What this covers

Russ Roberts of EconTalk interviews Riccardo Rebonato, then of the Royal Bank of Scotland and author of Plight of the Fortune Tellers, on the sources and lessons of the 2008 financial crisis. The conversation centers on Rebonato's core argument: that regulators and financial institutions placed unwarranted faith in quantitative risk models to forecast events so rare that historical data cannot support reliable estimates, while regulatory philosophy assumed self-interested market participants would police themselves despite the absence of systemic safeguards.

Rebonato traces how this worldview broke down in practice. Value at Risk measures short-term volatility competently but was misapplied to set capital reserves meant to cover once-in-thousands-of-years losses—a fundamental category error. The 99.975% solvency standard, he explains, was reverse-engineered from credit ratings rather than derived from catastrophe modeling. Banks gamed regulatory books by sheltering risky assets in trading accounts, attracting lower capital requirements through abundant liquidity until that liquidity evaporated. Competition and career risk locked prudent managers into the bubble: stepping out early meant years of explaining underperformance to shareholders before vindication. Securitization and rating-agency conflicts receive sustained examination, as does the tension between what is rational for an individual firm—accepting leverage and tail risk—and what is rational for the financial system as a whole. On regulation, Rebonato argues the problem was not markets' lack of self-interest but rather that self-interest, when rational at the firm level, need not be rational systemically. He pushes back on several post-crisis scapegoats: mark-to-market accounting, for instance, becomes problematic only when liquidity vanishes, not because the principle is wrong. The too-big-to-fail moral hazard argument, he suggests, is overstated when examined at the level of actual decision-makers facing real losses.

Sharpest takeaway

Rebonato argues that the financial crisis stemmed from misplaced faith in quantitative risk models that promised knowledge of extremely rare events the data cannot support, compounded by a regulatory philosophy trusting self-interested actors who, while individually rational, ignored systemic externalities.

  • Risk models calibrated on a few years of data cannot credibly estimate once-in-thousands-of-years events
  • Value at Risk measures day-to-day volatility well but was misused to set capital meant to absorb extreme losses
  • A risk rational for a private firm need not be rational for the system, which is where regulators must intervene

The claims · ranked34 claims · weighted by value

This asset isn't compiled yet

You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.

0.90

A prudent manager who stepped out of the bubble early would face a disastrous competitive and career penalty: with Greenspan's 'irrational exuberance' warning coming years before the burst, anyone exiting early would have to explain years of lagging returns to shareholders while everyone else got rich, and would likely be fired before being vindicated.

causalhigh valueestablishednovelty 3/4durability 4/4· Riccardo Rebonato

“if someone steps out a bit too early it is absolutely disastrous”

0.86

There is an inherent tension between bondholders and shareholders over capital levels: bondholders care only about solvency and would prefer triple-A safety, while shareholders care about return on capital and dislike holding excess capital; for most banks the equilibrium settled at the Double-A level because shareholders, holding a call option on the firm's assets, like upside volatility that bondholders never share.

causalhigh valueestablishednovelty 3/4durability 4/4· Riccardo Rebonato

“there is this compromise between the shareholder and the bond holder and for most bank the the equilibrium has been found around the double-a level”

0.86

As risk premia were compressed across all risky assets, investors had to take more and more risk to earn the same absolute return, since less and less compensation was being paid for the same amount of risk.

causalhigh valueestablishednovelty 3/4durability 4/4· Riccardo Rebonato

“for the same amount of risk less and less premium less and less compensation was being paid so if you wanted to make the same absolute return you had to take more and more recently”

0.81

Banks can choose to place a security in the trading book (which attracts less capital but requires daily mark-to-market) or the banking book (higher initial capital but value recognized at par absent impairment); in the blue-sky years to 2006-07 banks preferred the trading book for its lower capital and abundant prices, then got trapped when liquidity and prices for those instruments disappeared.

causalhigh valueestablishednovelty 3/4durability 3/4· Riccardo Rebonato

“typically the trading book attracts less capital but it forces you to the discipline of a daily marked market so perhaps people have been placing into the trading book when the sky was blue”

0.80

Bear Stearns took such a catastrophic gamble on the US housing market because of extreme leverage: having borrowed so much to finance the position, losing the bet did not merely mean a bad quarter—it meant the death of the firm.

causalhigh valueestablishednovelty 2/4durability 4/4· Russ Roberts

“the simple answer is well because they were so leveraged they had borrowed so much to finance that gamble that losing the gamble didn't just mean a bad quarter it meant death”

0.79

The 99.975% (one-in-4000-years) economic-capital standard was not derived from extreme-event modeling but reverse-engineered from credit ratings: historically about one in 4000 Double-A institutions defaulted within a year, so holding enough capital to default only that rarely signals to bondholders that the bank is Double-A and deserves cheap funding.

causalhigh valuecontestednovelty 4/4durability 3/4· Riccardo Rebonato

“if I divide this very small number companies which defaulted in one year from Double A down to default over the total number of companies I come up with this one in four thousand approximately”

0.78

A perfectly rational, enlightened shareholder in mid-2007 could rationally accept a level of risk a regulator would not allow, because insuring against an extreme repeat-of-the-1930s scenario would lock in near-certain losses they likely never have to bear—and this is precisely where regulators must step in, since shareholders' risk-reward calculus ignores systemic impacts.

causalhigh valuecontestednovelty 3/4durability 4/4· Riccardo Rebonato

“a shareholder can be perfectly enlightened can be perfectly rational but can be willing to accept a level of risk that a regulator might not like”

0.78

The argument that going public created a damaging asymmetry of incentives between traders/executives and stockholders (who would never have gambled with their own money) sounds right after the fact, but before the fact—in 2006 to early 2007—the risk was reasonable for a private firm; the problem is that a reasonable risk for a private institution need not be reasonable for the system.

causalhigh valuecontestednovelty 3/4durability 4/4· Riccardo Rebonato

“a reasonable risk for a private institution needn't be a reasonable risk for a system and it is where the regulator should step in creating a level playing field”

0.78

We are not in a position to say anything reliable about extremely rare events because risk data only cover a few hundred to a few thousand relevant points spanning a few years, and even abundant historical data require an enormous leap of faith to assume relevance to present market conditions.

factualhigh valuecontestednovelty 3/4durability 4/4· Riccardo Rebonato

“we really are not in a position of saying anything about extremely rare events because we are collecting data with different frequencies”

0.78

Blaming mark-to-market accounting for the crisis is like saying a patient wouldn't have a fever without a thermometer; mark-to-market gives good information about fundamental value only when there is a willing buyer and seller in a liquid market, and ceases to be meaningful when liquidity disappears and the 'pseudo-arbitrageurs' lack firepower to push prices back toward fundamentals.

causalhigh valuecontestednovelty 3/4durability 4/4· Riccardo Rebonato

“it's like saying if I didn't have a thermometer this patient wouldn't have a fever to me that doesn't sound quite right”

0.78

Value at Risk works for typical 20-day or 100-day horizons but was misused for capital requirements, because capital exists to absorb losses occurring once every hundreds of years to protect solvency, not the frequent losses VaR measures—so the two purposes are fundamentally mismatched.

causalhigh valuecontestednovelty 3/4durability 4/4· Riccardo Rebonato

“capital is there not to absorb losses that occur once every month or once every few months but it should be there in order to absorb losses that occur well once every hundreds of years”

0.78

There is a destructive interaction between mark-to-market accounting and regulatory capital: an asset intended to be held to maturity and fine on a cash-flow basis can have its market value drop when foreclosure rates rise, pushing the holder out of capital compliance and forcing an economically unnecessary sale—the one part of the mark-to-market critique that is genuinely compelling.

causalhigh valuecontestednovelty 3/4durability 4/4· Russ Roberts

“I may not be in compliance with my capital requirements today but over time I might be fine in a cash flow basis I think that's the only part of the mark-to-market argument that I find somewhat compelling because of how it interacts with my capital buffer”

0.77

The too-big-to-fail moral-hazard argument is weak when examined at the level of individual decision-makers rather than abstract institutions: while a firm might be bailed out, the individuals who made the choices are not, and the CEOs who made those decisions are faring badly today, so they were unlikely to be gambling on a rescue.

causalhigh valuecontestednovelty 3/4durability 3/4· Riccardo Rebonato

“perhaps the institution might survive but as we've seen the individuals don't and it is the individuals in know the institution who made those choices”

0.75

The lack of consensus in economics about the causes, recovery, and lessons of the Great Depression illustrates the difficulty of theorizing from one-time events: it is easy to construct a theory of a single event but hard to be confident it applies today, which should induce humility.

factualhigh valueestablishednovelty 2/4durability 3/4· Russ Roberts

“it's rather remarkable that we don't really have a consensus in economics or economic history about the cause of the depression”

0.75

Rating agencies faced a corrupting conflict of interest: they were paid only when a securitization went through and actively worked with investment houses to structure portfolios to meet the very rating criteria they had set—analogous to a teacher who sets a tough test but hands the students the questions in advance.

causalhigh valueestablishednovelty 2/4durability 3/4· Riccardo Rebonato

“the rating agencies only cost paid even securitization went through and the rating agencies actively worked with the investment houses to structure the portfolio so that it would meet the criteria”

0.75

Regulators sensibly applied a multiplier of three to VaR-based calculations, following the engineering tradition of safety factors (e.g., multiplying by ten for elevator cables, 1.1 for aircraft landing gear), translating a meaningful statistical number into a safety-adjusted one.

factualhigh valueestablishednovelty 2/4durability 3/4· Riccardo Rebonato

“regulators I think with a lot of common sense slapped on a multiplier multiplier of three”

0.75

Securitization works by pooling diversified assets (e.g., mortgages from different regions) to create a portfolio less risky than its components, then tranching it so the lowest tranche bears first losses for high return up to a Triple-A top tranche—with rating agencies meant to be impartial judges of whether the diversification truly justified those ratings.

definitionhigh valueestablishednovelty 2/4durability 3/4· Riccardo Rebonato

“securitisation is the process whereby in banks accumulate assets accumulate mortgages or other type of acid package them”

0.73

The defense of the rating-agency conflict at the time was a franchise-value argument rooted in the prevailing enlightened-self-interest mantra: agencies were assumed to protect the currency of their name, since debasing it by giving away Triple-A would destroy their credibility in an efficient self-regulating market.

factualhigh valuecontestednovelty 3/4durability 3/4· Riccardo Rebonato

“there is a value of our franchise remember who it was retained we were only talking about 18 to 24 months ago when the general mantra was the enlightened self-interest of market players was the best safeguard”

0.73

Basel II was a worthy attempt to align required capital more closely with actual risk, fixing Basel I's crude rules (which demanded the same capital for lending to IBM as to a corner shop, creating an incentive to lend to riskier borrowers for more bang per buck), but it suffered from overconfidence in how much risk can be quantified using frequentist statistical techniques.

causalhigh valuecontestednovelty 3/4durability 3/4· Riccardo Rebonato

“if I have to set aside the same amount of money for $1,000 lent to IBM or lent to the corner shop well I get a much bigger bang for the buck if I lend the same amount of money to the corner shop”

0.73

Risk aversion will rebound only modestly if a quick V-shaped recovery occurs—disappearing faster than expected—whereas a protracted period of decline will make actors more fundamentally cautious; the same search for yield that created the crisis can recreate instabilities if recovery is rapid.

forecasthigh valuecontestednovelty 3/4durability 3/4· Riccardo Rebonato

“if we get out in as if plan a works beautifully and we get out with the v-shape recovery I would expect the risk aversion will disappear much more quickly than we think possible today”

0.71

Models promising to characterize events occurring once every 100, 1000, or even 4000 years cannot be validated because we cannot know that the data used to calibrate them are sufficiently homogeneous—belonging to the same patch of history—to be relevant today rather than to a period when the world was quiet.

causalhigh valuecontestednovelty 3/4durability 4/4· Riccardo Rebonato

“these models were promising to tell you about event that should happen once every 100 years once every 1000 years once every four thousand years”

0.70

Banks are unlike other industries because they carry systemic implications: asymmetry of information and the possibility of bank runs mean an individual bank's insolvency can spread and paralyze the entire financial system, justifying a regulatory goal of preventing insolvency.

causalhigh valueestablishednovelty 1/4durability 4/4· Riccardo Rebonato

“a bank is not like any other industry but a bank as a systemic implication and there is asymmetry of information there are possibility of runs of the bank”

0.69

Central bankers were making it up as they went along, engaged in intense internal debate over whether to let markets work out their own problems or intervene; those who opposed the Bear Stearns rescue and were more market-oriented gained the upper hand by the time Lehman came around, only to reverse course again after witnessing the consequences.

factualhigh valuecontestednovelty 3/4durability 2/4· Riccardo Rebonato

“everybody including the regulator's including the best minds in central banks they will make any stuff as they were going along”

0.68

Banking regulation in the 1990s came to rely increasingly on quantitative models, reflecting an underlying philosophy of faith in the self-regulating properties of a system managed by enlightened, self-interested managers.

