nancial stability at the Bank of England,has described as �disaster myopia�. Likedrivers who slow down after seeing acrash but soon speed up again, investorsexercise greater caution after a disaster, butthese days it takes less than a decade tomake them reckless again. Not having seena debt­market crash since 1998, investorspiled into ever riskier securities in 2003­07to maintain yield at a time of low interestrates. Risk­management models rein­forced this myopia by relying too heavilyon recent data samples with a narrow dis­tribution of outcomes, especially in sub­prime mortgages.

A further hazard was summed up bythe assertion in 2007 by Chuck Prince, thenCitigroup’s boss, that �as long as the musicis playing, you’ve got to get up and dance.�Performance is usually judged relative torivals or to an industry benchmark, en­couraging banks to mimic each other’srisk­taking, even if in the long run it bene­�ts no one. In mortgages, bad lendersdrove out good ones, keeping up with ag­gressive competitors for fear of losing mar­ket share. A few held back, but it was noteasy: when JPMorgan sacri�ced �ve per­centage points of return on equity in theshort run, it was lambasted by share­holders who wanted it to �catch up� withzippier­looking rivals.

An overarching worry is that the com­

1Borrowed time

Source: Federal Reserve

US financial-industry debt as % of GDP

0

20

40

60

80

100

120

1978 1988 1998 2008

The Economist February 13th 2010 A special report on �nancial risk 3

2

1

plexity of today’s global �nancial networkmakes occasional catastrophic failure in­evitable. For example, the market for creditderivatives galloped far ahead of its sup­porting infrastructure. Only now are seri­ous moves being made to push these con­tracts through central clearing­houseswhich ensure that trades are properly col­lateralised and guarantee their completionif one party defaults.

Network overloadThe push to allocate capital ever more e�­ciently over the past 20 years created whatTill Guldimann, the father of VAR andvice­chairman of SunGard, a technology�rm, calls �capitalism on steroids�. Banksgot to depend on the modelling of prices inesoteric markets to gauge risks and becameadept at gaming the rules. As a result, capi­tal was not being spread around as e�­ciently as everyone believed.

Big banks had also grown increasinglyinterdependent through the boom in de­rivatives, computer­driven equities trad­ing and so on. Another bond was cross­ownership: at the start of the crisis, �nan­cial �rms held big dollops of each other’scommon and hybrid equity. Such tightcoupling of components increases thedanger of �non­linear� outcomes, where asmall change has a big impact. �Financialmarkets are not only vulnerable to blackswans but have become the perfect breed­ing ground for them,� says Mr Guldimann.In such a network a �rm’s troubles canhave an exaggerated e�ect on the per­ceived riskiness of its trading partners.When Lehman’s credit­default spreads

rose to distressed levels, AIG’s jumped bytwice what would have been expected onits own, according to the InternationalMonetary Fund.

Mr Haldane has suggested that theseknife­edge dynamics were caused not onlyby complexity but also�paradoxically�byhomogeneity. Banks, insurers, hedge fundsand others bought smorgasbords of debtsecurities to try to reduce risk through di­versi�cation, but the ingredients were sim­ilar: leveraged loans, American mortgagesand the like. From the individual �rm’sperspective this looked sensible. But forthe system as a whole it put everyone’seggs in the same few baskets, as re�ected intheir returns (see chart 2).

E�orts are now under way to deal withthese risks. The Financial Stability Board,an international group of regulators, is try­ing to co­ordinate global reforms in areas

such as capital, liquidity and mechanismsfor rescuing or dismantling troubledbanks. Its biggest challenge will be to makethe system more resilient to the failure ofgiants. There are deep divisions over howto set about this, with some favouringtougher capital requirements, othersbreak­ups, still others�including Ameri­ca�a combination of remedies.

In January President Barack Obamashocked big banks by proposing a tax ontheir liabilities and a plan to cap their size,ban �proprietary� trading and limit theirinvolvement in hedge funds and privateequity. The proposals still need congressio­nal approval. They were seen as energisingthe debate about how to tackle dangerous­ly large �rms, though the reaction in Eu­rope was mixed.

Regulators are also inching towards amore �systemic� approach to risk. The oldsupervisory framework assumed that ifthe 100 largest banks were individuallysafe, then the system was too. But the crisisshowed that even well­managed �rms,acting prudently in a downturn, can un­dermine the strength of all.

The banks themselves will have to �nda middle ground in risk management,somewhere between gut feeling and num­ber fetishism. Much of the progress madein quantitative �nance was real enough,but a �rm that does not understand the�aws in its models is destined for trouble.This special report will argue that ruleswill have to be both tightened and betterenforced to avoid future crises�but that allthe reforms in the world will never guaran­tee total safety. 7

2In lockstep

Source: “Banking on the State” by Andrew Haldane andPiergiorgio Alessandri; Bank for International Settlements

Weighted average cumulative total returns, %

2000 01 02 03 04 05 06 07 08 0950

0

50

100

150

200

+

–

Large complexfinancialinstitutions

Banks

Insurers

Hedge funds

IT PUT noses out of joint, but it changedmarkets for good. In the mid­1970s a few

progressive occupants of Chicago’s op­tions pits started trading with the aid ofsheets of theoretical prices derived from amodel and sold by an economist calledFisher Black. Rivals, used to relying on theirwits, were unimpressed. One model­based trader complained of having his pa­pers snatched away and being told to�trade like a man�. But the strings of num­bers caught on, and soon derivatives ex­changes hailed the Black­Scholes model,which used share and bond prices to calcu­

late the value of derivatives, for helping tolegitimise a market that had been deridedas a gambling den.

Thanks to Black­Scholes, options pric­ing no longer had to rely on educatedguesses. Derivatives trading got a hugeboost and quants poured into the industry.By 2005 they accounted for 5% of all �­nance jobs, against 1.2% in 1980, says Thom­as Philippon of New York University�andprobably a much higher proportion of pay.By 2007 �nance was attracting a quarter ofall graduates from the California Instituteof Technology.

These eggheads are now in the dock,along with their probabilistic models. InAmerica a congressional panel is investi­gating the models’ role in the crash. Wired,a publication that can hardly be accused oftechnophobia, has described default­prob­ability models as �the formula that killedWall Street�. Long­standing critics of risk­modelling, such as Nassim Nicholas Taleb,author of �The Black Swan�, and Paul Wil­mott, a mathematician turned �nancialeducator, are now hailed as seers. Models�increased risk exposure instead of limit­ing it�, says Mr Taleb. �They can be worse

Number­crunchers crunched

The uses and abuses of mathematical models

4 A special report on �nancial risk The Economist February 13th 2010

2

1

than nothing, the equivalent of a danger­ous operation on a patient who wouldstand a better chance if left untreated.�

Not all models were useless. Those forinterest rates and foreign exchange per­formed roughly as they were meant to.However, in debt markets they failed ab­jectly to take account of low­probabilitybut high­impact events such as the gut­wrenching fall in house prices.

The models went particularly awrywhen clusters of mortgage­backed securi­ties were further packaged into collateral­ised debt obligations (CDOs). In traditionalproducts such as corporate debt, ratingagencies employ basic credit analysis andjudgment. CDOs were so complex thatthey had to be assessed using specially de­signed models, which had various faults.Each CDO is a unique mix of assets, but theassumptions about future defaults andmortgage rates were not closely tailored tothat mix, nor did they factor in the tenden­cy of assets to move together in a crisis.

The problem was exacerbated by thecredit raters’ incentive to accommodatethe issuers who paid them. Most �nancial�rms happily relied on the models, eventhough the expected return on AAA­ratedtranches was suspiciously high for suchapparently safe securities. At some banks,risk managers who questioned the ratingagencies’ models were given short shrift.Moody’s and Standard & Poor’s were as­sumed to know best. For people paid ac­cording to that year’s revenue, this was un­derstandable. �A lifetime of wealth wasonly one model away,� sneers an Ameri­can regulator.

Moreover, heavy use of models mayhave changed the markets they were sup­posed to map, thus undermining the valid­ity of their own predictions, says DonaldMacKenzie, an economic sociologist at theUniversity of Edinburgh. This feedbackprocess is known as counter­performativ­ity and had been noted before, for instancewith Black­Scholes. With CDOs the mod­els’ popularity boosted demand, whichlowered the quality of the asset­backed se­curities that formed the pools’ raw materi­al and widened the gap between expectedand actual defaults (see chart 3).

A related problem was the similarity ofrisk models. Banks thought they were div­ersi�ed, only to �nd that many others heldcomparable positions, based on similarmodels that had been built to comply withthe Basel 2 standards, and everyone wastrying to unwind the same positions at thesame time. The breakdown of the models,which had been the only basis for pricing

the more exotic types of security, turnedrisk into full­blown uncertainty (and thusextreme volatility).

For some, the crisis has shattered faithin the precision of models and their inputs.They failed Keynes’s test that it is better tobe roughly right than exactly wrong. Onenumber coming under renewed scrutiny is�value­at­risk� (VAR), used by banks tomeasure the risk of loss in a portfolio of �­nancial assets, and by regulators to calcu­late banks’ capital bu�ers. Invented by egg­heads at JPMorgan in the late 1980s, VAR

has grown steadily in popularity. It is thesubject of more than 200 books. Whatmakes it so appealing is that its complexformulae distil the range of potential dailypro�ts or losses into a single dollar �gure.

Only so far with VARFrustratingly, banks introduce their ownquirks into VAR calculations, making com­parison di�cult. For example, MorganStanley’s VAR for the �rst quarter of 2009by its own reckoning was $115m, but usingGoldman Sachs’s method it would havebeen $158m. The bigger problem, though, isthat VAR works only for liquid securitiesover short periods in �normal� markets,and it does not cover catastrophic out­comes. If you have $30m of two­week 1%VAR, for instance, that means there is a 99%chance that you will not lose more thanthat amount over the next fortnight. Butthere may be a huge and unacknowledgedthreat lurking in that 1% tail.

So chief executives would be foolish torely solely, or even primarily, on VAR tomanage risk. Yet many managers andboards continue to pay close attention to itwithout fully understanding the caveats�

the equivalent of someone who cannotswim feeling con�dent of crossing a riverhaving been told that it is, on average, fourfeet deep, says Jaidev Iyer of the Global As­sociation of Risk Professionals.

Regulators are encouraging banks tolook beyond VAR. One way is to use Co­VAR (Conditional VAR), a measure thataims to capture spillover e�ects in trou­bled markets, such as losses due to the dis­tress of others. This greatly increases somebanks’ value at risk. Banks are developingtheir own enhancements. Morgan Stanley,for instance, uses �stress� VAR, which fac­tors in very tight liquidity constraints.

Like its peers, Morgan Stanley is also re­viewing its stress testing, which is used toconsider extreme situations. The worst sce­nario envisaged by the �rm turned out tobe less than half as bad as what actuallyhappened in the markets. JPMorganChase’s debt­market stress tests foresaw a40% increase in corporate spreads, buthigh­yield spreads in 2007­09 increasedmany times over. Others fell similarlyshort. Most banks’ tests were based on his­torical crises, but this assumes that the fu­ture will be similar to the past. �A repeat ofany speci�c market event, such as 1987 or1998, is unlikely to be the way that a futurecrisis will unfold,� says Ken deRegt, Mor­gan Stanley’s chief risk o�cer.

Faced with either random (and there­fore not very believable) scenarios or sim­plistic models that neglect fat­tail risks,many �nd themselves in a �no­man’s­land� between the two, says Andrew Free­man of Deloitte (and formerly a journalistat The Economist). Nevertheless, he viewsscenario planning as a useful tool. A �rmthat had thought about, say, the mutationof default risk into liquidity risk wouldhave had a head start over its competitorsin 2008, even if it had not predicted pre­cisely how this would happen.

To some, stress testing will always seemmaddeningly fuzzy. �It has so far been seenas the acupuncture­and­herbal­remediescorner of risk management, though per­ceptions are changing,� says Riccardo Reb­onato of Royal Bank of Scotland, who iswriting a book on the subject. It is notmeant to be a predictive tool but a meansof considering possible outcomes to allow�rms to react more nimbly to unexpecteddevelopments, he argues. Hedge funds arebetter at this than banks. Some hadthought about the possibility of a largebroker­dealer going bust. At least one,AQR, had asked its lawyers to grill thefund’s prime brokers about the fate of itsassets in the event of their demise.

3Never mind the quality

Source: Donald MacKenzie, University of Edinburgh

CDOs of subprime-mortgage-backed securitiesIssued in 2005-07, %

Estimated Actual 3-year default default rate rate

AAA 0.001 0.10

AA+ 0.01 1.68

AA 0.04 8.16

AA- 0.05 12.03

A+ 0.06 20.96

A 0.09 29.21

A- 0.12 36.65

BBB+ 0.34 48.73

BBB 0.49 56.10

BBB- 0.88 66.67

The Economist February 13th 2010 A special report on �nancial risk 5

2 Some of the blame lies with bank regu­lators, who were just as blind to the dan­gers ahead as the �rms they oversaw.Sometimes even more so: after the rescueof Bear Stearns in March 2008 but beforeLehman’s collapse, Morgan Stanley was re­portedly told by supervisors at the FederalReserve that its doomsday scenario wastoo bearish.

The regulators have since becometougher. In America, for instance, bankshave been told to run stress tests with sce­narios that include a huge leap in interestrates. A supervisors’ report last October�ngered some banks for �window­dress­ing� their tests. O�cials are now asking for�reverse� stress testing, in which a �rmimagines it has failed and works back­wards to determine which vulnerabilitiescaused the hypothetical collapse. Britainhas made this mandatory. Bankers are di­vided over its usefulness.

Slicing the EmmentalThese changes point towards greater use ofjudgment and less reliance on numbers infuture. But it would be unfair to tar all mod­els with the same brush. The CDO �ascowas an egregious and relatively rare caseof an instrument getting way ahead of theability to map it mathematically. Modelswere �an accessory to the crime, not theperpetrator�, says Michael Mauboussin ofLegg Mason, a money manager.

As for VAR, it may be hopeless at signal­ling rare severe losses, but the process bywhich it is produced adds enormously tothe understanding of everyday risk, whichcan be just as deadly as tail risk, says AaronBrown, a risk manager at AQR. Craig Bro­derick, chief risk o�cer at Goldman Sachs,sees it as one of several measures which,although of limited use individually, to­gether can provide a helpful picture. Like aslice of Swiss cheese, each number hasholes, but put several of them together andyou get something solid.

Modelling is not going away; indeed,number­crunchers who are devising newways to protect investors from outlying fat­tail risks are gaining in�uence. Pimco, forinstance, o�ers fat­tail hedging pro­grammes for mutual­fund clients, usingcocktails of options and other instru­ments. These are built on speci�c risk fac­tors rather than on the broader and in­creasingly �uid division of assets betweenequities, currencies, commodities and soon. The relationships between asset class­es �have become less stable�, says Mo­hamed El­Erian, Pimco’s chief executive.�Asset­class diversi�cation remains desir­

able but is not su�cient.�Not surprisingly, more investors are

now willing to give up some upside for thepromise of protection against catastrophiclosses. Pimco’s clients are paying up to 1%of the value of managed assets for thehedging�even though, as the recent crisisshowed, there is a risk that insurers willnot be able to pay out. Lisa Goldberg ofMSCI Barra reports keen interest in the an­alytics �rm’s extreme­risk model fromhedge funds, investment banks and pen­sion plans.

In some areas the need may be for morecomputing power, not less. Financial �rmsalready spend more than any other indus­try on information technology (IT): some$500 billion in 2009, according to Gartner,a consultancy. Yet the quality of informa­tion �ltering through to senior managers isoften inadequate.

A report by bank supervisors last Octo­ber pointed to poor risk �aggregation�:many large banks simply do not have thesystems to present an up­to­date picture oftheir �rm­wide links to borrowers andtrading partners. Two­thirds of the bankssurveyed said they were only �partially�able (in other words, unable) to aggregatetheir credit risks. The Federal Reserve, lead­ing stress tests on American banks lastspring, was shocked to �nd that some ofthem needed days to calculate their expo­

sure to derivatives counterparties.To be fair, totting up counterparty risk is

not easy. For each trading partner the cal­culations can involve many di�erent typesof contract and hundreds of legal entities.But banks will have to learn fast: undernew international proposals, they will forthe �rst time face capital charges on thecreditworthiness of swap counterparties.

