Information Processing

Just another weblog

Archive for July 2008

WSJ compensation survey

with 7 comments

The WSJ has posted some interesting compensation data which can be sorted by college and degree type. The data covers people less than 5 years out of school and more than 10 years out (“mid-career”), in all cases excluding those that earned advanced degrees — so that the outcomes are most sensitive to the undergraduate background of the respondents. Note you can click the column heading to sort by that variable (starting salary, 75 percentile mid-career salary, etc.).

There’s also an accompanying article. The author notes that schools (Ivies) at which a large fraction of students head into finance tend to have the highest starting and mid-career averages. Engineers have high starting salaries but not as much appreciation.

A social scientist can now regress the salaries on avg SAT and school / major to calculate the economic “value add” for a particular major or institution. (See UT Austin data here.) A significant confound is that people who *do* get advanced degrees would not appear in this data. (Almost half of all Caltech undergrads get PhDs and probably well over half get other advanced degrees.)

Some random results:

Caltech grads had the highest median starting salaries at $75k; by midcareer Dartmouth is number one with median compensation of $134k (MIT $126k, Caltech $123k).

Salaries by degree | starting | 10th-90th percentile midcareer range

Physics | $50k | $56k — $178k
EE | $60k | $69k — $168k
English | $38k | $33k — $136k
Economics | $50k | $50k — $210k
Philosophy | $40k | $35k — $168

• Survey respondents included two sets of U.S. bachelor’s degree graduates: Full-time workers with 5.5 years of experience or less and full-time employees with 10 or more years of experience.

• The survey excluded respondents who reported having advanced degrees, including M.B.A.s, M.D.s and J.D.s. Self-employed, project-based, and contract employees were also not included.

• Salary included annual cash compensation, including base salary or hourly wages, combined with commissions, bonuses, profit sharing and other forms of cash earnings.

Written by infoproc

July 31, 2008 at 3:18 pm

Still no mark to market

with 2 comments

Mortgage-backed assets with face value of $30.6B were just sold by Merrill for $6.7B.

1) obviously, this reflects huge losses and perhaps forward looking expectations of a very high default rate on the underlying mortgages (can’t tell unless I know what CDO tranches are in the pool they are selling).

2) but, amazingly, this 80% loss on the securities only gives a kind of upper bound on their value: Merrill had to provide 75% of the financing to the buyer! Even with an 80% haircut Merrill might not have found a real buyer for the assets at this price.

In the final days of the technology bubble we saw the same thing: vendor financing, in which sellers would loan customers the money to buy the product, just to make their quarterly numbers.

WSJ: …Much of what Merrill is sitting on is known as collateralized debt obligations on Merrill’s books. CDOs are securities backed by pools of mortgages or other assets, which have plummeted in value during the credit crunch. In all, Merrill has written down more than $46.7 billion in exposure to such mortgage-related assets since mid-2007.

The biggest single action Merrill took Monday was the sale of mortgage assets to an affiliate of Lone Star Funds with a face value $30.6 billion which it was carrying on its books at $11.1 billion as of the end of June. The sale, at a purchase price of just $6.7 billion, which represented just 22 cents on the dollar, reduced Merrill’s holding of such assets by more than half, from $19.9 billion to $8.8 billion.

Merrill said it would finance about 75% of the value of the deal, a similar action to other banks that loaned money to firms willing to take assets off its hands.

News that Merrill was able to sell these assets, even at a steep discount, bodes well for other firms on Wall Street like Lehman Brothers Holdings Inc., which are also sitting on hard to sell assets.

Written by infoproc

July 28, 2008 at 11:14 pm

Skype backdoor

with 2 comments

I knew it was too good to be true! As a for-profit company, Skype/EBay eventually had to cave in to spook pressure and allow for eavesdropping. In fact, the back door might have been in from the beginning.

Heise online: According to reports, there may be a back door built into Skype, which allows connections to be bugged. The company has declined to expressly deny the allegations. At a meeting with representatives of ISPs and the Austrian regulator on lawful interception of IP based services held on 25th June, high-ranking officials at the Austrian interior ministry revealed that it is not a problem for them to listen in on Skype conversations.

This has been confirmed to heise online by a number of the parties present at the meeting. Skype declined to give a detailed response to specific enquiries from heise online as to whether Skype contains a back door and whether specific clients allowing access to a system or a specific key for decrypting data streams exist. The response from the eBay subsidiary’s press spokesman was brief, “Skype does not comment on media speculation. Skype has no further comment at this time.” There have been rumours of the existence of a special listening device which Skype is reported to offer for sale to interested states.

Here’s what I wrote back in 2005:

…I just learned that Skype connections are encrypted using 256 bit AES, negotiated using 1024 bit RSA. This level of encryption is essentially unbreakable with current computing power. The Feds (with the possible exception of the NSA, and they would have to work very hard to break even a single session) have no chance of eavesdropping on any Skype conversation.

It is true that Skype is closed-source, so it isn’t easy to verify that the crypto implementation doesn’t have any holes or backdoors. However, given the number of users and the negative consequences for the company of any privacy issues, I suspect that it works as advertised.

Well, although you are probably safe from your neighbors or local network admin, the Feds apparently don’t have any problems listening in on your Skype calls.

Written by infoproc

July 26, 2008 at 7:12 pm

Asset-backed oversold? Paulson ready to get back in!

with 2 comments

John Paulson made $15 billion for his fund ($3.7 billion for himself!) betting against subprime securities last year. Now Bloomberg reports that he’s ready to get back in on the upside!

I’m as dismayed as anyone else about taxpayer dollars going to bail out mortgage holders, commercial banks and the GSE’s (Fannie, Freddie). I complained about Fannie back in 2004 when Franklin Raines was still CEO and the story first got out about their smoothing of earnings and use of derivatives accounting.

But, on the other hand, maybe Uncle Sam can actually generate positive return by buying oversold securities! Rumor says that current prices of subprime mortgage backed securities suggest implied default rates which are unrealistically high (e.g., like 50% !). Barring a complete economic meltdown these assets are underpriced and will generate a handsome return for an intelligent investor willing to take some risk. Is Uncle Sam smart? Let’s hope that the other Paulson negotiates some upside for us taxpayers in any bailout organized by the Treasury.

Bloomberg: John Paulson, the money manager whose wagers against the U.S. housing market helped him earn an estimated $3.7 billion last year, is now seeking to profit from Wall Street’s search for capital to offset mortgage writedowns.

Paulson plans to open a hedge fund by December that will provide capital to the world’s biggest banks and brokers as they add to the $345 billion they’ve raised in the past year, according to two people with knowledge of the matter. His Paulson & Co., which oversees $33 billion, hasn’t set a size target for the fund, said the people, who declined to be identified because the plans aren’t final.

The New York-based firm’s credit funds rose as much as sixfold last year, helped by bets that rising defaults on subprime home loans would pummel the value of mortgage-backed securities. The meltdown has forced the world’s biggest banks and securities firms to take $467 billion in asset write-offs and credit losses and led to the collapse of Bear Stearns Cos.

“Investors who are able to make money in a declining market and then rapidly turn around and profit from a rising market is highly unusual,” said Thomas Whelan, president of Greenwich, Connecticut-based Greenwich Alternative Investments, which advises clients on investing in hedge funds.

Paulson declined to comment. His 2007 earnings made him the highest-paid hedge-fund manager, according to Institutional Investor’s Alpha magazine.

Written by infoproc

July 23, 2008 at 9:22 pm

The joy of gender imbalances on campus

with 3 comments

The Chronicle of Higher Education reports on the social effects of gender imbalances. As we know, teenage girls are more likely than boys to have their acts together, hence make up a larger and larger percentage of those who attend college. I see this all the time in my intro classes — a majority of the most organized students are female, and a majority of the least organized are male. Many universities will have to use affirmative action for male applicants in order to preserve 50-50 gender ratios.

Math puzzle: (optional for readers who have trouble with distributions) If most of the least able students are male, and yet men and women perform equally on average (actually this may not be true anymore, but see PISA data :-), what does that imply about the most able students on campus?

Chronicle: American colleges are undergoing a striking gender shift. In 2015 the average college graduating class will be 60-percent female, according to the U.S. Education Department. Some colleges have already reached or passed that threshold, which allows anecdotal insights into how those imbalances affect the pickup culture. What can be seen so far is not encouraging: Stark gender imbalances appear to act as an accelerant on the hookup culture.

…In 2006 I visited James Madison University, a public university with 17,000 students. At the time, women made up 61 percent of the campus population.

…A senior added: “The guys see that there are a lot more girls, and they’re not interested in having a relationship longer than the next girl to come along. Men know how to take advantage of that competition. They’ll set things up at parties to get girls to do stuff, such as having a slip and slide contest,” in which girls strip to their underwear and get wet sliding through water on a plastic sheet.

As a result of the rising gender imbalances, the university has become “female centric.” But while women may run the clubs, dominate in classes, and generally define the character of the university, the law of supply and demand rules the social scene. That’s why the women are both competitive in seeking men and submissive in lowering their standards.

Men at the university don’t dispute what the women say. “Since there’s such an overwhelming number of girls, they have such competition between each other to get a guy,” a male junior admitted. “The guys here aren’t stupid. They’re plenty aware of that and know that girls have to get into a fight over them, instead of what’s normal with guys courting girls.”

I wonder what a FaceBook search on the keywords “slip and slide” brings up?

Written by infoproc

July 22, 2008 at 6:09 pm

Bay area housing market begins to crack

with 8 comments

It’s about time! See Calcuated Risk for further discussion. (Note this data covers some regions which are quite far from the bay itself.)

Standard bubble wisdom: they last longer than you think is possible, then pop faster than anyone expects.

I’ve been calling the bay area bubble since as far back as 2004 / 2005 (actually, from before I started this blog!). I expect the last bubble market to pop will be Manhattan, but pop it will.

It must hurt to think that a house you purchased a year ago might be worth a third less today, with still further to go.

SF Chronicle: Across the nine counties, the median price paid for resale homes, new homes and condos in June plunged 27.1 percent from a year ago to $485,000, dipping below the half-million-dollar mark for the first time in four years, DataQuick Information Systems of La Jolla (San Diego County) reported Thursday.

Among resold homes, bank-repossessed foreclosures – which usually are discounted – accounted for 28.7 percent of all existing-home sales, up from just 3.5 percent in June 2007. Solano County, with foreclosures at 57.7 percent of all resales, had the highest percentage; San Francisco, at 3 percent, had the lowest.

Affluent areas such as Marin County and San Francisco, which until now had resisted most price erosion, saw existing single-family home median prices fall by about 11 percent. Including new homes and condos, the Marin County and San Francisco medians fell about 12 percent to $846,000 and $726,750, respectively.

“This is pretty grim; double digits across the board,” said Christopher Thornberg, principal at Los Angeles’ Beacon Economics. “It was eminently predictable if you had a realistic view of the world. I heard a lot of people say the Bay Area was never going to see prices fall, San Francisco was untouchable; in San Mateo, it was impossible; San Jose, not with all the tech money, blah, blah, blah. But prices at the peak relative to people’s incomes never made any sense.”

Note the 27% figure for change in median price is a bit tricky to interpret: it’s alway possible that the composition of units sold has changed, with a lot of owners of high end homes sitting on them, refusing to accept the current market price. In that case the average price decline would be less than 27%. It is typical in a real estate crash for sellers to remain in denial for some time, during which the big decline is in number of transactions rather than actual price levels. It’s only after the sellers have capitulated that the big price drops occur. In light of that, it’s rather ominous that only 7,178 new and resale homes changed hands in June – the lowest June figure since 1993 🙂

Written by infoproc

July 20, 2008 at 9:38 pm

Wonderlic fun

with 12 comments

The Wonderlic is a simple intelligence test widely administered to job candidates, most famously by the NFL in its annual draft. Try this sample test and report your score in the comments 🙂

Note: actual test is 50 problems over 12 minutes, so ESPN is probably incorrect to give 5 minutes for 15 questions. Give yourself 3.6 minutes and multiply your score by 50/15 = 3.33 to get a Wonderlic score, W. Wikipedia estimates: (very roughly) IQ = 2*W + 60

ESPN: …Each year, about 2.5 million job applicants, in every line of work, take the Wonderlic. The average NFL combiner scores about the same as the average applicant for any other job, a 21. A 20 indicates the test-taker has an IQ of 100, which is average.

Some people disagree with the whole idea of IQ testing because they believe the tests are culturally biased and inaccurate. But Charlie Wonderlic doesn’t make grand claims for the score derived from his test. “What the score does is help match training methods with a player’s ability,” he says. “It could be a playbook — what is the best way to teach a player a play? On the field, the higher the IQ, the greater the ability to understand and handle contingencies and make sound decisions on the fly.”

Yes, we know painfully well how “discredited” IQ tests are. But, evidently, many employers and virtually all universities, not to mention NFL franchises, think that your IQ score has some predictive validity…

The diagram below illustrates how scores vary by position (created by Ben Fry). Note even the “brainy” O-lineman fall short of the average score for college graduates, 28. The overall US average is 21, implying 1 SD is something like 7 Wonderlic points, so the spread between halfback and left tackle is about 1.5 SD.

From Every year, around the time of the NFL draft, there is a sudden surge of interest in the Wonderlic Personnel Test (WPT). The Dallas Cowboys first used the WPT for the selection of football players in the 1960’s. Today, it is one of the standard measures used by the combine and NFL teams when considering draft picks.

Although Wonderlic does not score the WPT for the combine, nor does it receive score reports, there are several sources that publish all of the draft statistics, including Wonderlic scores.

The average WPT score for player positions in the NFL are the same type of scores used by employers when hiring people for specific jobs. In his book Paul Zimmerman’s “The New Thinking man’s Guide to Pro Football,” Paul Zimmerman has published what he believes to be the average scores for NFL players by position.

NFL Position / Wonderlic Score / Job Title

Offensive Tackle / 26 / Marketing Executive
Center / 25 / Claims Examiner
Quarterback / 24 / Computer Operator
Tight End / 22 / Police Officer
Safety / 19 / Butcher
Middle Linebacker / 19 / Hospital Orderly
Cornerback / 18 / Machine Operator
Wide Receiver / 17 / Laboratory Assistant
Fullback / 17 / Dock Hand
Halfback / 16 / Material Handler

Of course, the average score by player position is a group statistic. An employer would use an individual player’s score to determine potential job success. As a point of reference, the average score across the United States is 21, while the average for college graduates is about 28.

Written by infoproc

July 18, 2008 at 8:35 pm