How Engineers Destroyed the Economy

[Savonarola]

Sorry guys   

http://www.spectrum.ieee.org/at-work/tech-careers/the-rise-and-fall-of-the-quants

The Rise and Fall of the Quants

When the U.S. financial system melted down, fingers were quickly pointed at the "quants"—the physicists, mathematicians, and engineers who had devised the computer programs, statistical tools, and financial instruments that were supposed to help investors manage risks. Critics said that it was the flawed assumptions of those financial models that brought banking to the brink of Armageddon.

You might guess that Wall Street is now shunning physicists, mathematicians, and engineers, but you'd be wrong. Talented people with quantitative backgrounds are more welcome than ever, says Petter Kolm, deputy director of New York University's master's program for mathematics in finance and formerly a quant for Goldman Sachs. Before the financial mess started in the fall of 2007, he says, investment banks typically hired fresh MBA grads as entry-level analysts. "Many investment banks I talk to today say they want to replace that portion of people with MBAs with people who have quantitative analysis skills, such as engineering and math."

The culture of Wall Street is already relying less on traditions and personal connections and more on complex technologies. Juhua Zhu, who has a Ph.D. in electrical engineering from Princeton and is now a vice president at Morgan Stanley, saw the change coming shortly after joining the firm four years ago. She says it was soon clear that "traditional trader jobs would vanish as many transactions would be transferred to computers."

Those changes are carving out a need for more math whizzes and data crunchers. "Quantitative jobs demand research talent—people who can read any text in a technical field and reach a high level of expertise in a short amount of time," says Alp Atici, a Columbia University math Ph.D. who works as a quantitative researcher at hedge fund Citadel Investment Group. "People with Ph.D.'s in science and pure math are usually accustomed to much harder and deeper research texts."

One of the first quants was Robert Merton, who started out as an applied mathematician at Caltech before switching to economics at MIT. In 1969 he tried to work out the pricing of stock options with the help of stochastic calculus—a branch of mathematics used to model random systems such as Brownian motion, the movement of particles in liquid. A high point for quants came when Merton and Myron Scholes won the 1997 Nobel Prize in Economics for their option-pricing model. (When it comes to low points, Merton, the quant trendsetter in so many ways, was ahead of his time again. He and Scholes were members of hedge fund Long-Term Capital Management's board of directors when that company lost more than US $4 billion in 1998; the Federal Reserve, fearing a liquidity crisis, put together a restructuring plan that presaged its aggressive liquidity interventions in 2008.)

The trickle of physicists and mathematicians eschewing low-paying academic jobs in favor of Wall Street bonuses turned into a flood in the Reagan years. The 1980s fads for junk bonds and leveraged buyouts came and went, but the quants stayed. Their direct role in today's recession started in the booming 1990s housing market as they helped banks package mortgages, credit cards, and other credit assets, slice up the packages, and sell them as instruments known as asset-backed securities. That took risky assets off companies' balance sheets, freed up capital, and let the companies borrow more money. Quants are credited with creating models that helped investors understand, manage, and price the risk associated with these securities.

Asset-backed securities are a great innovation if used sparingly. But this did not happen. Emanuel Derman, industrial engineering and operations research professor at Columbia and author of the memoir My Life as a Quant: Reflections on Physics and Finance, says banks "got too big for their boots and borrowed too much money." The quants' mathematical machinations didn't so much dilute risk as hide it. And then came the breaking point, one that quantitative models did not take into account: record numbers of subprime borrowers defaulting on their mortgages.

So was this a lack of foresight on the quants' part? The simple truth is, there is no one right model, Derman says. Inputs to financial models are related to how people will behave in the future. But patterns of behavior change over time.

"If you're designing a global positioning system, the distance from here to Waterloo doesn't change," he says. "Financial engineering isn't based on financial science. It's scientific methods applied to human variables."

But why didn't quants raise the alarm about risks associated with their innovations? NYU's Kolm says that business decision makers, not quants, were calling the shots and did not always care about the risks. At a major investment bank, he says, risk managers who tried to warn about risks were unpopular: "Either you played along or you left because you were the bad guy at the party." It's also becoming clear that mortgage fraud played a big role in distorting the data that the models used to predict foreclosures—a case of garbage in, garbage out.

Despite the role that physicists and engineers played in the economic crisis, the relationship doesn't seem to have soured. Quants aren't being recruited less or fired more—though there are fewer jobs overall as some companies, like Lehman Brothers, disappear, and others, like Bank of America and Merrill Lynch, merge. Beverly Principal, assistant director of employment services at Stanford, says that finance companies are still coming to campus, while a large number of students are still interested in financial careers. Career service representatives at other schools echo this sentiment.

Companies have started to be selective, and people with advanced degrees have the upper hand. Credit Suisse Securities has spent the past year trying to attract people with doctorates in physics, math, and engineering from top schools, says Ilias Tagkopoulos, who got his Ph.D. in EE from Princeton last summer. A trifecta of skills—programming, mathematical proficiency, and an ability to communicate—led to his job at Credit Suisse as a relationship manager. Most members of his team are also Ph.D.'s from top schools, he says.

The lack of jobs, though, has started to make itself felt. Ming Zhong, a portfolio manager at Lazard Asset Management, an investment bank in New York City, says that when he graduated from Columbia with a master's in EE in 2004, 70 percent of the jobs available through the campus-recruiting program were finance related. Now, he says, graduates are having a hard time finding internships, even unpaid ones. "One year ago, I'd say keep trying, always have hope. Six months ago, I suggested students look around. In the spring, I told them, 'You should think about changing careers.' "

But things won't always be so bleak. When the economic pendulum eventually swings back, predicts Morgan Stanley's Zhu, "the whole pie will have shrunk, but the portion of jobs for people with a technical background will have continued to grow."

When I was studying linear algebra in my graduate courses the teacher said that what we were learning actually had greater use in the social studies than engineering or computer science; since those fields deal with much larger and less predictable systems than networks, circuitry or computers.   Yet everyone in the class was an engineer, mathematician or computer scientist.  I thought about that when I read this article; finance is an even more obvious application for advanced mathematics.

[DGuller]

True engineers have a vague idea about how much play there is in their products, and built some safety in.  Unfortunately, in a race to the bottom, conservatism is the first virtue to go.

[The Minsky Moment]

Not so. To wit:

But why didn't quants raise the alarm about risks associated with their innovations? NYU's Kolm says that business decision makers, not quants, were calling the shots and did not always care about the risks

It's like when an engineer says - sure we can rejigger the doohickie to carry twice the load, but it will increase the risk of catastrophic failure.  If the honcho says to do it anyway, the engineer is not to blame.

[garbon]

rejigger the doohickie to carry twice the load

[DGuller]

Not so. To wit:

But why didn't quants raise the alarm about risks associated with their innovations? NYU's Kolm says that business decision makers, not quants, were calling the shots and did not always care about the risks

It's like when an engineer says - sure we can rejigger the doohickie to carry twice the load, but it will increase the risk of catastrophic failure.  If the honcho says to do it anyway, the engineer is not to blame.

I guess that depends on what quants told the upper management about the risks.  I wouldn't be surprised if the quants got way over-confident in their models, but then I wouldn't be surprised if the upper management just did not want to hear about the risks.  The article doesn't make it perfectly clear what was the more common occurrence.

[Admiral Yi]

It's like when an engineer says - sure we can rejigger the doohickie to carry twice the load, but it will increase the risk of catastrophic failure.  If the honcho says to do it anyway, the engineer is not to blame.

I see no evidence that the engineers said that.

[Baron von Schtinkenbutt]

I see no evidence that the engineers said that.

Its not clear what they said at the lower levels.  The upper levels of management over the engineers, be they engineers themselves or not, may well have been filtering the results coming up from the lower levels based on what they thought their bosses wanted to hear.  Its something that happens all too frequently in pure engineering companies, too.

[Neil]

True engineers have a vague idea about how much play there is in their products, and built some safety in.  Unfortunately, in a race to the bottom, conservatism is the first virtue to go.

For one thing, that's not the engineer's call to make.

For another, that's not the way they do it anymore.  Overdesigning things was something they did before computer modelling allowed them to figure out exactly how strong to make something.

[citizen k]

Didn't the engineers also get blindsided by the Asia meltdown in late 90's?

[Admiral Yi]

Its not clear what they said at the lower levels.  The upper levels of management over the engineers, be they engineers themselves or not, may well have been filtering the results coming up from the lower levels based on what they thought their bosses wanted to hear.  Its something that happens all too frequently in pure engineering companies, too.

I read an article a while back in Time, talking about some dude with a Chinese name who came up with a model for pricing CDOs that didn't even use default rate as an input.  Can't remember where he worked.

[Warspite]

I see no evidence that the engineers said that.

Its not clear what they said at the lower levels.  The upper levels of management over the engineers, be they engineers themselves or not, may well have been filtering the results coming up from the lower levels based on what they thought their bosses wanted to hear.  Its something that happens all too frequently in pure engineering companies, too.

IT could surely be a bit of both - management contributing to an organisational culture that did not encourage appropriate respect for or modelling of risk?

[The Minsky Moment]

Its not clear what they said at the lower levels.  The upper levels of management over the engineers, be they engineers themselves or not, may well have been filtering the results coming up from the lower levels based on what they thought their bosses wanted to hear.  Its something that happens all too frequently in pure engineering companies, too.

I read an article a while back in Time, talking about some dude with a Chinese name who came up with a model for pricing CDOs that didn't even use default rate as an input.  Can't remember where he worked.

I assume you are referring to David Li's "Gaussian copula" model, which is the most famous (or infamous).  He worked at both CIBC and Citibank.  Li did warn about the limitations of the model, but it was misused and abused by those who should have known better.

[Admiral Yi]

I assume you are referring to David Li's "Gaussian copula" model, which is the most famous (or infamous).  He worked at both CIBC and Citibank.  Li did warn about the limitations of the model, but it was misused and abused by those who should have known better.

What did he say about the limitations of the model?

[ulmont]

I assume you are referring to David Li's "Gaussian copula" model, which is the most famous (or infamous).  He worked at both CIBC and Citibank.  Li did warn about the limitations of the model, but it was misused and abused by those who should have known better.

What did he say about the limitations of the model?

"Very few people understand the essence of the model"

"The most dangerous part is when people believe everything coming out of it."

"it's not the perfect model."

[viper37]

When I was studying linear algebra in my graduate courses the teacher said that what we were learning actually had greater use in the social studies than engineering or computer science; since those fields deal with much larger and less predictable systems than networks, circuitry or computers.   Yet everyone in the class was an engineer, mathematician or computer scientist.  I thought about that when I read this article; finance is an even more obvious application for advanced mathematics.

when we study calculus, we are told that integrals have more use for the engineers and physicists than us
I'd say we're even