The banks with the most dysfunctionalsystems are generally those, such as Citi­group, that have been through multiplemarriages and ended up with dozens of�legacy� systems that cannot easily com­municate with each other. That may ex­plain why some Citi units continued topile into subprime mortgages even as oth­ers pulled back.

In the depths of the crisis some bankswere unaware that di�erent business unitswere marking the same assets at di�erentprices. The industry is working to sort thisout. Banks are coming under pressure toappoint chief data o�cers who can policethe integrity of the numbers, separate fromchief information o�cers who concen­trate on system design and output.

Some worry that the good work will becast aside. As markets recover, the biggesttemptation will be to abandon or scaleback IT projects, allowing product devel­opment to get ahead of the supportingtechnology infrastructure, just as it did inthe last boom.

The way forward is not to reject high­tech �nance but to be honest about its limi­tations, says Emanuel Derman, a professorat New York’s Columbia University and aformer quant at Goldman Sachs. Modelsshould be seen as metaphors that can en­lighten but do not describe the world per­fectly. Messrs Derman and Wilmott havedrawn up a modeller’s Hippocratic oathwhich pledges, among other things: �I willremember that I didn’t make the world,and it doesn’t satisfy my equations,� and �Iwill never sacri�ce reality for elegancewithout explaining why I have done so.�Often the problem is not complex �nancebut the people who practise it, says Mr Wil­mott. Because of their love of puzzles,quants lean towards technically brilliantrather than sensible solutions and tend toover­engineer: �You may need a plumberbut you get a professor of �uid dynamics.�

One way to deal with that problem is toself­insure. JPMorgan Chase holds $3 bil­lion of �model­uncertainty reserves� tocover mishaps caused by quants who havebeen too clever by half. If you can makeprovisions for bad loans, why not badmaths too? 7

6 A special report on �nancial risk The Economist February 13th 2010

1

IN A speech delivered to a banking­indus­try conference in Geneva in December

2006, Madelyn Antoncic issued a warningand then o�ered some reassurance. Withvolatility low, corporate credit spreadsgrowing ever tighter and markets all but ig­noring bad news, there was, she said, �aseemingly overwhelming sense of com­placency�. Nevertheless, she insisted thatthe �rm she served as chief risk o�cer,Lehman Brothers, was well placed to rideout any turbulence, thanks to a keenawareness of emerging threats and a rock­solid analytical framework.

Behind the scenes, all was not well. MsAntoncic, a respected risk manager withan economics PhD, had expressed uneaseat the �rm’s heavy exposure to commer­cial property and was being sidelined, bitby bit, by the �rm’s autocratic boss, DickFuld. Less than two months after herspeech she was pushed aside.

Lehman’s story ended particularly bad­ly, but this sort of lapse in risk governancewas alarmingly common during theboom. So much for the notion, generallyaccepted back then, that the quality ofbanks’ risk regimes had, like car compo­nents, converged around a high standard.�The variance turned out to be shocking,�says Jamie Dimon, chief executive ofJPMorgan Chase.

The banks that fared better, includinghis own, relied largely on giving their risk­managing roundheads equal status withthe risk­taking cavaliers. That was not easy.In happy times, when risk seems low, pow­er shifts from risk managers to traders.Sales­driven cultures are the natural orderof things on Wall Street and in the City. Dis­couraging transactions was frowned upon,especially at �rms trying to push their wayup capital­markets league tables. Riskmanagers who said no put themselves ona collision course with the business headand often the chief executive too.

At some large banks that subsequentlysu�ered big losses, such as HBOS and Roy­al Bank of Scotland (RBS), credit commit­tees, which vetted requests for big loans,could be formed on an ad hoc basis from apool of eligible members. If the commit­tee’s chairman, typically a business­linehead, encountered resistance from a risk

manager or other sceptic, he could adjournthe meeting, then reconstitute the commit­tee a week or two later with a more pliablemembership that would approve the loan.

Another common trick was for a busi­ness line to keep quiet about a proposal onwhich it had been working for weeks untila couple of hours before the meeting to ap­prove it, so the risk team had no time tolodge convincing objections. Exasperatedroundheads would occasionally resort topleading with regulators for help. In theyears before the crash the Basel Commit­tee of bank supervisors reportedly re­ceived several requests from risk managersto scrutinise excessive risk­taking at theirinstitutions that they felt powerless to stop.

Many banks’ failings exposed the tri­umph of form over substance. In recentyears it had become popular to appoint achief risk o�cer to signal that the issue wasreceiving attention. But according to LeoGrepin of McKinsey, �it was sometimes acase of management telling him, ‘you tickthe boxes on risk, and we’ll worry aboutgenerating revenue’.�

Since 2007 banks have been scram­

bling to convince markets and regulatorsthat they will continue to take risk serious­ly once memories of the crisis fade. Someare involving risk o�cers in talks aboutnew products and strategic moves. AtHSBC, for instance, they have had a biggerrole in vetting acquisitions since the bank’sAmerican retail­banking subsidiary,bought in 2003, su�ered heavy subprime­mortgage losses. �Everyone should nowsee that the risk team needs to be just as in­volved on the returns side as on the riskside,� says Maureen Miskovic, chief risk of­�cer at State Street, an American bank.

Glamming upMs Miskovic is one of an emerging breedof more powerful risk o�cers. They areseen as being on a par with the chief �nan­cial o�cer, get a say in decisions on payand have the ear of the board, whoseagreement is increasingly needed to re­move them. Some report directly to aboard committee as well as�or occasional­ly instead of�to the chief executive.

For many, the biggest task is to disman­tle cumbersome �silos�, says Ken Chalk of

Cinderella’s moment

Risk managers to the fore

The Economist February 13th 2010 A special report on �nancial risk 7

2

1

America’s Risk Management Association.Risks were often stu�ed into convenientbut misleading pigeonholes. Banks wereslow to re�ne their approach, even asgrowing market complexity led some ofthe risks to become interchangeable.

Take the growth of traded credit pro­ducts, such as asset­backed securities andCDOs made up of them. Credit­risk de­partments thought of them as market risk,because they sat in the trading book. Mar­ket­risk teams saw them as credit instru­ments, since the underlying assets wereloans. This buck­passing proved particu­larly costly at UBS, which lost SFr36 billion($34 billion) on CDOs. Many banks arenow combining their market­ and credit­risk groups, as HSBC did last year.

For all the new­found authority of riskmanagers, it can still be hard to attract tal­ent to their ranks. The job is said to havethe risk pro�le of a short option positionwith unlimited downside and limited up­side�something every good risk managershould avoid. Moreover, it lacks glamour.Persuading a trader to move to risk can be�like asking a trapeze artist to retrain as anaccountant�, says Barrie Wilkinson of Oli­ver Wyman, a consultancy.

A question of cultureBesides, there is more to establishing a sol­id risk culture than empowering risk o�­cers. Culture is a slippery concept, but itmatters. �Whatever causes the next crisis,it will be di�erent, so you need somethingthat can deal with the unexpected. That’sculture,� says Colm Kelleher of MorganStanley. One necessary ingredient is a tra­dition of asking and repeating questionsuntil a clear answer emerges, suggestsClayton Rose, a banker who now teachesat Harvard Business School.

The tone is set at the top, for better orworse. At the best­run banks senior �guresspend as much time fretting over risks asthey do salivating at opportunities (seebox). By contrast, Lehman’s Mr Fuld talkedof �protecting mother� but was drawn tothe glister of leveraged deals. Stan O’Neal,who presided over giant losses at MerrillLynch, was more empire­builder than riskmanager. But imperial bosses and soundrisk cultures sometimes go together, as atJPMorgan and Banco Santander.

