Friday, August 08, 2008

I pretend to be a statistician


So, I had time to kill at work, and while reading a Wikipedia article on the minimum wage, came across a chart that was begging for a regression line. I plotted unemployment rates vs. the ratio of the minimum wage to the average hourly wage, for the years 1980-2008. (Sources: IMF for unemployment, BLS for wages.) I had Excel add a best-fit line.

I'm sure there are all sorts of flaws in this very basic analysis (Pete, feel free to point them all out). Still, considering the huge effect business cycles have on unemployment, an R-squared of .36 for the minimum wage seems very significant.

Of course, to anyone who's studied basic economics, the idea that the minimum wage would cause unemployment is nothing less than obvious. Says the ever-reliable Wikipedia, "Price floors set above equilibrium market prices cause surpluses." When the government guarantees a minimum price for farm produce, too much is grown and not enough is demanded, and the excess supply has to be dumped on foreign markets. When the government sets a minimum price for labor, supply exceeds demand, which leads to a whole bunch of people who can't find anything else to do all day but sit at home and watch X-Files reruns. Which reminds me...

Edit: I just realized that business cycle effects could be showing up in the model; lower wages during recessions would make the ratio of the minimum wage to the average wage higher. Oh well.

2 comments:

Sean said...
This comment has been removed by the author.
fatrat said...

Business cycle component probably should be added to the regression model. Plot the residuals against your X-axis variable. Perhaps add a second variable to the regression, like economic growth rate.

There's also the Austrian Economic theory. Maybe see if bursts in the money supply or low interest rates are part of that business cycle and can account for a pattern in the resulting residuals.