"Crises like the current one are inherently unpredictable. "

Greg Mankiw makes a good point:

"Lessons from the Crisis As seen by Michael Spence.

Mike says a lot of smart stuff in this article. But this sentence seems to veer off in the wrong direction, or at the very least could be easily misinterpreted:

we need a commission of top industry professionals and academics to address the challenge of measuring and detecting systemic risk and provide the underpinning of an effective “early warning” system.

I see little hope of creating any kind of "early warning" system, if by that Mike means better forecasting. Crises like the current one are inherently unpredictable. If they were predictable, hedge funds and other money managers would not lose so much money during them."

This is a good point, but not dispositive.

"True, a few people were sounding an alarm in advance of the current crisis: Nouriel Roubini, in particular, comes to mind. And a few hedge funds have made money during the crisis. Yet that fact is not very meaningful. Given the diversity of opinion at any point in time, someone will always look right ex post. The key question is whether the event is reliably predictable ex ante.

Policymakers at the Fed and Treasury cannot do better than rely on the consensus judgment of experts, and a couple years ago the consensus opinion was not predicting anything like what is now occurring. To suggest a regulatory system that gives an "early warning" is like saying we need to find a better crystal ball. Good luck with that."

I think that there's a difference between an early warning system, and deciding what the early warning means. I can certainly think of early warning signs to alert us that we're in need of taking a serious inventory of our circumstances, but whether or not anyone would listen to it any more than any other commission or policy group or whatever is dubious. So, in that sense I agree.

"In my view, the key to regulatory reform is not trying to predict the future with more accuracy but, instead, making the system more robust so that the economy functions better when the unpredictable inevitably occurs. In other words, our focus needs to be not on what will happen but on what might happen."

I'm not sure about the difference between "will" and "might" here. I think that he means we shouldn't try to prevent anything, but be better at reacting to anything. But that can't be right. For one thing, the policies you put in place to react to a crisis are open to the same indeterminacy as trying to predict the exact crisis. You might well prepare for exactly the wrong thing. Also, some things can surely be prevented, so why not prevent them, if you can. True, it's hard to prove that you prevented something that didn't occur, but, it's certainly conceivable that an early warning system could prevent two or three crises, while a robust system fails miserably when the crisis actually occurs.

As an epistemological point, then, I don't see that he has drawn any kind of clear or determinative distinction. The solution is to take a look at both preventative measures and crisis management measures, and employ the most sensible mix of measures that we can agree on, knowing that, well, we don't really know exactly what will happen and when.

One note of major disagreement with Spence:

"Systemic risk escalates in the financial system when formerly uncorrelated risks shift and become highly correlated. When that happens, then insurance and diversification models fail. There are two striking aspects of the current crisis and its origins. One is that systemic risk built steadily in the system. The second is that this buildup went either unnoticed or was not acted upon. That means that it was not perceived by the majority of participants until it was too late. Financial innovation, intended to redistribute and reduce risk, appears mainly to have hidden it from view. An important challenge going forward is to better understand these dynamics as the analytical underpinning of an early warning system with respect to financial instability."

See, I don't buy that financial innovation was intended to reduce risk. It was quite clearly meant to increase risk, in the hopes of higher returns. Lowering capital requirements, avoiding regulations that are intended to dampen risk, selling products to people unable to afford them under any reasonable scenario, are in no meaningful sense capable of lessening risk.

They were intended to redistribute and hide risk, but not in the way Mr. Spence says. And this is my main disagreement with Prof. Mankiw; namely, one can imagine an oversight group that examines all products that transfer risk to third parties or magnify risk. Since that is why they will, in fact, be created, it is not inconceivable that they can be monitored or, if deemed necessary, regulated. If they can be created by human comprehension, then they can be understood by human comprehension as well. This is actually a version of the fact that whatever can be meant can be understood.

" Except that last part wasn't true."

Barbara Kiviat of Curious Capitalist fame, around these parts, did a wonderful post, I like calling everything a post nowadays, on the recent evaluation of financial risk:

"Over the past two decades, wall street and the rest of the financial ecosystem became obsessed with the quantification of risk. Assigning numbers to the chance of something bad happening is a centuries-old endeavor--mortality tables have been used to devise insurance premiums since the 18th century. With modern computing power, though, financial engineers captured, packaged and sold risk exposure in startlingly new ways. Buying protection against a bad corn harvest or a spike in interest rates was just the beginning. Over time, as instruments became more complex, a huge shift occurred. Risk itself became the thing to trade--and to make money on. In the process, risk was redistributed to the people who could best handle it, making everyone safer.

