The 2011 revisions of the Labour Force Survey (LFS)
The Labour Force Survey (LFS) provides estimates of employment and unemployment which are among the most timely and important measures of performance of the Canadian economy. With the release of the survey results only 13 days after the completion of data collection, the LFS estimates are the first of the major monthly economic data series to be released.
Statistics Canada has an established history of applying a standard revision to its LFS estimates following the release of final population estimates from each census. Along with this revision, other changes have also been introduced. It should be noted that these changes do not involve modifications to the questionnaire or content of the survey. The purpose of this document is to explain each of these changes as well as a few other changes made to estimates.
In brief, starting with the release of LFS data for January 2011 (in February 2011), estimates will be based on the 2006 Census population counts; sub-provincial estimates will be based on 2006 Census boundaries; industry estimates will be based on North American Industry Classification System 2007; and occupation estimates will be based on National Occupational Classification – Statistics 2006. Specifically:
- Population rebasing: Until December 2010, labour force estimates had been based on population data from the 2001 Census. As of January 2011, the estimates have been adjusted to reflect population data from the 2006 Census. These revisions have had some minor impacts on the LFS estimates, while rates of unemployment, employment and labour force participation were essentially unchanged. Because the changes to the estimates were minimal, revisions were necessary only back to 1996.
- Geography boundaries: Census metropolitan areas (CMAs) are now based on 2006 census boundaries rather than the 2001 boundaries. With this change, six new CMAs have been added: Moncton, New Brunswick; Peterborough, Brantford, Barrie and Guelph in Ontario; and Kelowna, British Columbia. At the same time, the boundaries of seven CMAs were modified. Boundaries for economic regions and employment insurance regions remain unchanged.
- Industry and occupation: The LFS has also moved to more recent classification structures for industry and occupation data. These are updates from the current North American Industry Classification System 2002 (NAICS 2002) to NAICS 2007 and from the National Occupational Classification – Statistics 2001 (NOC-S 2001) to the NOC-S 2006. Changes to NAICS 2007 are minimal, and revisions were extended to January 1987. Changes to the NOC-S 2006 had no impact on the historical estimates.
2.0 Population rebasing
The LFS uses estimates of the target population, which are derived independently from the survey, as benchmarks for producing survey estimates. These population estimates start with a Census base and are then updated using administrative data between censuses to reflect the current population of Canada. Using these population counts reduces the sampling variability and coverage bias of survey estimates. Proper population numbers are crucial in determining estimates from a sample survey like the LFS. In order to translate the results of the sample into estimates, each individual in the sample is assigned a weight indicating the number of persons in the population that individual represents.
The Census base used for obtaining these estimates is updated several years after each new Census is conducted. Beginning with the release of the January 2011 survey, population estimates used by the LFS will change from a 2001 Census base to a 2006 Census base.
These updated population counts result in more accurate labour force estimates than can be expected when using those with a 2001 Census base. As the population estimates move away from their original Census base over the years, inadequacies in the administrative data used to update the numbers become more pronounced. For example, in December 2010, the 2001-based estimate of the target population is 0.3% higher (+79,000) than the 2006-based estimate. This means that population estimates were overestimated for that survey month. The differences in the two sets of population estimates can be more pronounced for some age and sex groups and sub-provincial areas.
The LFS uses population counts that include an adjustment for net Census undercoverage. In any census, there is both overcoverage and undercoverage: some people are counted more than once or should not be counted, while others are not counted but they should be. The net result is usually undercoverage, which was approximately 3% in the 2006 Census.
Since the difference between the old estimates (based on the 2001 Census) and the new estimates (based on the 2006 Census) was relatively small, a historical revision back to the start of the series was deemed unnecessary. The year 1996 was chosen based on the sub-provincial estimates, specifically census metropolitan areas (CMAs) and census agglomerations (CAs). Since CMA/CA boundaries have changed extensively beyond the past fifteen years, and since forcing comparability would affect data quality, it was decided to revise these series back to 1996.
3.0 Impact on estimates
3.1 Changes in population counts
Estimates of the population have been revised from January 1996 to December 2010. In general, estimates have been revised downward and the magnitude of the revision increases over the period.
