Tuesday, February 23, 2010

Freezing Weather Freezes Economies

Last month, I told you about how unusually cold weather was a factor increasing the probability of a "double dip" recession in the U.K.

Now we are seeing more examples of how the cold winter in the northern part of the northern Hemisphere is reducing economic activity.

The U.S. payroll employment number for February will likely be negatively affected by the blizzards that struck much of the North East United States, while the German ifo-index fell as cold weather reduced both retail sales and construction activity. Meanwhile, much of the Swedish rail system (and to a lesser extent some car/truck and air traffic) was paralyzed by extreme cold and large amounts of snow, something which will likely negatively affect many business activities.

If only we had more "global warming"....

Aside from that, another interesting observation one could make is that this makes "seasonal adjustments" more problematic. While I think "seasonal adjustments" makes sense in principle (assuming the absence of below mentioned practical problems), it is often a lot more problematic than many people think in practice applied to actual statistics.

Here we have a clearly seasonal (in the northern hemisphere, cold- and snow related problems only arise during the last and the first months of the year) factor depressing economic activity, yet because the severity of this seasonal factors varies from different years, the standard statistical method of historical averages will easily become misleading. If it was possible, one should let the factor vary from different years instead, but the risk is that this could allow statisticians to make arbitrary judgments, possibly influenced by political factors. Finding a way to objectively assess the different seasonal effect each year is something which aspiring statisticians should think about.