Lake Wobegon
Jul 16, 2012

"...where all the women are strong, all the men are good-looking, and all the children are above average."

If all the children are of the same ability, then they are all (weakly) above average.

Suppose that all the children are of ability 2 except Bob, who is only of ability 1. Then all the children except Bob are strictly above average.

Have you ever heard humorous observation that 90% of drivers think that they are better than average? The statement is potentially true, if there are a few really horrible drivers.

If you have gone to Starbucks in the last month, you have seen the red, white, and blue friendship bracelets with a piece of metal woven in hanging next to the register. The word "indivisible" is engraved on the metal. This is all part of the Starbucks "Create Jobs for USA" campaign. If you buy a bracelet for five bucks, the money is donated to the Starbucks Create Jobs for USA Fund. I unsuccessfully inspected a bracelet at some point to see if I could determine where it was made, since outside of this campaign, such a trinket would almost certainly be made in China--probably in some Bracelet City. I can't help but wonder if the jobs that are being created by the Starbucks campaign are mostly bracelet manufacturing jobs.

Reading the documentation, I found that donated money goes out as grants to lending institutions which focus on small business--community development organizations, for instance. In effect, the donations act as a subsidy to small business in the United States. This might create jobs, or it might not. Almost by definition, donations will make small business owners better off and some of this will probably filter through to their employees.

All this is fine as far as it goes, although I think this is a great example of donating to feel good rather than do good. Credit markets generally work well in the United States. Why should I subsidize small business owners rather than, say, give donation directly to people who have lost their jobs? My real worry is that this kind of donation crowds out those that do a lot more good in the world. (side note: a quick search of the literature shows a bunch of papers looking at crowd out of private donations by public subsidy, but I didn't see much about the crowd out of private donations by other private donations. Maybe there is room for an easy experiment?)

Prediction vs Theory
Feb 17, 2012

According to a recent nytimes article, Target has its ways of statistically uncovering your personal information:

"As [Target statistician] Pole’s computers crawled through the data, he was able to identify about 25 products that, when analyzed together, allowed him to assign each shopper a “pregnancy prediction” score. More important, he could also estimate her due date to within a small window, so Target could send coupons timed to very specific stages of her pregnancy."

Yesterday I blogged about the ethics of police using related methods to choose where to look for criminal activity. Target's goal is to predict the sort of products that customers will be interested in in the near future, and send relevant advertisements.

This sort of prediction makes me somewhat uncomfortable. Target doesn't have a micro model of pregnant mothers purchasing behavior. The company presumably doesn't know why pregnant mothers are making the product choices they do. What the company knows is that given present industry wide marketing practice, the state of the economy, and everything else, pregnant mothers buying habits are predictable. Now suppose that Target (or its competitor) changes marketing practices. Then pregnant mothers might buy a different set of products at Target, and the old statistical model may no longer hold.

If you don't understand "why?", it is hard to make accurate predictions. [HT Blattman]

Let me count the power laws:

1. My main work task was comparing a "networked" and "unnetworked" version of our search and learning trade model. The difference is that, as in the data, the networked version of the model gives rise to a power law tail in the client number distribution over firms. Here is the picture I made.

2. My office mate Felix and I were wondering whether the log-normal distribution had a fat tail or not. I looked up the definition of fat tail on wikipedia, and it turns out that fat tail means asymptotically proportional to a power law (A heavy tail, on the other hand, means not asymptotically bounded by any exponential distribution. You can be heavy but not fat, but all the fatties are heavy). A bit later, my other office mate Konstantin showed me why log-normal distribution does not have a fat tail. Log-normal is asymptotically proportional to x^-ln(x), which dies more quickly than a power law.

3. Robin Hanson's blog post yesterday was about power law distributions. I remember reading some time ago that it is hard to differentiate between true power law tails and other "almost" fat tail distributions like log-normal. A recent article in Science makes this point again, but Robin thinks that the evidence against power laws is weak.

When I was in college, I majored in political science. But if I were going through college today, I’d major in economics. It possesses a rigor that other fields in the social sciences don’t — and often greater relevance as well. That’s why economists are shaping national debates about everything from health care to poverty, while political scientists often seem increasingly theoretical and irrelevant.

Economists are successful imperialists of other disciplines because they have better tools. Educators know far more about schools, but economists have used rigorous statistical methods to answer basic questions: Does having a graduate degree make one a better teacher? (Probably not.) Is money better spent on smaller classes or on better teachers? (Probably better teachers.)

That is from the New York Times editorialist's column today.

Time to Buy!
Mar 23, 2011

The Mega Millions lottery jackpot is currently $304 million (annuity), the 12th largest jackpot in history (at least in nominal terms).  I consulted with a certain PSU faculty member very well versed in probability theory about buying a ticket.  He told me that when the Mega Millions jackpot gets to around 176 million, then the expected net return to buying a lottery ticket is zero.  Since the current jackpot is almost twice as high, the expected net winnings should be about the same as the price you pay for the ticket.  These are better odds than you will ever get at a casino!

The drawing is on Friday (3/25). I will see you in line at the gas station!

Edit: Corrected typo and added source.

Soliciting Donations
Mar 13, 2011

Here is a Google site for donating to Japanese Red Cross.

There goes my handshake chance. However he is going to be speaking at rec hall, which is right next to the econ building and is clearly visible from my office...I should at least be able to get a look at the guy.

It was a lot easier than I expected to match all the stuff from my website.  I guess everything is pretty much ready to go.