> Trivia question number two: 

>  there was a correlation between the lengths of women's skirts
> and the business cycle.  
> 
> (There is also supposedly a study linking the number of dairy cattle in 
> Kentucky and the business cycle.  Any help here?)
> 

Often, you will find similar  examples in Econometric textbooks to 
explain "spurious"  correlations (or sometimes "non-sense correlations", 
as denominated by G.Undy Yule). In particular, business cycles are 
very prone to misleading results in econometrics due to statistical
properties of cycles (random shocks will, in average, generate 
cyclical patterns; it is only a question of finding the "right" 
random examples to associate with the cyclical data under 
investigation...).

The best account of this is in Morgan, M.(1990) "The History of 
Econometric Ideas", Cambridge University Press. Chapter 3 of the
 book relates the pioneering works of Hooker (1901 !), Yule (1921 and 
1926 !), Slutsky (1927 ! ) and others, who presented evidence of 
"spurious" correlations such as marriage rate and international 
trade, experiments with harmonic curves, lottery results and the 
business cycle, etc. Not only did they present the evidence or 
experimental results, but also managed to demonstrate their 
statistical flaws.

To me, it is still amazing to see that for almost (already) a 
century, the economic literature is plagued by studies which "prove", 
based on "sound" or sophisticated  econometric techniques, a priori 
hypotheses which might just be "nonsense"... 

Probably is to keep busy the academia... 

Salud,

Alex

PS. I recently built a "sound" econometric model, for a workshop at the
ISS,  in which I "explain" the Kenyan GDP as  being correlated with 
a set of variables denoting a Keynesian expansion. The econometrics 
fit rather well, the only problem is that the exogenous variables 
were not Kenyan, but Ecuadorian... If anybody wishes it, I have no 
problem to send an attachment with the data and results to 
"reproduce" the experiment in a classroom (such things should be
public property, I guess...) 
 



Alex Izurieta

Institute of Social Studies
The Hague - The Netherlands
Email:  [EMAIL PROTECTED]
Tel.    31-70-4260480
Fax.    31-70-4260799


Reply via email to