At the national level, for the working-age population (aged 15 and over), the differences are negligible between January 1996 and December 2001 (Chart 1). Over that period, the new estimates are lower by 10,000 or less (-0.04%) than the old population count. Beginning in January 2002 to April 2005, differences are slightly higher, with changes generally ranging from -10,000 to -20,000 (or from -0.04% to -0.08%). From May 2005 to December 2010, the differences peak to about -0.3% (-79,000). This difference is about half the size of the previous rebasing in 2005, when LFS estimates were rebased from the 1996 to 2001 Census population counts (-0.7% or -170,000).
From 2007 to 2009, the difference between the revised and unrevised population estimates narrowed slightly before growing apart again. This is due to the fact that during this period, the number of non-permanent residents had been underestimated in the unrevised estimates and the difference was added back to the revised estimates. By adding this difference, the revised levels were brought closer to unrevised levels. From 2009 on, both the revised and unrevised population counts had similar levels of non-permanent residents, so the gap increased again.
As a result of the revision to the population estimates, the levels of employment and unemployment were revised downward and the estimates for those not in the labour force were revised slightly higher. When evaluating the data, it is important to keep in mind that some age groups among men and women have been affected differently by the population revision.
As can be seen from Chart 2, population estimates for both men and women of working age were revised downward, but more so for women than men. Chart 3 shows that population estimates for youths and those aged 55 and over were affected differently by the revision than for men and women aged 25 to 54 years. By December 2010, the new population count was higher by 1.1% for youths and 0.5% for those aged 55 and over, and lower by 1.1% for men and women aged 25 to 54. Because these groups also have different labour market characteristics (for example, a higher proportion of 25 to 54 year-olds are employed) the relationship between the new and the old estimates can be complex.
3.2 Overall impact on labour market estimates
Charts 5, 6 and 7 present monthly, seasonally adjusted revised and unrevised estimates of labour force characteristics at the national level by age and sex, while Tables 1 and 2 include differences based on annual averages for selected years.
As previously mentioned, employment and unemployment were revised downwards with the new population counts. This occurs for the total working-age population as well as for men and women aged 25 to 54, but slightly more so for women than men.
The revised estimates for persons aged 15 to 24 years show higher levels of employment, unemployment and participation in recent years. Despite these higher levels, the annual rates of employment and participation are the same between the revised and unrevised in 2010, while the revised unemployment rate is 0.1 percentage points higher than the unrevised (Table 1).
Similar to youth, the revised population estimates for those aged 55 and over are also slightly higher than the unrevised and had little to no impact on the rates of unemployment, participation and employment.
For half of the provinces, revised employment estimates were lower than the unrevised, with a difference of less than 1% (Tables 3, 4 and Charts 8, 9). Differences were beyond 1% in the other half of the provinces, including Newfoundland and Labrador, Prince Edward Island, New Brunswick, Alberta and British Columbia. In 2010, employment levels were revised downward by 1% or more for New Brunswick (-2.3%); British Columbia (-2.1%); Newfoundland and Labrador (-1.5%) and Prince Edward Island (-1.0%) while estimates for Alberta were revised upward, by 1.0%.
3.3 View of the labour market
When conducting an historical review of survey estimates, attention is focused on whether labour market trends have been altered. As can be seen from the employment chart in Chart 5, the new estimates of employment are slightly lower overall due to the lower population estimates. Despite the slightly lower levels, however, the new (revised) and old (unrevised) employment estimates track closely, following the same trends.
Between 1996 and 2010, both employment and population grew at a slightly slower pace for the revised versus unrevised levels. As a result, during this period, the revised employment rate increased by 3.1 percentage points, slightly lower than the 3.4 points increase in the unrevised rate.
Table 5 shows that the major areas of growth and the share of employment by age and sex, industry, class of worker, education, and immigrant status for the revised versus unrevised estimates are virtually unchanged from 1996 to 2010 (2006 to 2010 for immigrant status).
As mentioned earlier in the text, the further the population estimates move away from their original Census base in time, the more pronounced the changes. With the labour market downturn occurring in the past few years, the new and old estimates were reviewed to see if any of the milestones or trends changed during this period.
As with the old series, the new employment estimates also show that the downturn began in October 2008 and continued until July 2009 (Chart 4). During this period, the revised employment estimates declined slightly more (-2.5% or -428,000) than the unrevised estimates (-2.4%, or -417,000).
