- The collapse and ensuing rebound in economic data caused by the 2008/09 financial crisis seems to have affected the seasonal adjustment of two key US data series – the ISM manufacturing index and the payrolls report.
- The change in seasonal factors has generally lifted data during the winter months and now we expect to see some payback, which has already materialised in February’s manufacturing ISM.
- Next year, the distortion of data should fade markedly, as the models used to adjust for season generally give the highest weight to the most recent years. Watch out for seasonal quirk this spring
Seasonal adjustment of data series is used to separate cyclical fluctuations in data from those caused by pure seasonality (i.e. production usually slows somewhat in the last few months of the year and rises in spring). However, it seems that the bust caused by the financial crisis in 2008 and the following boom has caused some distortions to the seasonal adjustment. In other words, some of the cyclical swing in data has been interpreted by the models used to adjust data for season as being caused by seasonal factors. We have found evidence of this in two key US data series – the manufacturing ISM index and the goods-producing part of payrolls.
Comparing the seasonal factors in 2008 (before the distortion) with the current seasonal factors in general shows that both the ISM and payrolls got an extra boost from the seasonal adjustment in the months of November and December and this boost is now fading and will turn into a direct negative in March and April.
For payrolls, this means that the distortion will turn from adding a positive of 25-50K per month to the seasonally adjusted data to a drag of 40-60K in March and April. We are still positive on the underlying trend in payrolls, which is currently running close to 300K, and we are confident that the economic recovery will remain on track for the coming year. However, it does lower the chance that we will see a +300K gain in overall payrolls this spring.
According to our calculations, the seasonal quirk added just above one index point to the ISM in both November and December and we saw some payback in February. However, the main negative will come in April with a drag of approximately two index points. The seasonal adjustment should follow the normal pattern in the months thereafter. Next year, the distortion to data should fade markedly as the models used to adjust for season generally give the highest weight to the latest years.
Bottom line
If data turns out on the weak side of expectations over the next couple of months, one should keep this seasonal quirk in mind and not be caught up by springtime blues at once.