Slashdot Mirror
Greenspan Tells Congress Bad Data Hurt Wall Street
CWmike writes "Former Reserve Bank chairman Alan Greenspan has long praised technology as a tool to limit risks in financial markets. In 2005, he said better risk scoring by high-performance computing made it possible for lenders to extend credit to subprime borrowers. But today Greenspan told Congress that the data fed into financial systems was often a case of garbage in, garbage out. Christopher Cox, chairman of the Securities and Exchange Commission, told the committee that bad code led the credit rating agencies to give AAA ratings to mortgage-backed securities that didn't deserve them. Explaining in his testimony what failed, Cox noted a 2004 decision to rely on the computer models for assessing risks — a decision that essentially outsourced regulatory duties to Wall Street firms themselves."
If these people did not know what was going on, they are not professionals, they are just a schmuck who is being paid too much. To say that the computer models did not anticipate their stupidity is just denial.
If all else fails, blame your tools.
The data being flawed is very different than the code being flawed. In fact, what Greenspan is talking about has almost no connection to what Cox is talking about, and there's no real reason to put them both in the same article. Starting with bad data will abundantly suffice to explain the meltdown before any problems with the algorithms used have to be assumed.
Most of the bias that did the real damage is political. For example, the most recent figures on the economy show that in the months before the mortgage crash began, 68% of all spending was driven by individual consumers buying retail. If the last tax rebate had been aimed at 68% of the total going back to individual consumers, or the '700 billion bailout' had put 68% of the 200+ Billion actually committed so far into reducing the impact to non-institutional borrowers, those would be appropriately neutral positions - but in the current climate, those would both be classified as terribly liberal.
But that figure wasn't trumpeted about until after the bailout was passed. The same goes for the corrected inflation rates, which are still not accurate but are a bit better, and which again weren't corrected in releases to the general public until after the bailout was final.
Who is John Cabal?
Those bankster knew exactly what will eventually happen. But their modus operandi is to privatize profits and socialize losses. It's as simple as that. So why would they bother?
Bottom line, this had nothing to do with bad data. It was Greenspan's blindness to the consequences of easy monetary policy would have that caused much of the problems today.
Absolutely. Wait, rollercoaster interest rates are a bad idea? Really? And it took a genius to figure this out?
It's so easy to understand. Low credit and the push for home ownership at any cost led to insane price increases and speculation that it wasn't hard to see had to come to a crash stop. I had this figured out as of 2004 when I talked to a realtor who told me I needed to buy NOW with nothing down and use the guaranteed 2%/month price increase to refinance in a year. I can recognize a bubble when I see it.
That's why it pisses me off when Greenspan points the fingers elsewhere. He's the one who set the rates. He's the one who jacked them up, then down, waiting too long and overcorrecting to account for it. And he refuses to take the blame.
The funny thing is, this isn't the first time things have gotten sideways thanks to overspeculation. During the (mercifully) brief meltdown in 1998 due to the currency markets, he basically told the banks to do what they do, the government will help out if things go bad. The overcorrection to that mini-crisis and the post-9/11 slowdown sowed the seeds for what we have now. Gee, thanks Alan.
So now he blames bad data. Really, Alan, you're surprised that people selling certain securities said things about them that was overly rosy? Give me a break. At some point, you have to have some damned sense, and actually look at the securities without the computer models. When things defy common sense to that degree, something's wrong.
The funny thing is, it seems every crisis comes about because risk diversification models fail. Happened in 1929, happened in 1998, happened now. Investing houses have this theory that a lot of big risks can be less risky in totality, because the risks aren't correlated. Problem is, when the shit hits the fan, a lot of things become correlated that didn't use to be. Partly it's because everything's sitting on top of the same increasingly global economy. Part of it is that funds that are overly leveraged have to sell whatever they have to meet margin calls. The people who create the models study the risk correlation and assume things based on it that simply aren't valid in the real world. The book "When Genius Failed" has a good case study on this, where an investment house run by brilliant guys including Nobel Prize winners crashed and burned because they didn't understand that common sense trumps mathematical models.
To disclose, I actually see great value in statistical predictive models - indeed, that's what I do for a living. I design and implement mathematical models. But because of that, I also know what mathematical models can't do. Too much hubris by too many people, and we all suffer.
The problem with some old people is that they don't realize how much they don't know.
The problem with most people is that they don't realize how much they don't know.
From what I understand, they were giving loans to people who had no collateral and no income. If your computer model says that loan is a safe loan, then you have a bug.
Yeah, the last thing you want is someone who changes their mind in the light of new information.
I think the world got dumber the day the term "flip-flopper" began being used in public discourse.
Boffoonery - downloadable Comedy Benefit for Bletchley Park
Computers and programs do what they are built to do, exactly like they are programmed by programmers. and programmers code what they are TOLD to program.
this senile old bastard is trying to drop the blame ball on someone else than himself. he was the person who ushered the 'let everyone run around lawless' era in finance. he was praising it and saying that 'free market' was this and that. now he comes up saying he is 'shocked' to see the market not regulating ITSELF.
i have news for you, bastard, what you call market is comprised of PEOPLE, and its a social activity. just like life doesnt 'regulate' itself so that you still need laws and justice system, the social activity you call 'market' also is comprised of people and full of opportunists, schemers, bastards, exploiters, criminals and crooks. if you just let everything be, IT BREAKS. and IT DID.
any person with only a few decade of life experience under his/her belt would be able to realize this.
but you and your fellows in the church of holistic economists were SO zealots in your belief that, you were unable to realize this simple fact of social existence despite your 5-6 decades on the face of this world.
shame on you old man. shame on you for preparing the grounds for breakdown of ENTIRE global economy with your zealotry and foolishness, and your attempt to blame others for it.
blame the data !! after all, noone can do anything against it right ?! its not live, its nobody, and even if you hate its guts, you wont be able to remove it from business, so problem solved.
Read radical news here
Citation needed.
I don't think models were anywhere near that simple. The closest you could get to that is if you fed home appreciation data from a time period where house prices mostly went up, and had no examples of periods when they went significantly down. That's a plausible failure mode for many of these models (and it happens all the time with financial models, ugh, and the financiers don't seem to learn), but the models would have made different predictions with different data sets.
There's another assumption that people made that led to the problems with the ratings: the assumption that housing and mortgages from different parts of the country would have uncorrelated performance, so that packaging them all together would diversify risk away. The short catchy phrase for that was "all real estate is local": the assumption was that house prices can go down in some parts of the country at any given time, but that it was unlikely that they would go down in all of the country all at once. You can see how that one turned out, of course.
That one, again, turns out to be a recurring problem with financial modeling:
The financial model failures we're seeing now are remarkably similar to the crisis that led to the failure of LTCM 10 years ago. The industry doesn't seem to learn, which is a big problem.
More generally, there's a bigger problem here (and I'm paraphrasing Buffett in the following): it's not that the mathematical models of risk aren't valuable, it's that, by putting very precise-looking numbers to aggregates of thousands of highly uncertain estimates of future risks, they make it look like risk has been tamed. If you have a model that tells you that the current risk of your portfolio is, say, 15.72%, and you mechanically decide how to allocate your capital using a formula that doesn't build in a generous margin of safety against mistakes in that number, you're going to get burned by problems like this.
Are you adequate?
That's the whole point of TFA - that the repackaged debt was given AAA ratings despite being obviously toxic.
Of course the decision to force institutions to make these loans was stupid, but it's the subsequent repackaging and reselling of the debt, disguising its lack of real value, that was the cause of this crisis, and the methods used to create that false valuation that are the subject of Greenspan's criticism.
One swallow does not a fellatrix make