Retail sales fell just 0.1 percent in February, following a surprising 1.8 percent jump in January, observe the analysts at Wachovia. The better showing followed six consecutive monthly declines, the last three of which were monstrous. Since consumer spending makes up two-thirds of real GDP, the unexpected improvement in retail sales was heralded as a sign the economy might be stabilizing.

While we hate to rain on anyone's parade, claims that the past two months retail sales data are signs of stabilization in the broad economy are just wishful thinking. Seasonal adjustment accounts for all the improvement.

Retail sales are adjusted for seasonal variation and holiday and trading day differences. Such adjustments are necessary so that we can look at retail sales on a month-to-month basis and so that we can also adjust for holidays that move about the calendar a bit, such as Thanksgiving and Easter. In addition, this process helps take into account leap years.

While the intentions of seasonal adjustment are to smooth out distortions in the data, sometimes they serve to amplify them. We believe this is the case today. We have indexed two series: overall retail sales and retail sales excluding motor vehicles, gasoline and building materials stores, back to July. The bars shows the average change from the July level for the past seven years and the line shows what has happened this past year. Back-to-school sales normally push retail sales higher in August and then sales fall back in September. October and November typically see modest gains and then sales surge as the holiday season kicks into full swing in December. Retail sales then plummet in January and slide further in February.

This past year saw sales take a slightly different path. Overall sales fell slightly in August and then fell more sharply than usual in September. Instead of rising in October and November, sales fell, slightly at first and harshly in November. Sales did rise in December but nowhere near as much as usual. Since sales did not rise as much as they usually do, sales did not fall as much as usual either, dropping 19 percent in January instead of the usual 21.3 percent. The smaller than usual non-adjusted drop is what produced the large seasonally-adjusted gain reported that month.

Oddly enough, sales fell more than usual in February, falling 3.9 percent on a non-adjusted basis compared to a typical drop of around 0.8 percent, but the adjusted figures somehow show just a 0.1 percent drop.

We repeated the same process for retail sales excluding motor vehicles, gasoline and building materials and found similar results. Sales for this key core spending measure did not rise anywhere near as much as usual going into the holiday season, so the subsequent declines look less spectacular than usual. The underlying trend, however, is unmistakably weak and casts real doubt on the notion retail sales are stabilizing.

For comments and feedback: contact