Monday, December 7, 2009

Watch out for a rebound in seasonally adjusted economic data for winter

I have been of the opinion that autumn would see a weaker growth picture due to technical factors (see for example Autumn growth weakness is finally becoming evident dated Oct 2). A key reason has been that seasonal adjustment factors are likely to overstate the extent of the true seasonality at the current juncture. Given the massive amount of job losses and the huge decrease in industrial production during the recession (and especially following the collapse of Lehman), the usual economic slowing into summer should have been less pronounced than is generally the case. With that, seasonally adjusted data are likely to overstate the dynamic of the economy going into summer. However, it directly follows that the usual seasonal strengthening into autumn should be less pronounced than is usually the case and therefore seasonally adjusted data for autumn are likely to understate the true dynamic of the economy. Indeed, while economic data during the summer months surprised positively on average, economic data was in line to slightly negative on average during the period of September-November. The chart below shows the Citigroup economic surprise indicator which confirms this picture. However, with winter having started this story is turning around again.
Citigroup US economic surprise indicator: autumn slowdown seems to be over

Source: Citigroup via Bloomberg

I think that the change in the seasonal adjustment factors could already be seen with Friday's employment report. First, this was a positive employment report. Most notably the pace in the hiring of temporary workers has increased and the development of temporary workers is usually a good leading indicator for the hiring/firing of permanent employees. Furthermore, the index of aggregate weekly hours moved from 98.5 to 91.1, the largest uptick since March 2007, i.e. before the start of the recession. However, it seems that seasonal adjustments played only a relatively small role for the November report. According to the establishment survey, the seasonal adjustments assumed a "normal" job creation for November of 91k vs. 820k for October. The chart below shows the seasonal adjustments of the past 12 months (in dark blue) as well as the average of the seasonal adjustments of the 9 years prior (in light blue). As can be seen the seasonal adjustments assume a significant slowing of the job market during summer as well as winter (December and especially January).
Seasonal adjustment factors for employment
Source: BLS, Research Ahead

Furthermore, it can be seen that the current seasonal adjustment factors are not much different from the usual seasonal adjustments. This makes sense from a statistical point of view and clearly during periods where economic volatility is not much pronounced. However, in the wake of the financial crisis, we do still not have a 'normal' economic environment. Just to repeat, I think that given the magnitude of the previous job losses, the seasonal adjustment factors overplay the current importance of seasonal economic weakness and strength. Given the large seasonal adjustment to the October jobs data (which was even larger than usual), the weakness in the job market during October was likely overplayed and with seasonals being amongst the lowest for November, the latest jobs data might show a truer picture about the underlying dynamic.
Looking ahead, we are about to enter the period of the most pronounced seasonality. In turn, what we witnessed during summer and autumn might be even more pronounced during the upcoming months. Just one example is retail hiring and firing. Usually retail hiring is strong during October and November in the run up to Christmas (see chart below, courtesy of CalculatedRisk). However, especially last year but also this year has seen less hiring than usual. If less people are hired for the pre-Christmas period, less people will be fired again going into January.

Given the large likelihood that the usual seasonal economic weakness is likely to be less pronounced than seasonal adjustment factors assume, seasonally adjusted economic data for December and January is likely to show a healthier momentum than the data for autumn.

No comments:

Post a Comment