Specialization is for insects

Impossible is Nothing

Great story on graduate assistant at Hawaii who suffers from cerebral palsy in the NY Times.  Yet another reason to root for them against Georgia.

NY Times front page

From the front page of nytimes.com:

Mike Huckabee, pictured campaigning in Iowa, used the volatile situation in Pakistan to argue for building a fence on the U.S. border with Mexico.

That's not the Liberal Media misquoting him.  Read the full article:

On Thursday night he told reporters in Orlando, Fla.: “We ought to have an immediate, very clear monitoring of our borders and particularly to make sure if there’s any unusual activity of Pakistanis coming into the country.”

On Friday, in Pella, Iowa, he expanded on those remarks.

“When I say single them out I am making the observation that we have more Pakistani illegals coming across our border than all other nationalities except those immediately south of the border,” he told reporters in Pella. “And in light of what is happening in Pakistan it ought to give us pause as to why are so many illegals coming across these borders.”

In fact, far more illegal immigrants come from the Philippines, Korea, China and Vietnam.

The craft and rhetoric of economics paper-writing

I read Chay and Greenstone (JPE 2005) very carefully yesterday.  It is a great paper.  Since Levitt lists it as a starred reading in the introduction to his "how to write a paper" course syllabus, he must consider it a stellar example of empirical paper-writing.  And I definitely agree.  It's very clear and easy-to-follow.  The paper is written in such a way that my mother (who has a good statistics background but a less-than-graduate-level economics background) could get through almost all of it.  It is clear and convincing, and at the end of the paper you feel like the authors have made an important contribution to knowledge.

What I noticed in my close reading is that the paper crafts some very nifty arguments when things don't really "work":

Exhibit A: Region-specific time-trends (that's what the Region FE in First Differences are doing) are the final column in the main results table (Table 5).  But the analysis is at the county-level; what about state-specific time-trends?  Their omission as a robustness test leads me to assume that they must not have worked out very well, otherwise they would almost certainly have been included.  There are many environmental regulations at the state level that might have changed during this time period and also contributed to the observed changes in housing values.

Exhibit B: The results aren't robust to small changes in which year nonattainment status is used an instrumental variable.  The authors construct a fairly convincing argument for why they use mid-decade nonattainment status, which might not have been the instrument one would have chosen a priori.

Exhibit C: The RD doesn't really "work" at all (Table 6).  There is no statistical significance in any of the nine results presented in the three panels.  How do the authors deal with this?  By noting that the point estimates are negative and their magnitudes fall roughly in-line with the results in Tables 4 and 5.  That's a nifty argument. 

My conjecture is that "with probability 1", all the specifications and robustness checks in your research project will not work out perfectly.  The "art" is constructing an argument that leads the reader through why you think the ones that didn't work out aren't a big deal (or why when they don't "work" in terms of statistical inference they are still generally consistent with your overall story).  It's an art because you won't (generally) be able to run statistical tests to convince the reader, but instead you will need to give rationalizations that don't seem like ex post justifications.

What's also nice about the Chay-Greenstone paper is that they benchmark what they do against the previous literature (and also attempt to replicate the prior work).  In the end, you are convinced that they made big steps forward from the previous work, even if what they did didn't work out perfectly.

UPDATE: Paper can be freely downloaded here.

Innovation and Its Discontents

Excerpt from "Innovation and Its Discontents" by Jaffe and Lerner:

In 2001, Albie’s Foods, Inc., a small grocery and caterer in Gaylord, Michigan, received an unusual communication. The letter was from the law firm of giant jam and jelly maker J.M. Smucker Co. In the correspondence, Albie’s—which markets pastries and sandwiches in the northern Michigan region—was accused of violating Smucker’s intellectual property by selling crustless peanut butter-and-jelly sandwiches.

In particular, Smucker’s claimed that Albie’s had infringed Smucker’s recently granted U.S. Patent 6,004,596 granting it broad protection on the "sealed crustless sandwich." Probably surprising the jam magnates, Albie’s decided to defend itself in federal court. In their filings, Albie’s lawyers noted that the “pasty”—a pocket sandwich with crimped edges and no crust—has been popular fare in northern Michigan since the immigration of copper and iron miners from Cornwall, England, in the nineteenth century.

A battle in federal court over peanut butter sandwiches may seem merely funny, and a little pathetic. But it is symptomatic of the larger and more profound problems with the patent system.

Cool music video post-of-the-day

"Boyz", by M.I.A.  There is even a "making of" companion video.  Pretty sweet dancing skills. 

How many no money boys are crazy?
How many boys are raw?
How many no money boys are rowdy?
How many start a war?

Econometrics question

Guiteras?  Powell?  Anyone else who reads this blog and knows some econometrics?  Quick question:

I have a state-year panel data set with outcome variable Y, endogenous variable X and two candidate instrumental variables for X, Z1 and Z2.  I run the first stage of X on Z1 with state and year FE and compare it to the first stage of X on Z2 with state and year FE.  It does not make conceptual or theoretical sense to use both instruments, so I want to pick the "stronger" first stage before running the reduced-form regressions.  The question is do I compare "strength" using the t-statistics on the Z's or the R-squared's?

They are pretty close either way but I get much more precision in the reduced-form using Z2.  Choosing based on precision of reduced-form is clearly data mining, so I'd like some principled way to choose between Z1 and Z2 based on the first stage.  And, uh, it would be great if that principle resulted in me picking Z2 :-)  After all, you can't spell econometrics without CON.

UPDATE: Tal gives the textbook response in comments.  I confess that I came up with that as I was writing the post (and fortunately for me, the partialled-out t-statistic in the first stage is higher for Z2!), but I wanted an excuse to link to Leamer's paper, which is a quick read and very entertaining.

Cool music video post-of-the-day

This could be the best thing out of France in months.  "D.A.N.C.E.", by Justice.  Incredibly catchy song, and very cool T-shirt effects in the video.

Do the D.A.N.C.E.
1 2 3 4 FIGHT
Stick to the B.E.A.T.
Get ready to ignite

It reminded me of The Go! Team, but according to the Wikipedia entry on the song, I should have instead thought of Michael Jackson.

Happy Boxing Day!

It turns out the day after Christmas Day is also a holiday.  But I'm going into work now, safely back from a short vacation in Davis, California.  I had a fun and relaxing time, which included winning some money on Tradesports by betting on the Heat-Cavs game (my rule: LeBron will not lose on Christmas Day).

Today I went to Tradesports.com to log on and check my winnings, and I got this message:

We are currently experiencing technical problems, which may prevent you from logging into your account. We are working on getting things back to normal and hope to do so ASAP. Please accept our apologies for any inconvenience this may cause.

It's not good for an options trading platform to prevent you from logging in and making trades: you are locked into any positions that expire while the technical problems persist.  Perhaps this explains [my perceived] premium on deep out-of-the-money options on Tradesports -- maybe it's a premium for "technical problems."  Of course it's not obvious that this premium would be larger and/or more noticeable the more deeply out-of-the-money the option is -- perhaps it would just result in larger bid-ask spreads across the board.  Maybe someone with actual knowledge of asset pricing can help me out.

Before the error message I just assumed that the premium on deep out-of-the-money options was a manifestation of the favorite-longshot bias that persisted in equilibrium (because of some limits to arbitrage).

Ok, this is way too much finance to start the day.  Besides, I am not going to be a financial economist.  I need to instead be reading the collected works of Ed Glaeser.  My interests are migrating towards what might be called urban economics, economic geography or what I like to call "a new combination of Labor Economics and Industrial Organization but without the Organizational Behavior, Personnel Economics, or anything to do with the internal workings of the firm."  More on that later.

Professor Lewin in the NYT

My freshman physics professor is profiled in the New York Times.  Along with Arthur Mattuck (my freshman calculus professor), Lewin was one of the two most inspiring professors I had at MIT.  They are both great examples of how raw enthusiasm is one of the most important traits of an effective teacher.

The pendulum lecture was recorded 10/1/99, when I was in the class!  Perhaps I'm in the video somewhere (I don't have time to watch now and check).

Cool music video post-of-the-day

This new video from Bloc Party doesn't make the cut.  But watching it reminded of one of my all-time favorite music videos: "Intergalactic" by the Beastie Boys.

As one youtube commenter notes, the robot is "doin the ... robot ... lol."

Also, the lyrics are pretty funny:

If you try to knock me you'll get mocked
I'll stir fry you in my wok
Your knees'll start shaking and your fingers pop
Like a pinch on the neck from Mr. Spock