Thursday, 14 January 2010


Haiti's tragedy is before our eyes as we look at the news. I find myself thinking about the Lisbon earthquake of 1755. At the time, the dismay was so general that it presented a real problem for theologians - how can a benign and all-powerful God let something like this happen? Voltaire used the earthquake to undermine the arguments of Dr Pangloss in Candide [Kant was also fascinated by it, and even wrote a book trying to explain the Lisbon quake as a result of subterranean caverns full of gas shifting -- one of the first serious attempts to produce a scientific explanation]. Today, there is little discussion along similar lines. In a way, the 18th century reaction to the 1755 quake is charmingly innocent. I guess the 20th century offered so many impressive horrors that it is difficult to think that a bad earthquake accentuates the need for a theodicy.

Why did piece rates disappear?

Rosario Macera from UC Berkeley was here yesterday, giving her jobmarket paper on loss aversion and wage contracts. I am not an expert in the field, and have to admit that I haven't thought much about the fact that monthly pay is mostly fixed, and variable pay arrives on a yearly basis. If you think about it, piece rates sound just like what Dr Smith ordered -- pay changes 1:1 with measured output. Sure, measuring output can be hard (try to assess the productivity of an academic, say...), but there must be many jobs where this is possible. Nonetheless, it's super-rare. Good science is all about making one look at something that seems obvious and not in need of thought, and turning it into an interesting research question. Rosario's talk did that for me (perhaps because I know less than most of the topic). She married a pretty standard setup with some loss aversion, and derived the optimal wage contract, which ... you guessed it, looks a lot like real life.

Now, as an economic historian, I couldn't help wondering... we used to live in a world with LOTS of piece rate work, plus a daily threat of dismissal (or not being re-hired). Why is it that 19th century labor markets didn't create the same kind of long-run stability, given that workers (in a way) are willing to "overpay" for stable paychecks? Standard stories of why we get long-term contracts focus on either the rise of unions, or on the different skill requirements in modern industry. But this can't be quite right -- even in textiles, where work is (for all I know) very similar today to what it was 100 years ago, there is much less work under piece rate contracts. Ditto in the mines. So are people more loss-averse today? Does the rise of mortgage debt and auto financing mean that, despite much fatter paychecks overall, people hate any decline in their income much more? Anyway, its nice to come away from a jobtalk and have many more things to think about than before... and good luck to Rosario.

Monday, 11 January 2010

It don't mean a thing...

As the flyout part of the recruiting season is about to start at UPF, I thought it might be a good thing to put a bit of a swing into everyone's step... from my colleague Michael Greenacre comes this rather splendid reinterpretation of Duke Ellington's "It don't mean a thing... if you don't do modelling":
It don't mean a thing if you don't do modelling,

It don't matter if you're frequentist or Bayesian,
You just need a model with some alphas and betas, x's and y's, and i's and j's and k's in!

So it don't mean a thing if you don't do modelling,
[I will suggest that ITFD statistics supremo Kurt Schmidheiny adopts it as the theme song of our quant methods class]