The employment increase from July 2009 to December 2010 based on the new estimates was not as strong as with the old estimates. With the new estimates, employment grew by 2.4% (+398,000) versus 2.8% (+463,000) with the old estimates. This difference is mostly due to less population growth among those aged 25 to 54 years in the revised estimates, the group with the highest employment rate.
Compared to the peak of October 2008, revised employment in December 2010 was down 30,000 (-0.2%). The unrevised estimate was an increase of 46,000 for the same period (+0.3%).
The new employment estimates also show that growth was slightly slower than the old estimates over the year 2010. From December 2009 to December 2010, the revised employment estimates posted a growth rate of 1.8% (+298,000) versus 2.2% (+369,000) with the old estimates. The unemployment rate was little changed between the new and old estimates over the year. The new estimates show a decline of 0.9 percentage points in the unemployment rate to 7.6% and the old estimates show a decline of 0.8 percentage points (also to 7.6%).
The national revised rates of employment, unemployment and participation followed the same trend line in the past few years as the unrevised. While the unemployment rates were generally the same, the new employment and participation rates were slightly lower than the old rates, by 0.1 to 0.3 percentage points in more recent years. This is due to slower population growth reflected in the new estimates.
Estimates for Yukon, Northwest Territories and Nunavut were also adjusted to reflect population counts based on the 2006 Census. Estimates were revised back to 1996 for Yukon; 2001 for Northwest Territories and 2004 for Nunavut. Boundaries for each territory remain unchanged since the start of each series.
Population and labour force estimates for the three territories were revised upward as a result of this rebasing. Employment estimates were revised up by 5.4% for Yukon; 1.9% for the Northwest Territories and 11.4% for Nunavut in 2010 (Table 7). Unemployment levels were little changed between the new and old estimates, however, the revised unemployment rate for Yukon and Nunavut was 0.4 percentage points lower than the old rate and 0.2 percentage points lower for the Northwest Territories in 2010.
There were also a few sampling methodology changes in the northern territories, beginning in January 2011. These included updates to the sample allocation and sampling rates based on the latest population estimates; expanded use of one-stage sample designs; and new cluster-labelling conventions. Further, in the Yukon, there was a re-stratification of outlying communities and the village of Pelly Crossing was added to the coverage. The goals of these changes were to achieve sample sizes closer to original targets; to distribute response burden more equitably, and to improve the efficiency of the overall design.
Industry and occupation estimates for the three territories were updated to NAICS 2007 and NOC-S 2006 but from older versions for Yukon and Northwest Territories – from NAICS 1997 and SOC 1991.
3.5 Aboriginal estimates
The weights applied to the Aboriginal working-age population were also updated to reflect population changes. As was announced in the Daily release of data on Aboriginal peoples from the 2006 Census on January 15, 2008, the Aboriginal population increased at a faster pace than the non-Aboriginal population from the 2001 to 2006 Census, nearly six times faster. Several factors account for this growth of the Aboriginal population: demographic factors, such as high birth rates; more individuals identifying themselves as an Aboriginal person; and a decrease in the number of incompletely enumerated Indian reserves.
The revised LFS working-age population estimate for the Aboriginal population in 2010 was 20% higher (+110,000) than the old estimate. Growth was particularly strong among Métis and North American Indians and in Quebec, Ontario and the Atlantic region (particularly Nova Scotia) (Table 7).
As with most other population groups, there was little change in the revised and unrevised unemployment rates for the Aboriginal population. However, the revised employment rate was 0.8 percentage points lower than the old rate, as employment levels grew at a slower pace than the population.
4.0 Industry and occupation changes
The reclassification changes from the North American Industrial Classification System (NAICS) 2002 to 2007 were minor and had little impact on the estimates. For this reason, revisions were applied to estimates going back to 1987. Most of the published industry tables start in 1987, so there will be minimal changes in these tables. For the few tables with historical series going back to 1976, estimates will be based on NAICS 2002 prior to 1986 and NAICS 2007 starting in 1987.
Most of the changes were at the four-digit code level. One four-digit industry was moved from one major industry group to another: from professional, scientific and technical services (NAICS code 54) to business, building and other support services (NAICS codes 55-56). There were also a few title changes. The concordance table for the title and code changes is in Appendix I.