A soft­touch boss can be more danger­ous than a domineering one. Under ChuckPrince, who famously learned only in Sep­tember 2007 that Citigroup was sitting on$43 billion of toxic assets, the lunatics wereable to take over the asylum. Astonishing­ly, the head of risk reported not to Mr

Prince or the board, but to a newly hired ex­ecutive with a background in corporate­go­vernance law, not cutting­edge �nance.

Another lesson is that boards mattertoo. Directors’ lack of engagement or ex­pertise played a big part in some of theworst slip­ups, including Citi’s. The �so­ciology� of big banks’ boards also hadsomething to do with it, says Ingo Walterof New York’s Stern School of Business: asthe members bonded, dissidents felt pres­sure to toe the line.

Too few boards de�ned the parametersof risk oversight. In a survey last year De­

loitte found that only seven of 30 largebanks had done so in any detail. Everyoneagrees that boards have a critical role toplay in determining risk appetite, but a re­cent report by a group of global regulatorsfound that many were reluctant to do this.

Boards could also make a better job ofpolicing how (or even whether) banks ad­just for risk in allocating capital internally.Before the crisis some boards barelythought about this, naively assuming thatprocedures for it were well honed. A for­mer Lehman board member professeshimself �astonished�, in retrospect, at how

JPMORGAN CHASE managed to avoidbig losses largely thanks to the tone setby its boss, Jamie Dimon. A voracious

reader of internal reports, he understands�nancial arcana and subjects sta� to de­tailed questioning. PowerPoint presenta­tions are discouraged, informal discus­sions of what is wrong, or could gowrong, encouraged. These �soft� princi­ples are supplemented by a hard­headedapproach to the allocation of capital.Though the bank su�ered painful lossesin leveraged loans, it was not tripped upby CDOs or structured investment vehi­cles (SIVs), even though it had been in­strumental in developing both products.Nor was it heavily exposed to AIG, an in­surance giant that got into trouble.

This was not because it saw disastercoming, says Bill Winters, former co­headof the �rm’s investment bank, but be­cause it stuck by two basic principles:don’t hold too much of anything, andonly keep what you are sure will generatea decent risk­adjusted return. The bankjettisoned an SIV and $60 billion of CDO­related risks because it saw them as toodicey, at a time when others were stillkeen to snap them up. It also closed 60credit lines for other SIVs and corporateclients when it realised that these couldbe simultaneously drawn down if thebank’s credit rating were cut. And it took aconservative view of risk­mitigation.Hedging through bond insurers, whose �­nances grew shaky as the crisis spread,was calculated twice: once assuming thehedge would hold, and again assuming itwas worthless.

Goldman Sachs’s risk managementstood out too�unlike the public­relationsskills it subsequently displayed. Steeredby its chief �nancial o�cer, David Viniar,the �rm’s traders began reducing their ex­posure to mortgage securities months be­

fore subprime defaults began to explode.More willing than rivals to take risks,Goldman is also quicker to hedge them.In late 2006 it spent up to $150m�one­eighth of that quarter’s operating pro�t�hedging exposure to AIG.

The �rm promotes senior traders torisk positions, making clear that suchmoves are a potential stepping stone tothe top. Traders are encouraged to nurturethe risk manager in them: Gary Cohn, the�rm’s president, rose to the top largely be­cause of his skill at hedging �tail� risks.Crucially, Goldman generally does not�re its risk managers after a crisis, allow­ing them to learn from the experience. Yetdespite everything, it still needed govern­ment help to survive.

By contrast, UBS’s risk culture was aw­ful. Its investment bank was free to betwith subsidised funds, since transfersfrom the private bank were deeply under­priced. It confused itself by presentingrisk in a �net and forget� format. Tradingdesks would estimate the maximum pos­sible loss on risky assets, hedge it andthen record the net risk as minimal, inad­vertently concealing huge tail risks in thegross exposure. And it moved its best trad­ers to a hedge fund, leaving the B­team tomanage the bank’s positions.

Publicly humbled by a frank report onits failings, the bank has made a raft ofchanges. Risk controllers have been hand­ed more power. Oswald Grübel, the chiefexecutive, has said that if his newish riskchief, Philip Lofts, rejects a transaction hewill never overrule him. If the two dis­agree, Mr Lofts must inform the board,which no longer delegates risk issues to atrio of long­time UBS employees. A new,independent risk committee is bristlingwith risk experts. Whether all thisamounts to a �new paradigm�, as MrLofts claims, remains to be seen.

Why some banks did much better than others

A matter of principle

8 A special report on �nancial risk The Economist February 13th 2010

2

1

some of the risks in the company’s proper­ty investments were brushed aside whenassessing expected returns. The survivorsare still struggling to create the sort ofjoined­up approach to risk adjustmentthat is common at large hedge funds, ad­mits one Wall Street executive.

Board gamesRobert Pozen, head of MFS InvestmentManagement, an American asset manager,thinks bank boards would be more e�ec­tive with fewer but more committed mem­bers. Cutting their size to 4­8, rather thanthe 10­18 typical now, would foster morepersonal responsibility. More �nancial­services expertise would help too. Afterthe passage of the Sarbanes­Oxley act in2002 banks hired more independent direc­tors, many of whom lacked relevant expe­rience. The former spymaster on Citi’sboard and the theatrical impresario onLehman’s may have been happy to askquestions, but were they the right ones?

Under regulatory pressure, banks suchas Citi and Bank of America have hiredmore directors with strong �nancial­ser­

vices backgrounds. Mr Pozen suggests as­sembling a small cadre of �nancially �u­ent �super­directors� who would meetmore often�say, two or three days a monthrather than an average of six days a year, asnow�and may serve on only one otherboard to ensure they take the job seriously.

That sounds sensible, but the case foranother suggested reform�creating inde­pendent risk committees at board level�isless clear. At some banks risk issues arehandled perfectly well by the audit com­mittee or the full board. Nor is there a clearlink between the frequency of risk­relatedmeetings and a bank’s performance. AtSpain’s Santander the relevant committeemet 102 times in 2008. Those of otherbanks that emerged relatively unscathed,such as JPMorgan and Credit Suisse, con­vened much less often.

Moreover, some of the most importantrisk­related decisions of the next few yearswill come from another corner: the com­pensation committee. It is not just invest­ment bankers and top executives whosepay structures need to be rethought. In thepast, risk managers’ pay was commonly

determined or heavily in�uenced by themanagers of the trading desks they over­saw, or their bonus linked to the desks’ per­formance, says Richard Apostolik, whoheads the Global Association of Risk Pro­fessionals (GARP). Boards need to elimi­nate such con�icts of interest.

Meanwhile risk teams are being beefedup. Morgan Stanley, for instance, is increas­ing its complement to 450, nearly doublethe number it had in 2008. The GARP sawa 70% increase in risk­manager certi�ca­tions last year. Risk is the busiest area for �­nancial recruiters, says Tim Holt of Hei­drick & Struggles, a �rm of headhunters.When boards are looking for a new chiefexecutive, they increasingly want some­one who has been head of risk as well aschief �nancial o�cer, which used to be thestandard requirement, reckons MikeWoodrow of Risk Talent Associates, an­other headhunting �rm.

The big question is whether this inter­est in controlling risk will �zzle out as econ­omies recover. Experience suggests that itwill. Bankers say this time is di�erent�butthey always do. 7

STAMPEDING crowds can generate pres­sures of up to 4,500 Newtons per square

metre, enough to bend steel barriers. Rush­es for the exit in �nancial markets can bejust as damaging. Investors crowd intotrades to get the highest risk­adjusted re­turn in the same way that everyone wantstickets for the best concert. When someoneshouts ��re�, their �ight creates an �endog­enous� risk of being trampled by fallingprices, margin calls and vanishing capi­tal�a �negative externality� that adds tooverall risk, says Lasse Heje Pedersen ofNew York University.

This played out dramatically in 2008.Liquidity instantly drained from securities�rms as clients abandoned anything witha whi� of risk. In three days in March BearStearns saw its pool of cash and liquid as­sets shrink by nearly 90%. After the col­lapse of Lehman Brothers, Morgan Stanleyhad $43 billion of withdrawals in a singleday, mostly from hedge funds.

Bob McDowall of Tower Group, a con­sultancy, explains that liquidity poses �themost emotional of risks�. Its loss can prove

just as fatal as insolvency. Many of thoseclobbered in the crisis�including BearStearns, Northern Rock and AIG�werestruck down by a sudden lack of cash orfunding sources, not because they ran outof capital.

Yet liquidity risk has been neglected.Over the past decade international regula­tors have paid more attention to capital.Banks ran liquidity stress tests and drewup contingency funding plans, but oftenhalf­heartedly. With markets awash withcash and hedge funds, private­equity �rmsand sovereign­wealth funds all keen to in­vest in assets, there seemed little prospectof a liquidity crisis. Academics such as MrPedersen, Lubos Pastor at Chicago’s BoothSchool of Business and others were doingsolid work on liquidity shocks, but practi­tioners barely noticed.

What makes liquidity so important isits binary quality: one moment it is there inabundance, the next it is gone. This time itsevaporation was particularly abrupt be­cause markets had become so joined up.The panic to get out of levered mortgage in­

vestments spilled quickly into interbankloan markets, commercial paper, primebrokerage, securities lending (lendingshares to short­sellers) and so on.

As con�dence ebbed, mortgage­backedsecurities could no longer be used so easilyas collateral in repurchase or �repo� agree­ments, in which �nancial �rms borrowshort­term from investors with excesscash, such as money­market funds. Thiswas a big problem because securities �rmshad become heavily reliant on this market,tripling their repo borrowing in the �veyears to 2008. Bear Stearns had $98 billionon its books, compared with $72 billion oflong­term debt.

Even the most liquid markets were af­fected. In August 2007 a wave of selling ofblue­chip shares, forced by the need to cov­er losses on debt securities elsewhere,caused sudden drops of up to 30% forsome computer­driven strategies popularwith hedge funds.

Liquidity comes in two closely connect­ed forms: asset liquidity, or the ability tosell holdings easily at a decent price; and

When the river runs dry

The perils of a sudden evaporation of liquidity

The Economist February 13th 2010 A special report on �nancial risk 9

2

1

funding liquidity, or the capacity to raise �­nance and roll over old debts when need­ed, without facing punitive �haircuts� oncollateral posted to back this borrowing.

The years of excess saw a vast increasein the funding of long­term assets withshort­term (and thus cheaper) debt. Short­term borrowing has a good side: the threatof lenders refusing to roll over can be asource of discipline. Once they expectlosses, though, a run becomes inevitable:they rush for repayment to beat the crowd,setting o� a panic that might hurt themeven more. �Financial crises are almost al­ways and everywhere about short­termdebt,� says Douglas Diamond of the BoothSchool of Business.

Banks are founded on this �maturitymismatch� of long­ and short­term debt,but they have deposit insurance which re­duces the likelihood of runs. However, thistime much of the mismatched borrowingtook place in the uninsured �shadow�banking network of investment banks,structured o�­balance­sheet vehicles andthe like. It was supported by seemingly in­genious structures. Auction­rate securities,for instance, allowed the funding of stodgymunicipal bonds to be rolled over month­ly, with the interest rate reset each time.

The past two years are littered with sto­ries of schools and hospitals that came acropper after dramatically shortening thetenure of their funding, assuming that thesavings in interest costs, small as theywere, far outweighed the risk of market sei­zure. Securities �rms became equally com­placent as they watched asset values rise,boosting the value of their holdings as col­lateral for repos. Commercial banks in­creased their reliance on wholesale fund­ing and on �ckle �non­core� deposits, suchas those bought from brokers.

Regulation did nothing to discouragethis, treating banks that funded them­selves with deposits and those borrowingovernight in wholesale markets exactlythe same. Markets viewed the second cate­gory as more e�cient. Northern Rock,which funded its mortgages largely in capi­tal markets, had a higher stockmarket rat­ing than HSBC, which relied more on con­ventional deposits. The prevailing viewwas that risk was inherent in the asset, notthe manner in which it was �nanced.

At the same time �nancial �rms builtup a host of liquidity obligations, not all ofwhich they fully understood. Banks wereexpected to support o�­balance­sheet enti­ties if clients wanted out; Citigroup had totake back $58 billion of short­term securi­ties from structured vehicles it sponsored.

AIG did not allow for the risk that the in­surer would have to post more collateralagainst credit­default swaps if these fell invalue or its rating was cut.

Now that the horse has bolted, �nan­cial �rms are rushing to close the door, forinstance by adding to liquidity bu�ers (seechart 4). British banks’ holdings of sterlingliquid assets are at their highest for a de­cade. Capital­markets �rms are courtingdeposits and shunning �ighty wholesalefunding. Deposits, equity and long­termdebt now make up almost two­thirds ofMorgan Stanley’s balance­sheet liabilities,compared with around 40% at the end of2007. Spending on liquidity­managementsystems is rising sharply, with specialists�almost able to name their price�, says onebanker. �Collateral management� has be­come a buzzword.

Message from BaselRegulators, too, are trying to make up forlost time. In a �rst attempt to put numberson a nebulous concept, in December theBasel Committee of central banks and su­pervisors from 27 countries proposed a

global liquidity standard for international­ly active banks. Tougher requirementswould reverse a decades­long decline inbanks’ liquidity cushions.

The new regime, which could be adopt­ed as early as 2012, has two components: a�coverage� ratio, designed to ensure thatbanks have a big enough pool of high­quality, liquid assets to weather an �acutestress scenario� lasting for one month (in­cluding such inconveniences as a sharpratings downgrade and a wave of collater­al calls); and a �net stable funding� ratio,aimed at promoting longer­term �nancingof assets and thus limiting maturity mis­matches. This will require a certain level offunding to be for a year or more.

It remains to be seen how closely na­tional authorities follow the script. Someseem intent on going even further. In Swit­zerland, UBS and Credit Suisse face a tri­pling of the amount of cash and equiva­lents they need to hold, to 45% of deposits.Britain will require all domestic entities tohave enough liquidity to stand alone, un­supported by their parent or other parts ofthe group. Also controversial is the compo­sition of the proposed liquidity cushions.Some countries want to restrict these togovernment debt, deposits with centralbanks and the like. The Basel proposals al­low high­grade corporate bonds too.

Banks have counter­attacked, arguingthat �trapping� liquidity in subsidiarieswould reduce their room for manoeuvre ina crisis and that the bu�er rules are too re­strictive; some, unsurprisingly, have calledfor bank debt to be eligible. Under the Brit­ish rules, up to 8% of banks’ assets could betied up in cash and gilts (British govern­ment bonds) that they are forced to hold,reckons Simon Hills of the British BankersAssociation, which could have �a huge im­pact on business models�. That, some ar­

75

4Filling the pool

Sources: Federal Reserve; Goldman Sachs

US banks’ cash assets, $trn

0

0.25

0.50

0.75

1.00

1.25

1973 80 85 90 95 2000 05 09

10 A special report on �nancial risk The Economist February 13th 2010

2

1

gue, is precisely the point of reform.Much can be done to reduce market

stresses without waiting for these reforms.In repo lending�a decades­old practicecritical to the smooth functioning of mar­kets�the Federal Reserve may soontoughen collateral requirements and forceborrowers to draw up contingency plansin case of a sudden freeze. Banks that clearrepos will be expected to monitor the sizeand quality of big borrowers’ positionsmore closely. The banks could live withthat, but they worry about proposals toforce secured short­term creditors to takean automatic loss if a bank fails.

Another concern is prime brokerage,banks’ �nancing of trading by hedgefunds. When the market unravelled, hedgefunds were unable to retrieve collateralthat their brokers had �rehypothecated�,or used to fund transactions of their own;billions of such unsegregated money isstill trapped in Lehman’s estate, reducingdozens of its former clients to the status ofunsecured general creditors. Brokers suf­fered in turn as clients pulled whateverfunds they could from those they viewedas vulnerable. Temporary bans on short­selling made things even worse, playinghavoc with some hedge funds’ strategiesand leaving them scrambling for cash. Reg­ulators are moving towards imposing lim­its on rehypothecation.

Early reform could also come to the se­curities­lending market, in which institu­

tional investors lend shares from theirportfolios to short­sellers for a fee. Somelenders�including, notoriously, AIG�found they were unable to repay cash col­lateral posted by borrowers because theyhad invested it in instruments that hadturned illiquid, such as asset­backed com­mercial paper. Some have doubled theshare of their portfolios that they knowthey can sell overnight, to as much as 50%.

Regulators might consider asking themto go further. Bond markets, unlike stock­markets, revolve around quotes from deal­ers. This creates a structural impediment tothe free �ow of liquidity in strained times,argues Ken Froot of Harvard BusinessSchool, because when dealers pull in theirhorns they are unable to function properlyas market­makers. He suggests opening upaccess to trade data and competition toquote prices. Some senior �gures at the Fedlike the idea, as do money managers,though predictably dealers are resisting.

Twin realitiesThe other brutal lesson of the crisis con­cerns the way liquidity can a�ect solvency.In a world of mark­to­market accounting, asmall price movement on a large, illiquidportfolio can quickly turn into cripplingpaper losses that eat into capital. Highlyrated but hard­to­shift debt instrumentscan �nish you o� before losses on the un­derlying loans have even begun to hurtyour cash �ows. If markets expect �re

sales, potential buyers will hold o� for abetter price, exacerbating fair­value losses.

In future banks will be more alert tothese dangers. �We were looking at thebonds we held, focusing on the credit fun­damentals. We lost sight of the capital hitfrom illiquidity and marking to marketthat can seriously hurt you in the mean­time,� says Koos Timmermans, chief risko�cer at ING, a large Dutch banking andinsurance group. �We now know that youhave to treat the accounting reality as eco­nomic reality.�

Another lesson is the �opportunity val­ue� of staying liquid in good times, saysAaron Brown, a risk manager with AQR, ahedge fund. In an e�cient market dollarbills are not left lying around. But in the dis­located markets of late 2008 there werelots of bargains to be had for the small mi­nority of investors with dry powder.

For some, though, bigger liquidity pro­blems may yet lie ahead. Some $5.1 trillionof bank debt rated by Moody’s is due tomature by 2012. This will have to be re�­nanced at higher rates. The rates could alsobe pushed up by an erosion of sovereigncredit quality, given implicit state guaran­tees of bank liabilities. And, at some point,banks face a reduction of cut­price liquid­ity support from central banks�o�ered inreturn for often dodgy collateral�whichhas buoyed their pro�t margins. Mortgageborrowers on teaser rates are vulnerable topayment shock. So too are their lenders. 7

THE Delta Works are a series of dams,sluices and dikes built in the second

half of the 20th century to protect the low­est­lying parts of the Netherlands from thesea. They are considered one of the sevenwonders of the modern world. The taskfacing global regulators is to construct the�nancial equivalent of this protective net­work, said Jean­Claude Trichet, presidentof the European Central Bank, in an inter­view last November.

This will require success in three con­nected areas: reducing the threat to stabil­ity posed by �rms deemed too big to failbecause their demise could destabilisemarkets; ensuring that banks have biggercushions against losses; and improvingsystem­wide, or macroprudential, regula­

tion. The work is under way, but some bitsare hobbled by a surfeit of architects, oth­ers by a lack of clear plans. ¹ Too big to fail. Dealing with �systemical­ly important� giants is the thorniest pro­blem. Having once been cornered into achoice between costly rescues and gut­wrenching failures, governments are de­termined to avoid a repeat. When marketsswooned, they were obliged to stand be­hind the big and the highly connected (aswell as their creditors), but found them­selves ill­equipped. Tim Geithner, Ameri­ca’s treasury secretary, said his administra­tion had nothing but �duct tape and string�to deal with American InternationalGroup (AIG) when it tottered.

The problem has only worsened dur­

ing the crisis. After a quarter­century ofever­increasing �nancial concentration,the giants of �nance grew even more dom­inant in 2008­09 thanks to a series of shot­gun takeovers of sickly rivals (see chart 5,next page).

Regulators can tackle the issue either byaddressing the �too big� part (shrinking orerecting �rewalls within giants) or the �tofail� bit (forcing them to hold more capitaland making it easier to wind down bust�rms). Until recently the focus was on thesecond of these approaches. But since Pres­ident Obama’s unveiling of two initiativeslast month�a tax on the liabilities of bigbanks and the �Volcker rule�, which pro­posed limits on their size and activities�momentum has been shifting towards

Fingers in the dike

What regulators should do now

some combination of the two.The Volcker plan�named after Paul

Volcker, the former Federal Reserve chair­man who proposed it�calls for deposit­takers to be banned from proprietary trad­ing in capital markets and from investing inhedge funds and private equity. The Finan­cial Stability Board (FSB), a Basel­basedbody that is spearheading the internation­al reform drive, gave it a cautious welcome,stressing that such a move would need tobe combined with tougher capital stan­dards and other measures to be e�ective.

The Volcker rule does not seek a fullseparation of commercial banking and in­vestment banking. Nor is America pushingto shrink its behemoths dramatically; formost, the plan would merely limit furthergrowth of non­deposit liabilities (there isalready a 10% cap on national market sharein deposits). O�cials remain queasy aboutdictating size limits. Citigroup’s woes sug­gest a �rm can become too big to manage,but JPMorgan Chase and HSBC are strikingcounter­examples.

For all the hue and cry about theVolcker plan, America sees it as supple­menting earlier proposals, not supplantingthem. The most important of these is animproved �resolution� mechanism for fail­ing giants. Standard bankruptcy arrange­ments do not work well for �nancial �rms:in the time it takes for a typical case togrind through court, the company’s valuewill have evaporated.

America’s resolution plan would allow

regulators to seize and wind down basket­cases. The challenge will be to convincemarkets that these measures will not turninto life­support machines. Worse, there isno international agreement on how tohandle the failure of border­straddling�rms, nor is one close. That was a huge pro­blem with Lehman Brothers, which hadnearly 3,000 legal entities in dozens ofcountries. And the struggle to retrieve $5.5billion that a bust Icelandic bank owescreditors in Britain and the Netherlandsstill continues.

Questions also linger over the treat­ment of lenders. America’s plan wants itboth ways, giving regulators discretion tooverride private creditors but also to subor­dinate the taxpayer’s claims. This fuelsconcerns about handouts to politically fa­

voured groups, as happened in the govern­ment­orchestrated bankruptcy of GeneralMotors. Another worrying precedent wasthe generous treatment of troubled banks’derivatives counterparties in 2008. Allcounterparty trading exposures, to the ex­tent that they are uncollateralised, shouldbe at the bottom of the capital stack, not atthe top. Regrettably, the opposite hap­pened. This prompted a wave of credit­de­fault­swap buying because these contractswere underwritten by the state. �Today,too big to fail means too many counter­party exposures to fail,� says Peter Fisherof BlackRock, a money manager.¹ Overhauling capital requirements. Inthe hope of avoiding having to trigger theirresolution regimes in the �rst place, regula­tors will force banks to strengthen theircapital bu�ers. A number of countries areconsidering a punitive capital surchargefor the largest �rms. A report from the Bankof England last November suggested va­rious ways of designing this. It could varyby sector, allowing regulators to in�uencethe marginal cost of lending to some of themore exuberant parts of the economy. Or itcould re�ect the lender’s contribution tosystemic risk, based on its size, complexityand the extent of its connections to other �­nancial �rms.

How such a penalty would �t withbroader capital reforms is unclear. In De­cember the Basel Committee of supervi­sors and central banks laid out proposedrevisions to its global bank­capital regime.These could come into force as early as2012­13. The new standards, dubbed Basel3, are less reliant than the last set of reformson banks’ own risk models. Then the talkwas of capital �e�ciency�. Now it is allabout robustness. With markets alreadydemanding that banks hold more equity, areversal of a long trend of falling ratios isunder way (see chart 6, next page).

Before the crisis banks could get awaywith common equity�the purest form ofcapital�of as little as 2% of risk­weightedassets. The new regulatory minimum willnot be clear until later this year, but mar­kets now dictate that banks hold four to�ve times that level. Hybrid instruments�part debt, part equity�will be discouragedsince these proved bad at absorbing losses.Regulators are encouraging banks to issuea di�erent type of convertible capital:�contingent� bonds that automaticallyturn into common shares at times of stress.

In another acknowledgment that rely­ing too heavily on internal models was amistake, the new rules will be supple­mented by a �leverage ratio�. Not weighted

99 01 03 05 07

5Big banks get bigger

Source: “Banking on the State” by Andrew Haldane andPiergiorgio Alessandri; Bank for International Settlements

Top five global banks and hedge fundsAssets as % of industry total

0

5

10

15

20

25

30

1998 2000 02 04 06 08 09

Global banks

Hedge funds

The Economist February 13th 2010 A special report on �nancial risk 11

2

1

12 A special report on �nancial risk The Economist February 13th 2010

2

1

to risk, this measure looks appealingly sim­ple these days. One aim is to curb gamingof risk­based requirements: Europeanbanks, which unlike American ones werenot subject to a leverage ratio, could taketheir borrowing to dangerous heights be­cause many of their assets were highly rat­ed securities with low risk weightings.One o�cial likens the new approach toplacing a net under a trapeze artist.

In an equally big philosophical shift,the new measures will lean against �procy­clicality�, or the tendency of rules to exag­gerate both the good and the bad. Bankswill be required to accumulate extra capi­tal in fat years that can be drawn upon inlean ones. Until now the rules have en­couraged higher leverage in good timesand much lower in bad times, adding todistress at just the wrong moment. Securi­ties regulators contributed to the problem,frowning on boom­time reserve­buildingas possible pro�t­smoothing in disguise.The new proposals will encourage �dy­namic� provisioning, which allows banksto squirrel away reserves based on expect­ed losses, not just those already incurred.

Addressing procyclicality will also re­quire tackling issues that straddle capitalrules and accounting standards. Critics offair­value (or �mark­to­market�) account­ing, which requires assets to be held atmarket prices (or an approximation), com­plain that having to mark down assets tothe value they would fetch in illiquid mar­kets is likely to exacerbate downturns. Thesolution is not to abandon fair value,which investors like because it is less opento manipulation than the alternatives. Butthere is a case for decoupling capital andaccounting rules, says Christian Leuz ofthe Booth School of Business. This wouldgive bank regulators more discretion to ac­cept alternative valuation methods yet stillallow investors to see the actual or estimat­ed market value.

There are lots of potential devils in thedetails of the proposals. A leverage ratio ispointless without strict monitoring of as­sets parked o� balance­sheets. Contingentcapital, meanwhile, could have the oppo­site e�ect of that intended if the bank’strading partners �ee as its ratios near thetrigger point. There are also worries overincreases in capital charges for securitisa­tions, exposure to swap counterpartiesand the like. These make sense in theory;to treat mortgage­backed securities as al­most risk­free was nonsense. But the newrules swing too far the other way, threaten­ing to choke o� the recovery of asset­backed markets.

America’s large banks, having repaidtheir debts to taxpayers, are sure to wagewar on higher capital standards. An impactassessment stretching over several monthswill give them ample opportunity to lookfor holes�and to lobby. In Europe, wherebanks were more highly leveraged andthus face a more wrenching adjustment,even some supervisors are queasy. ¹ Improving macroprudential regulation.In the meantime regulators can make pro­gress in other areas, such as overhaulingday­to­day supervision. In both Americaand Europe they have stepped up compari­sons of pay, lending standards and the likeacross big �rms. They are also introducingpeer review. Within the agency that over­sees Swiss banks, for instance, the lead su­pervisors of Credit Suisse and UBS arenow expected to scrutinise each other’swork. America’s Securities and ExchangeCommission, whose failures included neg­ligible supervision of investment banksand the Mado� scandal, has set up a newrisk division packed with heavyweightthinkers such as Henry Hu, Gregg Bermanand Richard Bookstaber. Part of their jobwill be to scan derivatives markets, hedgefunds and the like for any emerging threatsto stability.

This stems from a recognition that tradi­tional oversight needs to go hand in handwith the macroprudential sort that takesaccount of the collective behaviour of �­nancial �rms, contagion e�ects and so on.�Finance is full of clever instruments thatwork as long as the risk is idiosyncratic, butcan wreak havoc if it becomes systemic,�says Frederic Mishkin of Columbia Uni­versity. Moreover, the crisis showed howrisk can cross traditional regulatory lines.Pension funds and insurers, previouslyseen as shock­absorbers, were revealed aspotential sources of systemic risk.

However, there is no broad agreement

on how systemic regulation might work,or who should do the regulating. Mosteconomists see the job falling naturally tocentral banks, because of their closeness tomarkets and because of the link betweencapital standards and monetary policythrough the price of credit. But there arepolitical obstacles, particularly in America,where a large and vocal contingent in Con­gress accuses the Fed itself of being a threatto stability, pointing to loose monetarypolicy as a cause of the housing mania.

International co­ordination is equallytricky. The FSB has singled out 30 of thelargest banks and insurers for cross­borderscrutiny by �colleges� of supervisors.There is, though, a natural limit to co­oper­ation. It remains to be seen how well na­tional risk regulators work with suprana­tional bodies such as the EuropeanUnion’s systemic­risk council and the FSB.Private­sector groups want to have theirsay too: the Market Monitoring Group, acollection of grandees linked to a banking­industry group, is already issuing warn­ings about fresh bubbles emerging.

Another reason for scepticism is the dif­�culty of identifying a systemic event.AIG’s liquidity crunch was thought tocount as one at the time, hence the o�er ofan $85 billion emergency loan from theFed. But what exactly was the danger? Thatmarkets would be brought to their kneesby the failure of its derivatives counterpar­ties (who were controversially paid o� atpar)? Or by trouble at its heavily regulatedinsurance businesses? More than a yearlater, no one seems sure.

Pricking bubbles�another mooted rolefor systemic regulators�is also fraughtwith danger. Many central bankers consid­er it unrealistic to make prevention of as­set­price bubbles a speci�c objective ofsystemic oversight. But thinking at the Fedhas been shifting. Under Alan Greenspanits policy had been to stand back, wait forthe pop and clean up the mess. But BenBernanke, the current chairman, recentlybacked the idea of intervening to take theair out of bubbles. This could be donemainly through stronger regulation, hesuggested, though he did not rule out mon­etary policy as a back­up option.

Mr Mishkin, a former Fed governor,draws a contrast between credit­boombubbles and irrational exuberance instockmarkets, such as the dotcom bubble.The �rst is more dangerous, and the casefor pre­emptive action stronger, he argues,because it comes with a cycle of leveragingagainst rising asset values.

In retrospect all crashes look inevitable.

6Threadbare cushion

Source: “Banking on the State” by Andrew Haldane andPiergiorgio Alessandri; Bank for International Settlements

Banks’ capital ratios, %

0

5

10

15

20

25

1880 1900 20 40 60 80 2005

United States

Britain

The Economist February 13th 2010 A special report on �nancial risk 13

2

1

RISK antennae twitch after a crisis. Bank­ers, regulators and academics, shaken

from their complacency, jostle to identifythe next tempest. Right now gusts areblowing from several directions. Manycountries’ �scal positions are deterioratingfast after costly interventions to shore up �­nancial systems and restore growth. Thereis talk of demanding collateral even ondeals with formerly unimpeachable sover­eign entities. Recent terrorist incidentshave raised the spectre of external shocks.

Yet at least a fragile sort of optimism hassurfaced, born of ultra­cheap money andrelief at having avoided a depression. Insome markets fresh bubbles may be form­ing. Stockmarkets have rebounded sharp­ly. America’s, though still well o� theirpeaks, are up to 50% overvalued on a his­torical basis. Banks are once again throw­ing money at hedge funds and private­equ­ity �rms (though with tougher marginrequirements). Issuance of structured­loanfunds, which a few months back lookeddead, is booming. Investment banks’ pro­�ts, and bonus pools, are back near pre­cri­sis levels. International regulators havebeen issuing warnings to chief executivesabout a return of irrational exuberance.Banks have been ordered to run stress testsinvolving a sudden jump in interest rates,

in preparation for central banks’ with­drawal of monetary adrenaline.