Except that last part wasn't true."

Now, Barbara accepts that the models fooled people. I don't. You can only be fooled by models if you're prepared to be fooled. This is similar to the idea that an invention can only work in the right circumstances, or, better yet, a scientific theory can only be accepted with the right outlook by scientists. It's a matter of explaining human agency.

"There are many lenses through which to view the worldwide financial crisis as it has rippled from soured loans to banks' balance sheets to the credit markets and, finally, into the real economy. Lenders were greedy, people lived beyond their means, government abdicated its responsibility to regulate. Through all those explanations runs the thread of risk--and how it was mismanaged. Financiers, as well as the investors who bought their wares and the ratings agencies that evaluated them, agreed that by applying the proper equations it was possible to, say, bundle a bunch of subprime mortgages, chop them up and sell the pieces as fairly safe securities, even as they were leveraged to the hilt. Why? The mathematical models--backward-looking and based on just a few years' data from an asset bubble--said so. "As an industry, we let financial tools be a substitute for human judgment," says Kevin Blakely, CEO of the trade group Risk Management Association. "There's been such an earthquake in the industry, people are saying we have to do it differently."

Actually, I think this says more or less what I'm saying. Let's try putting it another way. Why were people so loathe to accept that there was no ether, relativity, the uncertainly principle, and Godel's proof,were true, but they're jumping overboard in accepting math models predicting markets? I think that you see where I'm going.

"And so risk morphs from a slop trough of moneymaking into a business line like any other, one that requires human beings to ask the big-picture questions that computer models don't, like, What would happen if fundamental assumptions--house prices go up, institutions are too big to fail--prove to be wrong?"

I guess the good news is that we've at least answered that question.

"What was important was that the firm also made a deliberate decision that risk was not something that could be reduced to a number. "We have a much more holistic discussion about a business and why we have it," says vice president and treasurer Helen Shan. "It becomes strategic, instead of simply, Do we get insurance to cover a potential loss?" In a speech at an industry conference in October, Federal Reserve governor Randall Kroszner urged financial firms to take a similar tack. Weighing risks, as well as potential returns, he said, "should be part of the calculus for all decision-making," and "assessing potential returns without fully assessing the corresponding risks to the organization is incomplete, and potentially hazardous, strategic analysis."

I don't like holistic. How about more complete? It's amazing how useful the calculus is. A little note: The planets don't exactly follow the movements of the math. It's simply close enough to be useful to us.

"Taking on risk from instruments like credit-default swaps and collateralized debt obligations (CDOs) was treated as a profit center, often with little oversight of the mathematical models that spit out numbers about what it was all worth. The models proved spectacularly wrong because they precluded the possibility of an outsize event. Once big shocks, like declining home prices, started hitting, the models broke down. According to a report by a group of U.S. and international regulators, while some firms tried to understand what would happen to their models in truly adverse conditions, plenty of others didn't. In some cases, finance outfits barely examined their products, instead relying on the evaluations of outside ratings agencies, even as they accumulated more and more of those assets."

Okay. Let's be AIG. We're in the insurance industry. We now offer credit default swaps. What makes us think we're competent to do that. So, let's write a CDS: You're the lender of $600,000 to borrower to buy a house. If the borrower defaults, and you foreclose upon the house, and the house re-sells for $500,000, I , AIG, will pay you $100,000, for which protection you've been paying premiums. My, it looks like insurance. Which it basically is, except that, for technical reasons, it allows you, with your AAA rating, to put up less capital than on regular insurance, and, indeed, it's that lower capital threshold that led you to CDS's in the first place. Riskier? You bet. See, there's less capital. That's why they have those levels mandated on regular insurance. So, did I say that it was riskier.

Now, when you have insurance on your house, it can include disaster relief. If it does, the insurer usually hopes to have enough money to take care of this, and they charge premiums based on that. It is usually the case that a disaster happens in one particular area of the country, so an insurer can usually handle that. But what would happen if a hurricane hit the whole U.S.? That's basically what happened in the housing crisis.

