Canada

Title of the survey

Survey of Employment, Payrolls and Hours (SEPH)

Organization responsible

Statistics Canada, Labour Division

Periodicity of the survey

Monthly.

Objectives of the survey

To provide monthly estimates of the total number of paid employees, average weekly and hourly earnings and average weekly hours for Canada, the provinces and territories, at detailed industry levels. The results of the survey help to calculate Canada's Gross Domestic Product, to formulate regional economic development policy and manpower retraining policy, assist in selecting sites for industrial relocation or expansion and help in determining escalation clauses in long term contracts.

Main labour topics covered by the survey

Employment, earnings and hours of work.

Reference period

The last seven days of the month. However the questionnaire asks respondents to report for the last pay period in the month. When the reported pay period is longer than one week, the data are reduced to a week using the number of days the firm usually operates during a week divided by the number of working days covered by the reported pay period. Reference week adjustments are also made for strikes and lockouts and layoffs.

Coverage of the survey

Geographical

The whole country.

Industrial

All branches of economic activity, except agriculture, fishing and trapping, private domestic services, religious organizations and defence services.

Establishments

Firms of all sizes and public institutions (government services).

Persons

Employees. Excluded are owners or partners of unincorporated businesses and professional practices, the self-employed, unpaid family workers, persons working outside Canada, and military personnel.

Occupations

Data are not collected by occupation.

Concepts and definitions

Employment

Employees are all persons drawing pay for services rendered or for paid absences and for whom the employer must complete a Revenue Canada T-4 Supplementary Form. They include full-time employees, part-time employees (i.e. those who regularly work fewer hours than the standard workweek of the establishment), working owners, directors, partners and other officers of incorporated businesses. Excluded are casual workers for whom a T-4 form is not required, and persons who did not receive any pay from the employer for the entire survey reference period (e.g. persons on strike, on unpaid holidays, receiving remuneration from an insurance, Workmen's compensation or other related funds). However, employees paid by the employer for a part of the reference period and unemployed or on strike for the rest are counted as employed. Employees are classified into the following categories:

Employees paid by the hour: any employee whose basic wage is expressed as an hourly rate. With the exception of elementary and secondary education, post-secondary and university education, this category of employment represents non-teaching staff: all employees other than teachers and academic staff, i.e. president, vice-president, librarian, registrar, bursar, accountant, guidance counsellor, laboratory technician, personnel officer and other services (e.g. janitor, caretaker, bus driver).
Salaried employees: any employee whose basic remuneration is a fixed amount for at least one week. With the exception of elementary and secondary education, post-secondary and university education, this category of employment represents teaching staff: all those classified as teachers, instructors or academics in the payroll records (e.g. principal, professor, dean, head of department, master, teacher) whether they teach during the day or evening. Supply or substitute teachers are excluded.
Other employees: those employees whose basic remuneration is in the form of commissions, piece rates, mileage allowances, etc. With the exception of elementary and secondary education, post-secondary and university education, this category of employment represents supply or substitute teachers: teachers who are called in to teach whenever their services are needed either to augment the existing staff or as a short-term replacement for a full-time or part-time teacher.

Earnings

Data are collected on the total payroll for each employee category. The payroll includes regular pay (at straight-time), overtime payments, regularly paid commissions, regularly paid bonuses (e.g. production, incentive, isolation, hazard, underground), salaries of working directors, as well as payments for time not worked (e.g. for paid holidays, sick leave including industrial injury and personal leave such as bereavement leave, jury duty). Irregular payments or regular payments that do not relate exclusively to the reference period, such as cumulative payments, vacation pay covering more than the reported pay period, termination pay, quarterly or annual payments, retroactive payments, commission settlements, cost-of-living (COLA) allowances, etc. are collected separately and are adjusted to coincide with the reference period. Excluded are dollar amounts that are taxable allowances and benefits, certain types of non-wage compensation, annual special payments, as well as employer contributions to unemployment insurance, Canada or Quebec pension plans, provincial medical plans, workers' compensation and other welfare plans. Data are also separately collected on overtime pay included in the total payroll, for each category of employees; it includes all pay received for the number of hours worked in excess of the standard workday or workweek.

