Survey of Employment, Payrolls and Hours (SEPH)
The Survey of Employment, Payrolls and Hours provides a monthly portrait of the amount of earnings as well as the number of jobs (i.e., occupied positions), vacant positions, and hours worked, by detailed industry, at the national, provincial and territorial levels.
Detailed information for December 2014
Data release - February 26, 2015
- Questionnaire(s) and reporting guide(s)
- Data sources and methodology
- Data accuracy
The Survey of Employment, Payrolls and Hours (SEPH) provides a monthly portrait of the amount of earnings, as well as the number of jobs (i.e., occupied positions), vacant positions, and hours worked, by detailed industry, at the national, provincial and territorial levels.
SEPH data provide the principal input to labour income estimates: they also serve as a proxy output measure for about 15% of real gross domestic product and 'nominal' gross domestic product. SEPH data are also used by the Canada Revenue Agency (CRA), to revise the maximum pensionable earnings and retirement savings plan contribution limits, and by the private sector, for contract escalations and wage rate determinations.
Monthly survey estimates are produced by a combination of a census of payroll deductions, provided by the Canada Revenue Agency, and the Business Payrolls Survey (BPS), which collects data from a sample of 15,000 businesses. The BPS also collects information about job vacancies, supplemental to the SEPH (see Job Vacancy Statistics, record number 5202).
Together, four surveys tell a more complete story of current labour market events. These surveys are: the Labour Force Survey (LFS), the Survey of Employment, Payrolls and Hours (SEPH) Employment Insurance Statistics (EIS) and Job Vacancy Statistics (JVS). The LFS focuses on its strengths: timely data on the labour market, including the unemployment rate and demographic analysis. SEPH reports, which come out later each month, show greater detail on non-farm industry employment and earnings. EIS provides substantial detail on recipients of EI regular benefits by detailed geography, by socio-demographics and by former occupation. JVS offers information on labour demand by reporting on the number of job vacancies by industry.
- Employment and unemployment
- Hours of work and work arrangements
- Wages, salaries and other earnings
Data sources and methodology
The program's target population is composed of all businesses in Canada that have at least one employee and, thus issued at least one payroll deduction remittance during the reference month. Excluded are businesses that are primarily involved in agriculture, fishing and trapping, private household services, religious organizations, international and other extraterritorial public administration and military personnel of defence services.
The BPS has a regular questionnaire and a questionnaire adapted to educational services. Both exist as paper forms, forms for computer-assisted telephone interview, and as electronic questionnaires, which were introduced in December 2012.
The BPS questionnaire was developed and tested by a multi-disciplinary team made of participants from Labour Statistics Division, survey operations, systems, and methodology. The questionnaire was prescribed, according to Section 7 of the Statistics Act, and approved by the Chief Statistician. Data collection for the BPS is fully integrated into the Business Collection Portal of Statistics Canada.
Two of the variables of interest are collected by census method: total monthly payroll employment and monthly payrolls are extracted monthly from the Canada Revenue Agency Payroll Deduction (PD) administrative source. The payroll deduction source includes all employers with remittances for employee income taxes, CPP/QPP, and Employment Insurance contributions.
Payroll Deduction data are spread across provinces and industry using the enterprise profile in the Statistics Canada Business Register. The Business Register (BR), is a list of all businesses in Canada, updated each month by Statistics Canada, using data from various surveys, business profiling and administrative data.
Other key variables, including number of hours worked, are collected by the Business Payrolls Survey (BPS) which collects monthly data directly from a sample of establishments drawn from the BR. The initial BPS sample is made up of 15,000 establishments out of a population of 900,000 establishments. The sample is stratified according to geography (province), industry (North American Industry Classification System (NAICS), level 3 or 4) and number of employees in the establishments. Some units are permanently in the sample; most remain in the sample for one year only. Every month, one-twelfth of the sample is refreshed. Once removed from the sample, most units remain out of sample for at least one year.
Responding to this survey is mandatory.
Data are collected directly from survey respondents, extracted from administrative files and derived from other Statistics Canada surveys and/or other sources.
The statistics compiled by SEPH are based on a census of administrative records for all in-scope establishments with employees that can be found on the Business Register. The total payroll employment estimates and the monthly payrolls are derived from the administrative source. Administrative information for total gross monthly payrolls and the total number of employees for the last pay period in the month are obtained from payroll deduction (PD) accounts maintained by Canada Revenue Agency. Public Sector Statistics Division of Statistics Canada provides information for general government services at the provincial and federal levels (see record number 1713).
The Business Payrolls Survey (BPS) conducted monthly, is used to collect SEPH variables not available from administrative sources. The BPS uses a combination of methods for data collection to permit maximum flexibility for respondents. For Internet data collection, invitations and access codes are emailed each month to employers' payroll offices. For respondents using mail, questionnaires are mailed to the employer's payroll office. For respondents who prefer to be surveyed by telephone, computer-assisted telephone interviews (CATI) are used. Respondents can also send reports electronically. Reporting units that do not respond to the initial contact are telephoned by staff of Statistics Canada's regional offices.