One caveat that should scare us: Often, in a disaster area, the government spends a lot of money over and above personal insurance. In some areas, we make people whole who live in areas that can't really even be insured properly without enormous premiums.

So, the question is simple: Why didn't AIG prepare for this? It's the models. Fine. Did the models check for this U.S. hurricane? Well, sadly no. We forgot that. What's the model for that? You might want to model, or, at least study, human agency and behavior. Just a thought.

"When managers did articulate problems, they were often ignored. In August 2007, one of Merrill Lynch's top risk managers warned his boss that a decision to wager $3 billion on indexes of mortgage-related securities was too risky. The firm made the bet anyway; three months later, the risk manager left. "The psychology during a boom makes it very difficult to come up with large stress scenarios and get management to consider them to be credible," says Ed Hida, a partner in Deloitte & Touche's risk- and capital-management practice."

I agree with this. It's not a problem of math models though.

"When firms did steer clear of what was, in retrospect, excessive risk, the tone typically came from the top"

Barbara's just a more decent soul than I am. I don't buy it.

"Sage advice, that. But for how long will it be followed? "Risk gets forgotten in all bubbles," says Peter Bernstein, an investment adviser and the author of Against the Gods: The Remarkable Story of Risk. "We've been down this particular road before." Indeed, we have. After every other trauma--the 1987 stock-market crash, the savings-and-loan crisis, the meltdown of the Long-Term Capital Management hedge fund--boisterous, unchecked risk-taking eventually rushed back in. "In times like this, people do listen to risk managers," says John Hull, professor of derivatives and risk management at the University of Toronto. "The problem is, times will become good again, and then you listen to the trader who is making a big profit."

Once again, I agree. Nothing to do with math models.

"It is tempting to say that this time is different, that the pain has been so severe and ongoing, the lessons will be remembered."

I think they said that after WW I.

"But there is also reason to be skeptical. Take, for instance, the idea of risk-adjusted employee pay as a way to keep people, including rank-and-file traders, from following personal incentives to the exclusion of a company's broader interest. It is a compelling idea. But so far it hasn't happened."

Actually, there are fiduciary responsibilty laws on the books. We might try enforcing those.

"And maybe right now we shouldn't be trying quite so hard. With lending locked up the world round, we desperately need some risk-taking. We just don't have to risk everything."

That's one of the most intelligent statements I've read lately.

"that was so simple that even a first-year economics student (or Financial Times journalist) could understand it."

I awoke this morning to a post that agrees with me implicitly. I've finally found another use for this word I've been bandying about so much as of late. From Brad Setser:

“Give me yield, give me leverage, give me return” perfectly sums up how Wall Street bought itself close to financial ruin. JP Morgan’s William Winters didn’t just create help to create CDOs. He seems to have a way with words.

During the past few years, the Street bet – and bet big – on two theories in its quest for higher returns.

The first was that housing prices never fell. At least not on a nation-wide basis. That meant that lending against a diversified pool of housing collateral wasn’t that risky, no matter how risky the individual borrower might be.

The second was that macroeconomic – and financial – volatility had been vanquished. That, in effect, meant it was OK to try to improve returns through the use of borrowed money.

Neither assumption proved true."

Neither assumption was even reasonable for God's sake.

"It turns out that the rapid growth in CDOs stuffed with exposure to risky mortgages — including “synthetic” exposure from writing credit default swaps on bonds backed by subprime debt* — was facilitated by the broker-dealers willingness to hold more credit risk on their own balance sheets. If you have any doubts, read Gillian Tett’s reporting from over a year ago."

What isn't credible is that they didn't understand the risk. Why did they take it?

"Synthetic CDOs were composed out of CDS rather than actual bonds. Selling insurance against default on a bond has many of the same characteristics as actually owning the bond. Both are bets that pay off if nothing goes bad. If you own the bond, you get paid back in full with interest. If you sell insurance you collect the insurance premium and never have to pay out. Clever financial engineers substituted one for the other, as for a while, there was more demand for subprime exposure than actual subprime loans from investors eager for big returns (helped along by investment banks willing to take the parts of the structure that investors didn’t want on their own balance sheets) …"

If you're lending money out on risky terms, buying insurance makes sense.

"Most were not actual securities, but derivatives linked to triple-B-rated mortgage securities. Called credit default swaps, these derivatives worked like insurance policies on subprime residential mortgage-backed securities or on the CDOs that held them. Norma, acting as the insurer, would receive a regular premium payment, which it would pass on to its investors. The buyer of protection, which was initially Merrill Lynch, would receive payouts from Norma if the insured securities were hurt by losses. It is unclear whether Merrill retained the insurance, or resold it to other investors who were hedging their subprime exposure or betting on a meltdown.

Many investment banks favored CDOs that contained these credit-default swaps, because they didn’t require the purchase of securities, a process that typically took months. With credit-default swaps, a billion-dollar CDO could be assembled in weeks."

Still makes sense to me, absent overlooking the risk.

"But there was another, far more important, incentive: regulatory arbitrage.

Most notably, because super-senior debt carried the triple-A tag, banks were only required to post a wafer-thin sliver of capital against these assets - even though this debt has typically offered a spread of about 10 basis points over risk-free funds. Thus, banks such as UBS and Merrill have been cramming their books with tens of billions of super-senior debt - and then booking the spread as a seemingly never-ending source of easy profit. It is not just the CDO desks that have been playing this game; treasury departments have been playing along. So have many hedge funds, including those financed by . . . er . . . the major investment banks"

Now this is just plain risky. I actually mentioned this point about trying to invest in products with less capital/collateralization as being the motive for the investment, so that blaming the investment created to fill the need is silly. If not this product, it would have been something else give:

1) A taste for risk ( It's there, we just need to explain it )

2) A taste for lower capital/collateralization requirements

"It turns out that the super-senior tranches carried a lot more risk than many thought – Combing a bunch of CDOs composed of subprime securities into a new security didn’t make the underlying risk go away. Plus these instruments were illiquid and thus hard to sell.

I never thought that (formerly) highly-rated US securities would trade at the same price as Argentine bonds just after Argentina’s default. The ultimate recovery on those bonds may end up being higher as well.

Moreover, as Gillian Tett reports, the banks themselves weren’t only holding these securities to make the securitization process work. They also liked the extra yield that they could get on something that was rated triple-A."

By many, how about the buyers.

"It took a huge amount of sophistication – or at least computing power – to produce a lot of the securities that have caused so much trouble. But sophistication doesn’t explain why so many banks were holding something with risks that they didn’t fully understand to pick up a few basis points.

And it wasn't just pure regulatory arbitrage either . Unregulated institutions seem to have been doing much the same thing -- UBS's in-house hedge fund for example."

We've also talked about the search for the Holy Grail before: namely, how to use regulations to increase risk. Let's call this The Holy Grail Principle.

"My guess is that a bunch of unsophisticated reserve managers ended up doing a lot better over the past few years than a lot of sophisticated bank treasurers. They didn’t take on as much credit risk, so they didn’t get as much yield during the good times. But they also didn’t take on the risk of large losses for a few extra basis points"

Bingo!

"Of course, the bank treasurers who took on too much “holding complex securities that you cannot sell when you need to risk” generally avoided taking on much currency risk. Central banks generally avoided credit risk, but took on a ton of currency risk. The investment banks held the super-senior tranches no one else wanted; central banks have been holding the dollars no one else wants."

Banks: High credit risk, Low currency risk
Lenders of Last Resort: Low credit risk, High currency risk

Super-symmetry.

Here's my comment:

November 11th, 2008 at 11:23 am
1. Don the libertarian Democrat responds:
   

   “So there you have it: in the last resort, a key reason for these record-beating losses is not a failure of ultra-complex financial strategies or esoteric models; instead it arose from a humongous, misplaced bet on a carry trade that was so simple that even a first-year economics student (or Financial Times journalist) could understand it. It is a shocking failure of common sense and risk management. So the moral, in a sense, is also a simple one: if someone offers you seemingly free money, in seemingly infinite quantities, with a soothing new name, you really ought to smell a rat. Even - or especially - if you are in the position of running a supposedly sophisticated investment bank.”

   So, how come, when I keep saying this is on blogs, I keep getting the response that:
   1) It’s the complexity of the investments
   2) It’s too easy credit
   3) It’s too much incentive
   The answer is just as described. My only point is to try and understand how much implicit and explicit government guarantees went into these decisions, and to what extent the selling of these products was either fraud, negligence, or fiduciary incompetence.

   But thank you for linking to that post. I’ve finally found someone I basically agree with.