Section 8: Comparing the Survey of Employment, Payrolls and Hours and the Labour Force Survey
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Statistics Canada has two monthly surveys that measure employment levels and trends: the Labour Force Survey (LFS), referred to here as the household survey, and the Survey of Employment, Payrolls and Hours (SEPH), referred to here as the payroll or establishment survey.
The LFS provides the first timely picture of overall labour market conditions, with total employment, unemployment and unemployment rate, as well as information on which groups of Canadians are most affected by changes in the labour market.
Approximately two months later, the SEPH provides additional detail of the same month by industry, along with estimations of earnings and hours worked.
Statistics from both the household and business payroll surveys, along with those from the Employment Insurance program and the Job Vacancy program all contribute to understanding the supply and demand components of the labour market.
The household survey (LFS) provides a broader picture of employment, including employment in agriculture and the number of self-employed. The payroll survey (SEPH) provides a highly reliable gauge of monthly change in non-farm payroll employment.
Because the LFS has a broader definition of employment than the SEPH, the LFS employment level exceeds the SEPH employment level. See Chart 8.1 below for more details.
For comparability purposes, an adjusted LFS series was added to Chart 8.1 to be more similar in concept and definition to SEPH employment. This adjusted series is created by subtracting agriculture, fishing and hunting employment, non-agricultural self-employed, unpaid family and private household workers, as well as workers absent without pay from their jobs, and then adding non-agricultural wage and salary multiple jobholders.
The LFS adjusted series tracks much more closely with the SEPH measure; nonetheless, trend discrepancies occur occasionally. For example, during the 2008–2009 employment downturn, the decline was more pronounced in the LFS than the SEPH.
Some of the discrepancies are attributable to conceptual and measurement differences between the two surveys. The major features and distinctions of the two surveys are shown in Table 8.1.
Comparing employment trends from the two surveys
The LFS is the only survey conducted by Statistics Canada designed to provide the official unemployment rate every month, with a monthly sample size of approximately 56,000 households. It is the earliest and most timely indicator of the pulse of the labour market in Canada. The sample size makes it a very reliable source for different geographic levels. It provides a complete picture as it includes employees, self-employed people, as well as unemployed. Characteristics of all three groups, including age, sex and occupation are also available.
The SEPH, also a monthly survey, is designed to provide data on payroll employment as well as average earnings and hours worked. It is a census of all payroll employees in Canada. Consequently, the SEPH does not survey the self-employed or the unemployed. However, the employment payrolls data are available at a detailed industry level.
As mentioned earlier, SEPH and LFS estimates track well over the long-term, but discrepancies in trends occur occasionally.These discrepancies might be more common for sub-groups, for example, at the provincial level or in a specific industry.
Table 8.2 provides year-over-year employment change from the two sources of data.
Sampling error in the LFS
The LFS is subject to sampling and non-sampling error. While it is one of the largest sample surveys internationally, reaching 56,000 households, it covers a small portion of all employed persons. When looking at short-term trends, especially over-the-month changes, it is therefore essential to assess the statistical significance of the change.
Employment estimates from the SEPH, as they are derived from all the administrative payroll deduction forms submitted by employers to the Canada Revenue Agency, are not subject to sampling error. They are, however, subject to non-sampling error.
Worker classification in the LFS
An adjusted LFS employment series was created to be more similar in concept and definition to the SEPH employment series, as featured in Chart 8.1. This adjusted series is calculated by subtracting agriculture and related employment; unpaid family and private household workers; workers absent without pay from their job; as well as military reservists; and adding wage and salary multiple jobholders whose second job is either as an employee or as an incorporated self-employed with employees. For the SEPH, employment data from the three northern territories are removed to make it comparable to the LFS.
This adjustment process also has some limitations. For example, some independent contractors in LFS may have mistakenly reported themselves as wage and salary workers rather than self-employed. This could lead to some overstatement in the adjusted LFS employment estimates. Separately, the adjustment for multiple jobholders adds the number of workers whose primary job is non-agricultural wage and salary, but not necessarily their secondary job. However, there are no adjustments to account for the number of multiple jobholders with three or more jobs; the adjustment process presumes all multiple jobholders only hold two jobs. This introduces some understatement into the adjusted household survey employment. These types of worker classification issues limit the ability to fully reconcile the two employment measures.
‘Off the books’ employment
Workers may be working for pay but not necessarily declared in the SEPH (for example, ‘off-the-books’ employment). The LFS could possibly include some of these workers, but it is not possible to determine the extent to which they might be reflected in the survey estimates.
Estimates by province
Estimates from the LFS are based on where people usually reside. However, the SEPH counts employees in the province or territory where they work. This does not affect comparability at the national level, but can create differences at the provincial/territorial level.
Payroll estimates and detailed industry data
New businesses are included in the Statistics Canada Business Register, which is used by the SEPH to obtain the industry information. However, there can be delays before all new businesses are classified, which can mean temporary under-reporting in some of the detailed industry data. Also, the Business Register regularly reviews and updates the industry and provincial classification of current establishments. This can cause changes to employment levels of updated industries, not reflecting real growth or decline.
Employment industry data available from the household survey are based on information provided by survey respondents when asked about the detailed characteristics of their employment. That information is then processed and an industrial classification code is assigned by Statistics Canada.
Release schedule differences
The LFS interviews take place over the ten days following the reference week. This is followed by nine days of processing and analysis, enabling the release of the estimates 19 days after the Saturday of the reference week. The release usually takes place the first or second Friday of the month.
Businesses have until the 15th of the following month to file data from the last pay period of the reference month to the Canada Revenue Agency. These data, approximately one million records, are provided to Statistics Canada at the beginning of the next month, or five weeks after the reference period. This is followed by three and a half weeks of processing and analysis, bringing the release to eight and a half weeks after the reference period.
Estimates from both surveys are revised according to different schedules. While this does not impact the overall trends, it could affect the month-to-month change.
The seasonally adjusted LFS estimates are revised yearly, going back three years and are published around the end of January. Every five years, population controls are updated according to the latest census population projections and all LFS estimates are revised historically over a longer time span. This exercise is called a rebasing and the last one occurred in January 2015. At the time of the rebasing, the latest classifications for geography, industry and occupation are updated along with the latest seasonal factors.
With the SEPH, monthly estimates are revised the month after the estimates are first publicly released. For example, when estimates for May are released, estimates for April are revised.
Every year at the end of March, the SEPH estimates undergo a historical revision. The span and the breadth of the revisions vary depending on the year. The revisions to specific industries sometimes go back as far as 2001 and can include updates to new classification systems (i.e., moving from NAICS 2007 to NAICS 2012), or sometimes will span only a few years with minimal changes. At the same time, seasonally adjusted data are revised back three years.