View the Questionnaire(s) and reporting guide(s).
For the administrative portion of the survey, edits and verification procedures at the data capture stage ensure that the data is of the best quality possible. Significant and unusual monthly movements at the micro level are identified using the Hidiroglou-Berthelot method, and are manually corrected, if appropriate.
Both manual and automated editing procedures are employed to detect and correct problematic data provided by the respondent on the BPS questionnaire. Historical edits (weighted and unweighted) are performed at the data collection stage.
For the administrative portion of the survey, there are four methods of imputation:
- impute to zero - set both employment and pay to zero when information from the CRA does not indicate employment activity in the month;
- trend imputation - apply month-to-month movement from non-imputed records with characteristics similar to the prior month's data;
- ratio imputation-imputing by using the relationship among different variables from non-imputed records with similar characteristics;
- mean imputation-imputing by adopting the stratum averages.
For the BPS portion, only units that are permanently in sample are imputed using historical data. Imputation avoids respondent follow-up while using as much respondent-provided data as possible. Reweighting is employed to correct for all other missing establishments.
The estimation of population characteristics from a survey is based on the premise that each sampled unit represents, in addition to itself, a certain number of unsampled units in the population. A basic survey weight is attached to each record to indicate the number of units in the population that are represented by that unit in the sample.
Two adjustments are applied to the basic BPS weights to improve the reliability of the estimates. These basic BPS weights are first inflated to compensate for non-response. The non-response adjusted weights are then calibrated to ensure that estimates of total monthly payroll employment and monthly payrolls respect estimates from the Canada Revenue Agency Payroll Deduction (PD7) administrative source.
The calibration is done using a generalized regression estimator. The model groups are mostly defined at the national and sub-sector levels (i.e., three-digit North American Industry Classification System (NAICS) code or, in a few instances, four-digit); in a few cases, the enterprise size (employment) and the provincial dimensions are used. Regression coefficients, calculated at the model group level, are applied to the estimates of total employment and payrolls from the administrative sources to estimate the additional variables.
The information obtained from the BPS is used to estimate the weekly component of the gross monthly payrolls, the total number of paid hours (regular hours and overtime) and the allocation of hours, earnings and employment for three categories of employees: salaried, paid by the hour and others, such as commission workers.
Non-farm payroll employment data are for all hourly and salaried employees, as well as the 'other employees' category, which includes piece-rate and commission-only employees.
Average weekly hours data are for hourly and salaried employees only. They exclude businesses that could not be classified to a NAICS code by the time monthly processing was completed.
All earnings data include overtime pay, and exclude businesses that could not be classified to a NAICS code. Earnings data are based on gross taxable payroll before source deductions.
Average weekly earnings are derived by dividing total weekly earnings by the total number of employees.
Data collection for the BPS is monitored for changes in response rates and failed edit rates.
Coefficients of variation (CV) are analyzed every month to identify the domains having the least accurate estimates. Sampling fractions are adjusted occasionally, to obtain comparable coefficients of variation across domains.
A micro-match is performed every month to compare the BPS data with the administrative source data for employment and payrolls. Large differences are looked at and corrected if necessary.
Prior to the release, comparisons with independent sources such as the Labour Force Survey are performed.
Statistics Canada is prohibited by law from releasing any information it collects which could identify any person, business, or organization, unless consent has been given by the respondent or as permitted by the Statistics Act. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.
Cell suppression via the CONFID software is used to control disclosure of the data.
Revisions and seasonal adjustment
With each release, data are revised for the previous month to take into account late remitters or additional information from BPS respondents. Users are encouraged to request and use the most up-to-date data for each month.
Each year, additional revision processes are done at the same time as the December monthly revision:
- Seasonally adjusted data are revised back three years.
- Annual revisions to fine-tune some estimates of the previous 12 months.
- Occasionally, historical revisions are done to introduce changes related to concepts, new data sources, revised industrial or geographical classifications, as well as methodology.
For the BPS portion of the survey, response rates are produced every month. The total response rate for Canada as a whole usually varies between 80% and 90%.
Every month, SEPH coefficients of variation (CV) are produced for all variables and every domain. These CVs take into account the sampling variance coming from the BPS as well as the variance due to imputation of the administrative source.
The CVs are usually very low - less than 5% - for the administrative data component of SEPH (e.g., monthly number of employees and gross payroll). The coefficients are higher for those associated with the BPS (e.g., average weekly earnings, average weekly hours). Quality indicators are included in the CANSIM tables for both surveys published regularly by Statistics Canada.
- Use of SEPH earnings data for contract escalation
- Date modified: