THIS COLUMN IS not usually one to feature the gloomy side of the job and jobless figures, but a few words of caution are needed on the latest announcement that the unemployment rate last month fell to 8.6 from 8.7 per cent.
In the first place, does it not strike you as a little odd that a month in which the number of people with jobs declined and the number of people in the labour force rose should be a month in which unemployment was lower? Yet, there you have it. There were 500 fewer people with jobs but 200 more people in the labour force and yet the unemployment rate was down.
The reason of course is seasonal adjustment, courtesy of that piece of Canadian rocket science, X-11-ARIMA, which our statisticians now use to smooth out the figures over the course of the year.
August happens to be the month every year in which the X-11 pushes the raw unemployment figures down by the greatest extent. As the first chart shows, if you make the calculation on just the raw figures (unemployed as a percentage of labour force), you get a record high unemployment rate of 8.83 per cent, significantly higher than the X-11 figure and rising slightly, too.
Now, I do not wish to denigrate seasonal adjustment. It is a legitimate process for ensuring economic statistics do indicate real trends. It does not make a huge difference to the unemployment figures, but it does to other data series and I quite understand why statisticians should use it whenever they see fluctuations that are clearly seasonal. In this case, however, it may have been preferable if the notation sa stood for shock adjustment rather than seasonal adjustment. What we really want to know is how much of our unemployment rate is attributable to the Sars outbreak earlier this year.
This is not possible to calculate with any precision but Sars must still have had much more impact than the normal seasonal adjustment and skewed the finely calculated trajectory of the X-11. Perhaps we may want to put a little more emphasis than normal on the raw figures. There is another reason for doing so, too. The monthly unemployment figures are taken from the General Household Survey (GHS). In other words, there is not a complete count of employed and unemployed but only a partial survey from which these figures are derived.