Rosenberg: Seasonal Skew In February Data?

David Rosenberg, the departing economist at Merrill Lynch(ed), these days known as BofA (BAC; Shop of Barbie's Ken) pointed to some interesting observations last week:

To be sure, there have been several data releases in February that have lined up on the strong side of expectations. Caveat emptor on any February data point that is seasonally adjusted at a time of the year when winter weather typically forces most of the country into hibernation. This was no ordinary February. At an average of 37 degrees (F), the month was two degrees warmer than a year ago and four degrees balmier than two years ago. As we said, almost everyone likes to talk about how the latest data have all of a sudden signaled a turn in the economy, with retail sales, home sales, and this week’s durable goods report. Everyone was so excited about a 3.4% increase in February orders, and it seems as though the headline was taken completely at face value. But again, like so many indicators, the seasonal adjustment factor was extremely aggressive in providing a record boost (in this case) to the top-line figure. We calculate that if a typical February adjustment factor had been used, orders would have shown a 5% collapse last month. We are still in the process of trying to figure out why this happened – it could be due to the mild weather compared to the last two years. The YoY trend in the non-seasonally smoothed orders data shows that the pace is still testing unprecedented negative terrain (-29%); ditto for shipments (-20%). The durable goods inventory/shipments ratio at 1.88 is close to an all-time high. That spells more production cutbacks and deflation pressure as we move into the second quarter.
So, we do advise caution here because we have seen a very aggressive set of seasonal factors that made the raw data look extremely strong in February. The seasonal adjustment for new home sales, for example, was the strongest since 1982. For orders, it was the strongest since the data were first released in 1992. The retail sales number in February in non-seasonally adjusted terms was the worst, a 3% decline actually, on record, and yet again a strong seasonal adjustment factor made it look flat … or flattering, we should say. Beware of reading too much into the data in February when a 40,000 raw non-seasonally adjusted housing starts number suddenly becomes a headline seasonally adjusted figure of 583,000 at an annual rate.

Yves Smith at Naked Capitalism was elaborating more on related stuff ...