« New Media Innovations | Main | John Taylor on the Fed's "industrial policy" »
Good News
My chances of winning the John Bates Clark Medal just doubled. Oh wait, I am over 40. Damm...it's still zero. Still, if you are an economist under the age of 40, your chances just doubled since the executive committee of the AEA has voted to make it an annual rather than a biennial award.
Posted by Alex Tabarrok on January 4, 2009 at 04:19 PM in Current Affairs | Permalink
Comments
Good news! Your chances have doubled!
2 x 0 = 0
Your chances can double and still be zero. Just don't try dividing anything by your chances of winning, and you should be fine.
Posted by: apotheon at Jan 4, 2009 4:49:47 PM
Devalued would be another way of looking at it. And what about the incentive effects; why not make it annual effective 2019?
Posted by: John B. Chilton at Jan 4, 2009 4:52:31 PM
Definite dilution.
Posted by: liberalarts at Jan 4, 2009 5:17:58 PM
Definite dilution.
Posted by: liberalarts at Jan 4, 2009 5:21:26 PM
It's a meta-Keynesian prize inflation to stimulate the economists.
Posted by: Harkins at Jan 4, 2009 6:30:58 PM
Definite dilution.
Not really, the Pentagon awarded a number of medal higher than the numbers of soldier in the Gulf War. Based on a study that says that people dont care if everybody has one
Posted by: k at Jan 4, 2009 6:43:57 PM
I realize Alex was kidding about the doubling, but it's probably not even true for the 30-something faculty member. For one thing, if some hotshot 25-year-old under the old rules had a 55% probability of winning it, then obviously his or her probability didn't just double.
And as others have suggested, if the extra award induces more people to become economists...
Posted by: Bob Murphy at Jan 4, 2009 6:46:37 PM
Let p be the probability of winning the John Bates award in any given year. Assume there will be n awards given out before you turn 40.
p + p*(1-p) + p*(1-p)^2 + ... + p*(1-p)^(n)
If n was the previous number of awards, then 2n is the new number. Thus, the new probability minus the old is:
p*(1-p)^(n+1) + p*(1-p)^(n+2) + ... + p*(1-p)^(2n)
Which is not equal to 2.
This is why you should hire me as your research assistant Alex.
Posted by: Martin at Jan 4, 2009 7:53:15 PM
Martin,
How much do you charge? I might be interested?
David
Posted by: David R. Henderson at Jan 4, 2009 8:14:21 PM
if they can switch to give it every year now, they can switch later to give it twice yearly and so on until it is given multiple times to each person on earth every femtosecond. in the limit this award may be worth very little.
Posted by: babar at Jan 4, 2009 8:27:09 PM
David,
If you are David Henderson, the research fellow at the Hoover Institution, then I won't charge much. I am in the market for a full-time RA position starting this summer (continuing for one to two years). I can also work on an hourly basis if it can be done from Washington D.C. E-mail me at 58saavedra@cua.edu if you are interested.
Best,
Martin Saavedra
Posted by: Martin Saavedra at Jan 4, 2009 10:30:40 PM
Martin, youre assuming a constant probability of winning
Posted by: benid at Jan 5, 2009 12:17:15 AM
This will reduce (though not halve) the famed percentage of John Bates Clark Medal winner who go on to win the Nobel Prize, which, once the trend becomes clear, may devalue the JBC Medal. Currently its signalling properties are a major part of its prestige.
Posted by: Timothy at Jan 5, 2009 12:56:14 AM
Thus, the new probability minus the old is: (...) Which is not equal to 2.
p_new minus p_old should yield p_old if the chances double, not 2.
Martin, youre assuming a constant probability of winning.
Do you have a function of p over t in mind that doubles the chances for all p?
Currently its signalling properties are a major part of its prestige.
While the percentage of JBC medalists turned Nobelists will go down, the absolute number will likely go up. Since (media) queries are usually only done for JBC medalists who have gone on to win the Nobel, the prestige would then go up. Which is likely the impetus for the increase.
If you are David Henderson, the research fellow at the Hoover Institution, then I won't charge much.
So and if DRH is not the Hoover fellow, would you charge less or more?
Posted by: ogmb at Jan 5, 2009 5:03:11 AM
The chances increased by the factor (2 - p), where p is the old chance of winning when the award was given out every two years.
-------
If your chance of winning the award when it is awarded is p, and it is only awarded every two years, then your chance of winning in a two-year interval is p.
If it is awarded every year, and you can only win once, then your chance of winning in a two-year interval is p + (1-p)*p = 2*p - p^2. (p is the chance of winning in the first year; (1-p)*p is the chance of not winning in the first year, but winning in the second year.) This is 2 - p larger than your chances of winning when the award is given once every two years, provided p is not equal to zero.
(You should divide, not subtract, to see how much the probability increased.)
Posted by: flyfishr at Jan 5, 2009 8:09:29 AM
Never hire somebody who posts his email address in a blog.
Posted by: meanie at Jan 5, 2009 11:05:35 AM
To continue beating a dead horse:
Holding the probability of winning the award each award cycle constant, the total likelihood of winning the award over one's eligible time frame is:
s = 1 - (1 - p)^n
where p is the probability of winning in a given cycle, and n is the number of cycles over one's eligible time frame (this is just a restatement of Michael's summation above).
If we double the number of cycles in the eligible time frame, we get:
s2 = 1 - (1 - p)^2n
And the ratio of these:
R = s2/s
is the object of consideration. For starters, we can see that as n and p increase, R approaches (and ultimately equals) 1. So for very young, very talented economists, this change has between very little and no effect. But what about everybody else?
Well, R is undefined at n=0. So in that sense, Alex's chances of winning are a trivial case. But what about the generic new, budding economist? R is, of course, decreasing in p, and the limit of R as p approaches 0 is 2 irrespective of n. The question, of course, is how close to that limit a particular economist is. Let's use Alex back when he got his PhD in 1994. In the original JBC medal schedule, he had six shots at the medal. He'd get 12 shots under the new schedule. So we have:
R = [1 - (1 - p)^12]/[1 - (1 - p)^6]
So if most new PhD economists have between very little and absolutely no chance of winning the JBC medal in any given round (say, p=0.01), and n is somewhere between, oh, 5 and 8, then the new economist's chances increase somewhere between 92 and 95 percent. If one thinks that a new economist's chances of winning in any given round are even lower, Alex's statement improves in accuracy.
In short, while Alex's off-the-cuff comment was imprecise, it is not an unreasonable characterization of the change in total lifetime probability, given the assumption that p is invariant to n. If benid's point is that a change in the regularity of issuing the JBC medal may influence economists' behavior (and thus their chances of winning in a given cycle), it is well taken, and requires some explicit model of p itself.
Posted by: Jared at Jan 5, 2009 2:15:37 PM
>Never hire somebody who posts his email address in a blog.
Never take hiring advice from snarky blog commenters.
Posted by: liberty at Jan 5, 2009 3:33:11 PM
I just returned from the meetings. Word there was that this decision was made because of what happened during the last decision process. I guess some people were incensed that John List did not win the JBC medal last time. Proponents of an annual award used that as a springboard to argue that there are at least 2 people every other year who are deserving so why not give it to them. The List rule may not help List himself, but it will help future economists like Alex.
Posted by: justice at Jan 8, 2009 10:29:35 AM