Many will already be doing this as theytry to show that they have learnt their les­sons. Like the best chess players, bankersinsist that they are now concentrating ashard on avoiding mistakes as on winning.Those that sidestepped the worst mort­gage­related landmines now top the indus­try’s new order. Blackrock’s Mr Fisher de­�nes risk as �deviation from objective�, onthe upside as well as the downside. If yourmodels tell you that a security is safer thanits returns imply, as with CDOs, that mightjust suggest hidden risks.

Fancy mathematics will continue toplay a role, to be sure. But �nance is notphysics, and markets have an emotionalside. In their struggle with the quants,those who would trust their gut instincthave gained ground.

Learning to tie knotsGovernments are taking no chances.Bloomberg counts some 50 bills and otherserious proposals for �nancial reform inAmerica and Europe. Leaders in America’sSenate hope to pass new rules by March.But there are limits to what can be expect­ed from regulators and supervisors. Likebankers, they have blind spots. As the

mortgage �asco showed, they are vulner­able to capture by those they police.

Their job will be made easier if newrules tackle incentives for the private sec­tor to take excessive risk. It is human be­haviour, more than �nancial instruments,that needs changing, says Mr Mauboussinof Legg Mason. Like Odysseus passing thesirens, bankers need to be tied to a mast tostop them from giving in to temptation.Pay structures should be better alignedwith the timescale of business strategiesthat run for a number of years and shouldnot reward �leveraged beta�, unremark­

Blocking out the sirens’ song

Moneymen need saving from themselves

7Parti pris

Sources: Company accounts; Bloomberg; Benn Steil

Global asset-backed-securities issuanceand Moody’s profit per employee

0

0.25

0.50

0.75

1.00

1.25

0

50

100

150

200

250

2000 01 02 03 04 05 06 07 08

Total ABS issuance,$trn

Profit per employee,$’000

Even with the most insidious­looking bub­bles, though, it is impossible to know at thetime how devastating the pop will be.Many thought the economic fallout fromthe internet crash would be far greaterthan it turned out. The economic cost ofprematurely ending a boom can be high.Even so, the worry is that a systemic regu­lator would be biased towards interven­tion, because it would face less criticism forpuncturing a non­bubble than for failing tospot a real one.

Alex Pollock of the American Enter­prise Institute (AEI), a think­tank, is con­cerned that the creation of an o�cial sys­temic regulator would bring false comfort,just as the Fed’s founding in 1913 did. Ac­cording to the then Comptroller of the Cur­rency, it had rendered further crises �math­ematically impossible�. Mr Pollock wouldprefer to see the task go to an independentadvisory body, manned by economic

heavyweights to provide institutionalmemory of past crises. For similar reasons,Andrew Lo, director of MIT’s Laboratoryfor Financial Engineering, suggests sepa­rating regulation from forensics, as hap­pens in the airline industry. America’s Na­tional Transportation Safety Board is seenas independent because its job is to investi­gate crashes, not to set rules after the event.That gives it more moral clout.

Whatever form it takes, systemic polic­ing would face a problem. During booms,governments are loth to take the punch­bowl away, at least until the next election.Nor do they want to be criticised for theirown contribution to systemic risk. Theymay have become even touchier now thatthey are, as Pimco’s Mohamed El­Erianputs it, �market players as well as referees�.

A way round this would be to introducerules requiring the regulator to step in if,say, credit and asset prices are growing at

above­average rates. That would shield itfrom claims that the next boom is some­how di�erent and should be left to run itscourse. But it comes at the cost of �exibility.

All of this suggests that although thereis a strong case for a more system­wide ap­proach to oversight, it could do more harmthan good if poorly crafted. Meanwhiletaxpayers will continue to underwrite �­nancial giants; America’s reforms in theircurrent shape allow the authorities to pullapart those that pose a �grave threat�, butalso to bail out their creditors if they con­sider it necessary.

The danger is that the very existence ofa systemic regulator creates an illusion ofincreasing stability even though it does theopposite by strengthening the implicitguarantee for the biggest banks�a �perma­nent TARP�, as the AEI’s Peter Wallisonputs it. Too big to fail sometimes seems toohard to solve. 7

14 A special report on �nancial risk The Economist February 13th 2010

2

Previous special reports and a list offorthcoming ones can be found online

Economist.com/specialreports

Future special reportsManaging information February 27thGermany March 13thAmerica’s economy April 3rdManagement innovation in emergingmarkets April 17thTelevision May 1stInternational banking May 15th

Economist.com/rights

O�er to readersReprints of this special report are available at aprice of £3.50 plus postage and packing. A minimum order of �ve copies is required.

Corporate o�erCustomisation options on corporate orders of 100or more are available. Please contact us to discussyour requirements.

Send all orders to:

The Rights and Syndication Department26 Red Lion SquareLondon WC1R 4HQ

Tel +44 (0)20 7576 8148Fax +44 (0)20 7576 8492e­mail: rights@economist.com

For more information and to order special reportsand reprints online, please visit our website

able returns juiced with borrowed money.In securitisation, originators will have todisclose more information about loanpools and hold a slice of their products.

Some of the worst abuses in securitisa­tion stemmed from the use of credit rat­ings. Rating agencies systematically under­estimated default risk on vast amounts ofdebt, resulting in pu�ed­up prices and asurfeit of issuance. Paid by issuers, theyhad every incentive to award in�ated rat­ings and keep the market humming: aver­age pay at the agencies rose and fell in tan­dem with the volume of asset­backedissuance (see chart 7, previous page).

An obvious way to deal with thiswould be to eliminate the agencies’ o�cial�nationally recognised� status, openingthe business to unfettered competition.Raters would then have to persuade inves­tors of their competence, rather than rely­ing on a government imprimatur. This, inturn, would force investors themselves tospend more time analysing loans. Oddly,proposed reforms fall far short of this, withno sign of anything tougher on the hori­zon. CreditSights, a research �rm, awardedratings �rms its �Houdini was an Ama­teur� award for 2009.

Nor, alas, is there much appetite totackle some of the public policies that con­tributed to the crisis. The non­recourse sta­tus of mortgages in large parts of America,for instance, gives the borrower a highly at­tractive put option: he can, in e�ect, sell thehouse to the lender at any time for the prin­cipal outstanding. An even bigger problemis the favourable tax treatment of debt rela­

tive to equity. Phasing this out would dis­courage the build­up of excessive leverage.But the idea has little political traction.

There are, to be sure, risks to rushing re­form. Post­crisis regulation has a long his­tory of unintended consequences, fromthe pay reforms of the early 1990s (whennew limits on the deductibility from cor­porate tax of executive salaries merelyshifted the excesses to bonuses) to keyparts of the Sarbanes­Oxley act on cor­porate governance. Another danger is thepricing of risk by regulation, not markets.The credit­card act passed in America lastyear leaves providers with a choice be­tween underpricing for some products,which is bad for them, or restricting their

o�erings, which is bad for consumers.Most would agree, however, that mar­

kets need both tighter rules and better en­forcement. The biggest question markhangs over the fate of those institutionswhose collapse would threaten the sys­tem. America’s proposal to cap banks’ sizeand ban proprietary trading has forti�edthose calling for radical measures to tacklethe �too big to fail� problem.

The virtues of scepticismBy itself, though, the plan does little to backup Barack Obama’s promise to stop such�rms from holding the taxpayer hostage.Proprietary trading and investments are asmall part of most big banks’ activities andplayed only a minor role in the crisis. Nordoes the plan cover brokers, insurers or in­dustrial �rms’ �nance arms, all of whichhad to be bailed out. To persuade marketsthat the giants no longer enjoy implicitstate guarantees, whether they are banksor not, policymakers will need to present acocktail of solutions that also includetougher capital and liquidity standards,central clearing of derivatives and crediblemechanisms to dismantle �rms whoselosses in a crisis would overwhelm even astrengthened safety bu�er.

Together, intelligent regulatory reformsand a better understanding of the limita­tions of quantitative risk management canhelp to reduce the damage in�icted on the�nancial system when bubbles burst. Butthey will never eliminate bad lending orexcessive exuberance. After every crisisbankers and investors tend to forget that itis their duty to be sceptical, not optimistic.In �nance the gods will always �nd a wayto strike back. 7