The Rise of the (Financial) Machines

BartlebyScrivener writes "A New York Times Op-Ed quoting Freeman and George Dyson wonders if Wall Street geeks and 'quants' outsmarted themselves with computer algorithms to create the current financial debacle: 'Somehow the genius quants — the best and brightest geeks Wall Street firms could buy — fed $1 trillion in subprime mortgage debt into their supercomputers, added some derivatives, massaged the arrangements with computer algorithms and — poof! — created $62 trillion in imaginary wealth. It's not much of a stretch to imagine that all of that imaginary wealth is locked up somewhere inside the computers, and that we humans, led by the silverback males of the financial world, Ben Bernanke and Henry Paulson, are frantically beseeching the monolith for answers.'" The quoted essay from George Dyson is available at Edge.

Comments

Re:VaR anybody?

[dnwq]

An interesting article from Newsweek, The Monster That Ate Wall Street :

But what if JPMorgan could create a device that would protect it if those loans defaulted, and free up that capital?

What the bankers hit on [in the 1990s] was a sort of insurance policy: a third party would assume the risk of the debt going sour, and in exchange would receive regular payments from the bank, similar to insurance premiums. JPMorgan would then get to remove the risk from its books and free up the reserves. The scheme was called a "credit default swap," and it was a twist on something bankers had been doing for a while to hedge against fluctuations in interest rates and commodity prices... [JPMorgan] built up a "swaps" desk in the mid-'90s and hired young math and science grads from schools like MIT and Cambridge to create a market for the complex instruments. Within a few years, the credit default swap (CDS) became the hot financial instrument, the safest way to parse out risk while maintaining a steady return.

...

Before long, credit default swaps were being used to encourage investors to buy into risky emerging markets such as Latin America and Russia by insuring the debt of developing countries. Later, after corporate blowouts like Enron and WorldCom, it became clear there was a big need for protection against company implosions, and credit default swaps proved just the tool.

...

AIG's fatal flaw appears to have been applying traditional insurance methods to the CDS market. There is no correlation between traditional insurance events; if your neighbor gets into a car wreck, it doesn't necessarily increase your risk of getting into one. But with bonds, it's a different story: when one defaults, it starts a chain reaction that increases the risk of others going bust. Investors get skittish, worrying that the issues plaguing one big player will affect another. So they start to bail, the markets freak out and lenders pull back credit.

The problem was exacerbated by the fact that so many institutions were tethered to one another through these deals. For example, Lehman Brothers had itself made more than $700 billion worth of swaps, and many of them were backed by AIG.

Ever wonder where 'money' comes from?

[mrbill1234]

Check out this 47 minute video for a very easy to understand and clear explanation.

http://video.google.com/videoplay?docid=-9050474362583451279

Unless you've been through university on some Economics degree - you were probably unaware of this.

Re:VaR anybody?

[Anonymous Coward]

I work with this as well, and I can tell you that the way VaR is run is somewhat flawed. What they basically do is to run various scenarios, changing different factors and calculate what the impact on the investor's position is.

Now, there are several ways you can do this. One is to run a massive monte-carlo simulation on all the input data, something that all financial software supports. There is a problem with this though, and that is that it requires some massive CPU power. Even a smaller bank would require many hundreds if not thousands of CPU's chugging away for 12 hours or more to come up with the proper numbers.

The solution they've come up with is called "historical VaR". What they do is to use the historical market data for the last year instead of random data which would be used otherwise.

The obvious problem with historical VaR is of course that it doesn't take unexpected market movements into account. If a drop such as the ones that we have seen recently haven't happened in the last year, VaR won't take such a scenario into account.

So, in short, VaR only really works in "normal" market conditions. It doesn't take extreme movements into account.

Re:The million dollar question is...

[nomadic]

Only most of the loans that were rolled into the mortgage-backed securities were made by institutions that were not governed by the CRA.

Nice try though.

Re:More Leverage....

[martyb]

As such, though the leverage ratio was officially 12.5, somebody who held nothing but mortgages could be levered up 35:1. And if you owned some bank issues, you could get nearly infinite.

But I'm wondering... what makes you think that these limits were going to be further increased?

You obviously know more about these details than I. What I was working from was what I heard in the second program: Another Frightening Show About the Economy. You can download it for free at the moment from: here.

Here's the summary for part 2:

Act Two. Out of the Hedges and Into the Woods.

One more confusing financial product that's bringing down the global economy. And one of way to think about this product is this: If bad mortgages got the financial system sick, this next thing you're about to hear about, helped spread the sickness into an epidemic. These are "credit default swaps." Alex explains. (19 minutes)

The segment was so well done, it's hard to summarize, but I'll do my best. In essence, "Credit Default Swaps" (CDS) were presented as a form of insurance that a lender could by so as to minimize the risk that a loan would default. So far, so good. Then, somebody realized they could by a CDS for a loan they did not even own. So what? It got out of hand when someone realized they could make money by buying a CDS for something that was questionable. How so? Like buying life insurance. Like buying life insurance on somebody ELSE. Who is old and feeble. I have an opportunity to buy a policy for, say, $1M on this person. And YOU have an opportunity to buy the same kind of policy from your agent. The sicker the person, the more of an incentive there is to buy in.

Now, replace "old and feeble person" with "Lehman Brothers". And for "insurance company", try "AIG". And multiply just You and Me with hundreds or thousands of CDSs in play. With many, many other companies. As I understand it, there was no regulatory limit on how many CDSs could be purchased on the exact same debt issue. And, because there was no mandatory reporting or the like on what CDSs were out there, we really don't know just how many of these are out there. With all the repackaging of these as securities, sliced and diced and sold as yet more instruments, we don't know just how bad the situation is.

Again, I cannot do the show justice. Listen to the podcast. I'd love to hear your take on it once you have done so.

Re:Anti-math/science witch hunt

[anaesthetica]

HOW did the model not include some check

That's just the point. They didn't do the sociology of the people taking the NINA variable rate mortgages. A lot of them were speculators who were buying properties with no money down and then flipping them. This had two effects: driving the price up because they could; and, making it looks like NINA variable rate mortgages were getting paid off with only slightly higher risk than normal loans. Once the house-flippers got out and interest rates went up, the people with no assets and no income left holding the variable rate mortgages couldn't pay and the house of cards collapsed. The model was based on speculators flipping houses, not on real owners.

If the quant people had done the sociology instead of just running the numbers, they would know who was buying and why the data appeared the way it did.