Lies, Damn Lies and Seasonal Adjustments

The monthly US employment figures came out on Friday and several people have already commented on them, e.g. how the government jobs (+39k) were a very large part of the total new jobs (+97k) and that the private sector added just +58k jobs, the lowest monthly number since July 2004. All this is true; but keep in mind that the figures are seasonally adjusted, i.e. they bear no resemblance to the real numbers. Employment in the US is very seasonal due to holiday patterns, weather conditions, etc. so gross numbers are adjusted in an attempt to cancel out those factors and allow for meaningful comparisons. Both sets of data can be obtained from the US Bureau of Labor Statistics.

For example, the government jobs mentioned above were reported by the BLS on an unadjusted and (adjusted) basis, as follows:

Dec. 2006: 22.494.000 (22.114.000)
Jan. 2007: 22.003.000* (22.129.000)
Feb. 2007: 22.484.000 (22.168.000)

(*The actual loss of government jobs between December and January is seasonally concentrated on the state and local education sector, i.e. school holidays)

What was in fact a net loss of -10.000 jobs between December and February ends up being reported as a net gain of +54.000. Don't you just love statistics?

I hasten to clarify that I am not claiming there is hanky-panky going on - just that seasonal adjustments can frequently muddy the waters. Even small temporary variations in the raw numbers that run counter to the established pattern, can skew the adjustment very sharply and result in almost meaningless "adjusted" figures.

In the above numbers, the established pattern must have expected an even bigger loss of government jobs between February and December than the -10.000, so the seasonal adjustment shows instead a "phantom gain" of +54.000.

Unless you are an actuary, aren't you bored yet? Wait, there is a point to all of this: let's compare raw, not-seasonally adjusted employment figures. Over longer time periods the seasonal variations won't matter. Shown below is a chart of the rolling 12 month net change for all private sector jobs and further below is the change in annual percentage terms. I use the private sector jobs as they better reflect the "real" economy. (The patterns are the same when govt. jobs are included, anyway).

Source: BLS, figures in thousands

The immediate, relevant observation is this: jobs are not being created in anywhere near the same number or rate as previous times of expansion. We should have been seeing around 4-5% gains in employment instead of the current 1.5% - and it looks like the line is rolling over now, heading even lower.

Bottom line: Employment growth is crucial because a negative saving rate and huge debt load are pressuring US household finances like never before, i.e. earned income growth is key to averting a crisis. So far it does not look very good.