Are Seasonal Adjustments Affecting Employment Data?

posted in: Uncategorized | 1

As May approaches, 2013 continues to look like 2010-2012 on repeat, and this has many asking whether or not we can expect a seasonal slump once again.  Even at the Fed’s most recent news conference Bernanke opined on the pattern for spring weakness, and he gave one possible statistical explanation:

Well, you’re absolutely right that there’s been a certain tendency for a spring slump that we’ve seen a few times. One possible explanation for that—besides some freaky things, some weather events and so on, one possible explanation is seasonality. Because of the severity of the recession in 2007 to 2009, the seasonals got distorted. And they may have led—and I say “may” because the statistical experts—many of them deny it, but it’s possible that they led job creation and GDP to be exaggerated to some extent early in the year. Our assessment is, though, that at this point, that we’re far enough away from the recession that those seasonal factors ought to be pretty much washing out by now. So if we do, in fact,see a slump, it would probably be due to real fundamental causes and then we would obviously have to respond to that. As I said, we’re planning to adjust our tools to respond to changes in the outlook, and that can go either direction.

Going back to 1939, the monthly seasonal adjustment of payroll data does vary from year to year.  However, looking at the data, it doesn’t appear that there has been a major effect on the seasonal adjustment of payrolls in the last five years (highlighted in red).  If anything February and March have seen fewer jobs added by seasonal adjustment while the spring months have seen more jobs added.

Nonfarm Payrolls Seasonal Adjustment By Month

Since the above chart gets a little busy, below is a chart of the difference between seasonally adjusted (the headline number) and non seasonally adjusted payrolls for May alone since 1939.  Interestingly, before 1970, May payrolls were seasonally adjusted positively.  In other words seasonally adjusted payrolls were reported to be higher than actual payrolls.  However, since that time, May has become a month that is seasonally adjusted downward.  Seasonally adjusted payrolls are about half of a percent lower in May than NSA payrolls, and that adjustment hasn’t changed much in the last five years either.

Nonfarm Payrolls Seasonal Adjustment May