Inspired by the Netflix contest, Nicholas Gruen and Anthony Goldbloom have created Kaggle, a site where would-be predictors go head-to-head to build a model that best forecasts the future. You can read more about it at Club Troppo, and at the Kaggle site.
Their first competition is to build a model that predicts the results of Eurovision 2010. Given the large idiosyncratic component in this kind of contest, I’m not sure how well suited it is to prediction models (the model that best predicts the 2010 result is unlikely to be the model that best predicts the winners for 2010-2019). But it’s a clever way of getting the Kaggle idea out there quickly.
Statisticians of the world, unite. You have nothing to lose but your residuals.
(xposted @ Core Economics)
On 22 April, I’m launching a project at the ANU Crawford School titled ‘New Social Policy Approaches for Sharing Risk’. More information here, including a flyer.
My op-ed today is on the economics of sleep. Full text over the fold.
Rob Bray points out to me some interesting data from 1999, in which Peter Saunders (SPRC, UNSW) asked respondents to place themselves in an income decile. Of course, 1/10th of the population falls in each decile, so if people are accurate, then the result should be 10 bars, each containing 1/10th of the population. But here’s what Saunders finds:
And for those who like decimal points with their data, here are the percentages:
In other words, only 1/10th of those in the poorest decile know it, and only 1/100th of the top decile are willing to admit it.
A decade on, it’d be worth seeing whether Australian are any better informed about their true position in the income distribution than we were in 1999.
I’ve been writing for nearly a decade (much of it with Justin Wolfers) about the predictive power of election betting markets. So why is it that now I’m running for ALP preselection in Fraser, none of the election betting websites can tell me my odds?
Some years ago, I published a paper with Tony Atkinson looking at trends in Australian top incomes since 1921. We’ve now updated the results to the 2007-08 tax year (the latest available from the ATO). Here’s the Excel spreadsheet. The ANU media release is over the fold.
Cathy Gong, Xin Meng and I have a new paper out, looking at intergenerational mobility in urban China. After making a bunch of adjustments to the data, we find a strikingly high intergenerational elasticity (implying a very low level of social mobility). The abstract is below – click on the title for the full paper.
Intergenerational Income Mobility in Urban China
This paper estimates the intergenerational income elasticity for urban China, paying careful attention to the potential biases induced by income fluctuations and life cycle effects. Our preferred estimates are that the intergenerational income elasticities are 0.74 for father-son, 0.84 for father-daughter, 0.33 for mother-son, and 0.47 for mother-daughter. This suggests that while China has experienced rapid growth of absolute incomes, the relative position of children in the distribution is largely determined by their parents’ incomes. Investigating possible causal channels, we find that parental education, occupation, and Communist Party membership all play important roles in transmitting economic status from parents to children.