Understanding why some countries are poor and why some grow than others is probably one of the most important questions in Economics. The traditional tool to tackle this challenge has been growth regressions: use cross-country data and regress the GDP growth rate on various indicators that could be relevant in order to find which matter most. These regressions have been abused over the years, especially as there are obvious endogeneity and collinearity issues. Also, the results are driven by a multitude of (poor) countries where data quality is quite horrendous. The worst is probably all the data mining that is going on in this literature, which culminated with Xavier Sala-i-Martin's two million regressions.
Tatu Westling uses a variable the previous literature completely ignored: the average length of the erect human penis. Adding this variable to the regression shows a U-shaped relationship for the GDP level, explaining 15% of its variation. The optimal penile length is 13.5 cm, and 16cm is disastrous. For GDP growth, the relationship is negative, explaining 20% of the dispersion. This is not negligible, and more than institutional variables that are thought to be the key to growth and convergence.
I wonder how many people will take these results seriously and try to get policy recommendations from it. Westling hypothesize about the impact of self-confidence. The paper is very well written, taking the 'male organ hypothesis' very seriously, but in truth tongue-in-cheek. Very different from this study on flag colors I wrote about previously.
Tuesday, August 16, 2011
Subscribe to:
Post Comments (Atom)
9 comments:
With such a title, I fear for the spam comments now.
Interesting implications for genetic engineering if there is a causal relationship. The optimal length for an individual would likely deviate from that for society.
This reinforces my belief that migration is the solution to development problems. We should should have migration policies that target the optimal penile size. It is only the countrywide average that matters, right?
Is that a huge penis you have there or just the rule of law?!
Economic Logician already mentioned the key issues. Since I was invited here are my two cents.
---------
Irrespective of the avenue - movies, scholarly endeavors or indeed casual conversation - a healthy dose of sarcasm is always good. As we all know it can sometimes clarify things in a more digestible way than more serious approaches. I'm not at all sure whether 'Male Organ and Economic Growth' scores on this but at least hope so. As might be evident, in this case the target is the 'let's put this Buddhism variable here and see how many asterisks appear' mindset - with a zero knowledge of Buddhism I still guess it would deserve a slightly more elaborate treatment, however naive this sounds.
Some [though not many] have retorted that the approach is a thing of the past, and MOEG misses its point. Might be, but Google Scholar seems to disagree. Search for 'cross-country growth regressions' yields 56 700 hits today*. For reference, on August 1st 2011 the figure was 56 300. Despite the problems inherent in these approaches being very widely-known, the literature expands unabated. By instinct an economist must ask: how many similar papers are still needed?
If MOEG can through this publicity make the August 2012 figure 64 998 instead of 65 000 and especially take the Buddhism dummy out of the next two, most of us should be happy - I’m quite sceptical, however. Yet sometimes it takes few puns to make points like these.
Any comments?
*Not a scientific measure but for rhetorical purposes it suffices
I am completely with you. In this literature, it is so easy the add some random indicator, find it significant and then find someone willing to publish it. Yet we learn nothing from it.
Your paper is of course of a very different value. But it will take some courage for an editor to accept it.
Re: Sala-i-Martin's article.
The published version has a title with two million regressions. The working paper versions have four million. What happened?
One interesting, dare I say sociological, blogosphere-observation has been the strong confidence in economic data contra the 'less-formal' male organ variable. The latter evidently suffers from questionable quality but by biological factors the figures are likely to be in the right ballpark. Yet when it comes to economic data, it might be somewhat uncertain how well GDP measurement and say Chad 1960 mix. Nevertheless, judging by the blogosphere reaction, it seems that it is more reliable to measure the GDPs of Sahelian tribes than body parts. The general trust on anything economic or 'official' is quite amazing, which is something at least I had not fully realized until this MOEG incident.
I agree again. data quality is very much overrated. I discrussed previously how poor data quality can make that data revisions leads to result reversals in growth regressions. Yet everybody happily continues.
Post a Comment