causalhigh valuecontestednovelty 2/4durability 3/4· Riccardo Rebonato

“the regulator's who are look after the soundness of a banking system of the international banking system came to rely more and more on on models on quantitative models”

0.68

Events of this magnitude have no single cause but result from a 'perfect storm' of five or six factors arriving simultaneously—including trade-surplus countries like China investing in Treasuries, which depressed yields and pumped liquidity into the system—so US housing-subsidy policies were aiding and abetting but not the sole cause.

causalhigh valuecontestednovelty 2/4durability 3/4· Riccardo Rebonato

“when something of a magnitude of what we have been observing occurs there is no single cause it is a confluent it is a perfect storm you have five or six weather fronts arriving at the same point”

0.68

The Lehman bankruptcy triggered genuine panic and incipient bank runs worldwide because, once an institution as illustrious as Lehman was not rescued, no bank felt safe and the slightest noise prompted depositors to withdraw funds—contradicting the view that the TARP bailout, not Lehman, spiked spreads.

causalhigh valuecontestednovelty 2/4durability 3/4· Riccardo Rebonato

“when people saw that even a bank is illustrious as Lehman and not been rescued there was the world a sense of panic and ransom banks were beginning to happen all over the world”

0.68

By 2006 there was a generalized consensus that abundant liquidity, compressed risk premia, and inflated asset prices were unsustainable—like the fracture line of an earthquake—yet it was nearly impossible to predict where or when the crack would occur.

factualhigh valuecontestednovelty 2/4durability 3/4· Riccardo Rebonato

“in 2006 there were plenty of signs that money was too easy liquidity was too abundant risk premia had been compressed too much asset prices were too high”

0.63

The quants who built risk models typically came from physics and mathematics backgrounds with only a smattering of finance training; they were skilled implementers but not deep thinkers about banking systems, making them instruments rather than designers of the models.

factualhigh valuecontestednovelty 2/4durability 2/4· Riccardo Rebonato

“they tended to come from a background very often physics very often mathematics it was only peripherally touching very often they would have a smattering of a course in mathematical finance just before joining a bank but the quant were not deep thinkers at all”

0.61

Value at Risk is a risk statistic expressed in money: a one-day 95th-percentile VaR of $20 million means losses should exceed $20 million only five days out of a hundred, making it a meaningful and statistically estimable measure of typical day-to-day trading volatility.

definitionhigh valueestablishednovelty 1/4durability 3/4· Riccardo Rebonato

“if I say your value at risk one day 95th percentile is twenty million what does that mean it means that I should exceed a loss of 20 million only five days out of the hundred”

0.61

Housing became the focus of the bubble because, with risk premia compressed across all assets, investing in something as 'solid as houses' looked obvious; instruments like home-equity refinancing turned housing into an investment rather than a place to live, and this bubble burst with particular severity because of the magnitude and the complexity/leverage of the mortgage-backed instruments attached to it.

causalhigh valuecontestednovelty 2/4durability 3/4· Riccardo Rebonato

“for many investors the obvious things solid as houses was to invest in houses”

0.60

Banks that fared worst were often latecomers who began building their portfolios when the party was nearly over, while a few genuinely smart firms (like Goldman Sachs and JP Morgan) profited; the different levels of toxic inventory simply reflected the different stages institutions were at in entering the business.

causalhigh valuecontestednovelty 2/4durability 2/4· Riccardo Rebonato

“they were the late comers in these fared worse than everyone else because they're not made the money before they were seeing everybody at the party and they wanted to join the party so they begin building their portfolios basically when the party was about to be about to be over”

0.35

The 1998 LTCM/Russia-default episode proved, in retrospect, to be purely contained to the financial sector and did not spill over into the real economy, becoming just a forgotten blip within six to nine months.

factualestablishednovelty 1/4durability 3/4· Riccardo Rebonato

“there was the hiccup of 98 with SCCM Russia default but that a posteriori after the fact they proved to be purely contained to a financial sector it did not spill over to the real economy”

0.25

There is currently an enormous amount of uncertainty about whether recovery will be quick or whether we are at the beginning of a prolonged depression, and in such an unstable state every piece of arriving information can shift the market mood dramatically from week to week.

factualestablishednovelty 1/4durability 1/4· Riccardo Rebonato

“every piece of information that arrives as the ability to shift the mood very very dramatically because we're in a very uncertain in a stable state”

0.12

I'll now describe and define value at risk to go into more depth on it.

factual· Russ Roberts

“describe how it would cite the idea behind it and why it became so popular”