The specific NAICS code changes are as follows: some records from 5416 were transferred to 5613 to create a new six-digit category for human resources placement agencies. Embroidery contractors were removed from 3152 and a new category was created in 3149. Boat building was removed from 3261 and placed into 3366. Code 3391 was redesigned, with some records moved to 3332, 3334, 3339 and 3371. Some records were moved from 4163 to 4143 to reclassify wholesalers of household cutlery and pots and pans. New NAICS codes were also created in 5171, 5179, 5191 and 5311.
The structure of the National Occupational Classification - Statistics (NOC-S) 2006 remains unchanged from that of NOC-S 2001. No major groups, minor groups or unit groups were added, deleted or combined, though some unit groups have new names or updated content. Since LFS occupation tables are at the major and minor group level, there are no changes to estimates or titles. Appendix I provides the title changes at the unit group level.
5.0 Geography boundary changes
Geography boundaries have also been updated in this revision, moving to the 2006 Standard Geographical Classification (SGC) from the 2001 SGC. With the change to the 2006 Census boundaries, six new census metropolitan areas (CMAs) have been added: Moncton, New Brunswick; Peterborough, Ontario; Brantford, Ontario; Barrie, Ontario; Guelph, Ontario; and Kelowna, British Columbia. At the same time, the boundaries of seven CMAs have been modified: Québec City, Quebec; Sherbrooke, Quebec; Montréal, Quebec; Ottawa-Gatineau, Quebec part; Ottawa-Gatineau, Ontario/Quebec; London, Ontario; and Winnipeg, Manitoba.
New CMA tables have been created based on the 2006 Census boundaries and will date back to 1996 (CANSIM tables 282-0109, 282-0110, 282-0111, 282-0112, 282-0113, 282-0114, 282-0116 and 282-0117). Since boundaries of CMAs and census agglomeration (CA) have changed extensively beyond the past fifteen years, and since forcing comparability would affect data quality, it was decided to revise these series back to January 1996. Boundaries for economic regions remain unchanged, with published tables dating back to January 1987.
Historical comparisons dating prior to 1996 are still possible by CMA. Estimates based on the 2001 Census boundaries and population counts will continue to be available on CANSIM with the original table numbers (CANSIM tables 282-0052, 282-0053, 282-0056, 282-0057, 282-0058, 282-0059, 282-0090 and 282-0091). For comparisons before 1996, please use the concordance table in the CANSIM What's new announcement from January 28, 2011.
A new annual table on census agglomerations (CAs) based on the 2006 Census boundaries is now available, replacing the selected medium-sized cities table based on the 2001 Census boundaries (CANSIM table 282-0064). This new table (CANSIM table 282-0115) includes nine new CAs: Truro, Nova Scotia; Saint-Hyacinthe, Quebec; Leamington, Ontario; Timmins, Ontario; Brockville, Ontario; Grande Prairie, Alberta; Wood Buffalo, Alberta; Duncan, British Columbia; and Courtenay, British Columbia.
6.0 New seasonally adjusted series
Several new seasonally adjusted series have been added from the start of the series:
- Employment full and part time, for those aged 15 to 64, and 25 to 54 by sex and province;
- Labour force characteristics for those aged 55 and over by sex and province;
- Employment full and part time, for those aged 55 and over by province;
- Six new CMAs: Moncton, New Brunswick; Peterborough, Ontario; Brantford, Ontario; Barrie, Ontario; Guelph, Ontario; and Kelowna, British Columbia.
With these new series being seasonally adjusted, and new raking rules applied for the labour force characteristics series by age group, sex and province, all seasonally adjusted series have been revised back to the start of the series.
As part of our 2011 annual review of seasonally adjusted data, the educational services employment series was examined in light of changes to the seasonal pattern of this industry's employment levels during the past four summers. With five straight years of an unclear pattern in summer employment, a model without outlier identification was determined to be more appropriate for this data series and has been implemented at the same time as the revision of population estimates. As can be seen from the education services chart (Chart 11), the July and August changes over the last four years are not as pronounced in the revised estimates as they are in the unrevised estimates.
- Date modified: