•  About Me 
  •  FAQs 
  •  Mail Me 
  •  Links 
  •  Archives 
March 16, 2012

134,000 Fewer Jobs Than 2008

BC had 134,000 fewer full-time jobs in February 2012 than in July 2008. Part-time jobs increased by 100,000 over that period, so total employment in February 2012 was 34,000 below the July 2008 level.

Three or four years of little or no employment growth is not the picture Premier Clark wants to tell with her jobs plan. Using a North Vancouver shipyard as a backdrop, on March 13th she released her version of job growth in a report on the first six months of her jobs plan. It was a bad media day for her as reporters were more interested in the Telus-BC Place naming controversy, and under the heading "Clark's jobs plan worked better before it was announced", Andrew MacLeod of The Tyee was quick to criticize her job numbers. Clark claimed that BC added 39,900 net new jobs to the economy. The correct context for that number is to say seasonally adjusted employment as measured by Statistics Canada's Labour Force Survey in February 2012 was 39,900 greater than it was in February 2011. Clark first announced her jobs plan in late September 2011 through a series of events, concluding with a speech to the Vancouver Board of Trade. As MacLeod pointed out, on a seasonally adjusted basis the job gain since September was only 2,000, while 37,900 jobs were created between February 2011 and September 2011. The unadjusted numbers are more dramatic: a gain of 76,600 jobs between February 2011 and September, followed by a loss of 33,400 jobs between September and February 2012. This raises the question of when is it appropriate to use seasonally adjusted figures, and when is it more accurate to use unadjusted numbers; the unemployed might ask, "have you been seasonally adjusted lately?"

LFS graphs Jan 2006 - Feb 2012The black line in the graphs shown here connects the raw, unadjusted, employment numbers from the Labour Force Survey; the red line connects the seasonally adjusted numbers. The actual number of jobs in any month is what is shown by the unadjusted numbers, but for some purposes when different months are compared seasonally adjusted figures are used so as to attempt to eliminate differences that might be related to seasonal patterns. The graph shows the unadjusted numbers for total and full-time employment peaking every June and hitting relative low points in January. Notice that the seasonal pattern in part-time employment is counter cyclical to the pattern in full-time; part-time lows are in August. Seasonal adjustment is a complicated calculation designed to remove those fluctuations, but it doesn’t make the resulting number real in the sense of representing the actual number of jobs in any month, it just makes the resulting number useful for some comparisons by removing one or more factors that may regularly cause differences. Job growth is better indicated by the seasonally adjusted series, than by comparing January and June from the unadjusted data. The change between any January and June is a real gain in jobs, but that gain is dominated by seasonal factors rather than underlying long term growth and much of it is lost on the downward part of the seasonal cycle.

Statistics Canada seasonally adjusts data using a method called X-12-ARIMA (autoregressive integrated moving average). A Google search on seasonal adjustment reveals millions of hits. One useful article in plain language is from the Institut de la statistique du Québec; another plain language description of seasonal adjustment is found in an article by Statistics New Zealand. A technical review of the history of seasonal adjustment and its various methods is found in the January 1, 2002 Journal of Business & Statistics’s article "Issues involved with the seasonal adjustment of economic time series". While some methods, such as X-12-ARIMA, may be in common use by major statistical agencies, there is no single correct method and some argue that nothing beats being familiar with the raw, unadjusted, data before deciding whether and what should be done to adjust it.

Of course Clark's jobs plan isn't about the complexities of how to measure and analyze jobs or job growth. Her plan is a political document for the next election. In that context, the unadjusted job numbers are particularly important as they indicate the number of voters with paycheques. The next general election is scheduled for May 14, 2013. Given the trend shown in the graphs displayed here, it is highly likely that employment will be at a record high in May 2013, but the graphs also show that BC was headed for that record long before Clark's plan was announced in September 2011. That won’t stop her from taking credit, just as she could take credit for making it rain. In the unlikely event that the economy goes sour, it is certain that she will not take responsibility for that. An interesting concluding note is that May is a month when jobs are increasing due to the seasonal component, but in May actual unadjusted employment is usually one half of one percent above the seasonally adjusted level, meaning that about 12,000 more people will have jobs than indicated by seasonally adjusted statistics.