27 December 2010

Economic cycles

Here is an article that finds that cycles are real: http://bit.ly/hwUT8i.
They do a spectral analysis of GDP data. In other words, they
transform data in a time series to a spectrum and find cycles,
significant statistically speaking.

There are reasons to think that cycles might not be real.

There's the Slutsky proposition (getting "slut" and "proposition" in
the same name supposedly helps as a mnemonic device), which states
that "the moving average of a random series fluctuates." In other
words, a simple mathematical manuever on a random series can produce
meaningful-looking cycles.

There's the cluster illusion, where humans, given random data, will
think it's not random and will try to ascribe meaning to the data.

There's the efficient market hypothesis, which states that, if a
market has predictable cycles, people will act to remove the cycles.


No comments: