Seasonal adjustments

Doug Henwood dhenwood at
Sun Mar 5 14:30:33 PST 2000

Michael Perelman wrote:

>Seasonal adjustments are necessary, but their reliability is often
>making comparisons of monthly data quite treacherous.

True enough, but the annualized 3-month change in seasonally adjusted energy prices for January was 12.1%; the year-to-year change in unadjusted energy prices wsa 14.7%. Again, these are not small numbers, so it's hard to figure out what the hell Crude(le) was talking about.

Energy has less than a 7% weight in the CPI, so even a big spike like that didn't have *that* much of an effect on the overall index.

I should say that for people who aren't familiar with the arcana of seasonal adjustment, it's a statistical technique that tries to adjust for regularly recurring movements in prices or or other statistical series. For example, heating oil prices usually rise in the winter; gasoline prices, in the summer. There's usually an influx of graduating students into the labor markets in June; retailers typically hire workers for Christmas season then lay them off in January. These sorts of movements aren't analytically very interesting most of the time, so the point of seasonal adjustment is to smooth them out. Michael is right that they're not perfectly reliable. Wall Street sharpies often try to second-guess the SA factors; if, say, Easter is later than usual, or a winter milder than usual, the normal SA factors won't work right, so you may get unexpected and misleading blips in the data. These may spook uninformed traders, so if you're set up with your better knowledge in advance of the release, you can make a few bucks. Highly paid grownups actually spend lots of time and money figuring these things out.


More information about the lbo-talk mailing list