Tuesday, May 19, 2009

Optimization problems not appearing in AER

I was walking from Union Station to the University of Toronto about 8pm this evening, which is apparently when the homeless start settling down for the night on Bay street. I found this very surprising my first time in Toronto a couple of years ago, I wasn't expecting such a visible homeless presence.

However, what I found particularly strange today was that the people who had set up a sleeping spot had uniformly chosen highly visible spots, typically the middle of the sidewalk in front of a major building. It seems to be it would be preferable to find a more secluded spot on a number of fronts: noise, chance of being dislodged by a policeman, but perhaps it reduces the chance of being robbed while sleeping? A form of silent protest?

What variables enter into this optimization problem? Given the repeated nature of the game, it seems unlikely that these people are away from their maxima. Can anyone think of more compelling reasons to sleep directly outside of the TD building instead of in a park?

Friday, May 8, 2009

Links

Leading off, solid labour force news. Can I be one of the cool bloggers and say "green shoots" as well?

Freakonomics asks how to optimally divide the rent of a shared house and proposes some rather uneconomic methods, when a Rochester student already figured out this problem for his job market paper. Paper here. (He went to Texas A&M.)

EnvEcon reiterates the point that debating carbon tax versus cap-and-trade isn't that important relative to implementing either over a much-less-desirable policy. I had a New Years' resolution not to nitpick at the differences.

Rats more rational than humans. What you wouldn't expect is that William Easterly starts from there to segue into a warning about the dangers of data mining. I, for one, cannot explain why there should not be a standard algorithm for tackling any econometric problem.

Thursday, May 7, 2009

Lessons

Okay, because I can talk about this subject with a degree of legitimacy now.

I remember reading some sort of 'guide to graduate study in economics' a year or so ago, written by some people in the UC system. I don't really remember much, except one line. "In grad school, you are judged by the fruit, not the sweat, of your labours." Okay, I'm paraphrasing a little, but that's pretty close. Ir's incredibly true.

Don't stay up until 5am trying to fix your code so the robust standard errors on your weird model match up with the textbook's to the fourth decimal place when the first two already match. It's just not feasible to do everything and remain sane, so setting priorities becomes extremely important. You don't want to become the person who lives out of their office working 90-hour weeks. Memorizing the proof to show that the set of essential finite games under the sup metric is residual is just not worth it.

In particular, a crucial lesson I'm still working on learning is being able to put a wall between economics and leisure. I remain terrible at resolving to take the morning off and put in a decent rest of the day. Often I will just blankly click around the internet until lunch while fretting about what work I have to do, at which point I start an internal debate on whether I've taken any time off. This is not a good way to spend time.

Okay, math. The importance of mathematical preparation decays throughout the first year. There were very few new mathematical techniques introduced in the second semester. (Though the first micro II class was a little freaky when the prof claimed the Riesz representation theorem was "standard off-the-shelf" stuff we should know.) The first term, preparation helps you a lot. Math for economics, mathematical statistics, Bellmann programming. If you've seen them before, it makes a big difference. I can show you my <50% on the mathematical econ midterm early last October for proof.

What is much more important than being exposed to lots of high-level techniques is 'mathematical sophistication'. This gets thrown around a lot and it's hard to define. Conditional on being told "use a second-order expansion here to derive blah blah blah", it's not difficult to mechanically go ahead and solve. But staring at a problem you've never seen before under severe time constraints, the real skill is to recognize that a Taylor expansion is the preferred method of attack. (And since any problem on a final exam has probably not been seen before...)

Yes, this is correlated with exposure to high-level materials, but not perfectly. I don't know how to acquire this talent, and it's not something I'm particularly great at, but it's crucial. Being really comfortable with the material at the level of Rudin is a lot more important than having been in a course that worked through a lot of weird results in functional analysis.

After this, everything else I could talk about becomes a lot less important and/or obvious, if what I said already wasn't. Stay as healthy as possible. Spend as little time thinking about money as possible. Don't be a moron in social situations with your classmates. Try to remember why you're doing it all in the first place.

Why does the NHL prefer Phoenix to Hamilton?

Dunno. The court filings say because it "unreasonably restrains competition in violation of the antitrust laws".

Can't say I buy that. Moving a team from a monopoly position in Phoenix to compete with the Leafs hardly sounds anticompetitive. In fact, I have no idea how antitrust laws even come into play here, because it's all the same league. Do the Leafs have that much sway in the overall league?

Presumably the NHL has got to like the fact that a team in Hamilton probably wouldn't lose the $20 million the Coyotes lose. The only thing that I can think of is a continued effort to grow the game south of the border would be helped by the Coyotes staying put. Am I missing something?

Monday, May 4, 2009

Back in the fray

Whee, blogging. I haven't done this in awhile. Anyway, I'm finished exams. Summer vacation stretches forward, I have a serious playoff beard, and it would probably kill my grandmother to observe the state of my apartment. Will do some roundup posts of the first year in the near future and get back to econoblogging.