Wage/salary rates

Not relevant.

Hours of work

Data are collected on total hours paid for (including overtime) and total overtime hours of employees paid by the hour, and hours in the standard workweek of salaried employees. Total hours paid for are all hours worked, including overtime hours, plus hours of paid absence for holidays, sick leave, jury duty, etc. Overtime hours correspond to the total of hours worked for which overtime was paid. Hours in the standard workweek constitute the average number of hours of work normally scheduled in a workweek.

International recommendations

The notion of payroll corresponds to the concept of earnings in cash contained in the international guidelines and used to compile current statistics of earnings. It includes all regularly paid components of earnings and a prorated portion of irregular payments. The value of payments in kind is excluded from the statistics. Two concepts of hours of work are applied: hours paid for, for employees paid by the hour, and normal hours of work for salaried employees.

Classifications

Industrial

The 1980 Standard Industrial Classification (SIC) is used. Each establishment is assigned an industry code on the basis of its principal activity. The SIC is linked to the International Standard Industrial Classification of all economic activities (ISIC), Rev.2, 1968, at the two-digit level (and a few groups at the three-digit level). For analytical purposes, special groupings have been created, such as industrial aggregate, good-producing industries, service-producing industries, durable and non-durable goods manufacturing, commercial and non-commercial services.

Occupational

Not relevant.

Others

The survey data are classified by employee category and geographic region (province or territory), according to the 1986 version of the Standard Geographic Classification.

Sample size and design

Statistical unit

Information is obtained from firms and establishments. Establishment: for statistical purposes, the smallest entity capable of reporting statistics of economic production, typically a factory, mine, store or similar unit. Firm: any business or institution whether incorporated or not and comprises sole proprietorships, partnerships, companies and other forms of organization.

Survey universe / sample frame

The SEPH draws its samples from the Business Register (BR) maintained by the Business Register Division of Statistics Canada and from a list of all payroll deduction accounts maintained by Revenue Canada. The BR is a list of all businesses in Canada and is updated each month using data from various surveys, business profiling and administrative data. The payroll deduction source represents all employers with monthly remittances for employee income taxes, pension plans and unemployment insurance contributions that are less than 15,000 dollars in total.

Sample design

The statistics are based on a combination of surveys and records: a census of establishments of 300 employees and more, a sample survey of medium size establishments employing 100 to 299 employees and a sample of administrative records for employers with less than 99 employees. Establishment survey: To obtain information on large and medium size employers, a sample survey is conducted for about 30,000 establishments from the 100,000 contained in the universe. The universe is stratified by industry, 12 geographic areas (provinces) and four size groups, for a total of 768 strata. All establishments with 300 and more employees (size 4 strata) are included every month (take-all units), while enterprises with less than 300 employees have a known probability of selection (take-some units). The take-some size groups are 0 to 19, 20-49 and 50-299. On average, size 3 strata have a sampling fraction of 80 per cent, size 2, 50 per cent and size 1, about 4 per cent. The total sample of 30,000 units consists of about 20,000 take-all establishments, plus about 9,700 establishments from the take-some population. Each month, the universe and the sample are updated to reflect new establishments created and those that no longer exist. Generally, after 12 months in sample, take-some units are replaced and are not eligible for reselection for at least 12 months. Approximately 1/12 of the take-some sample is replaced every month. Out-of-business establishments detected in the survey are retained in the sample until they normally rotate out. Administrative survey: An arrangement between Revenue Canada and Statistics Canada allows Statistics Canada to access information on the total gross monthly payrolls and the total number of employees for the last pay period of the month from the monthly payroll deduction forms for employers who remit on the PD7 form. This information covers the majority of small businesses which represent about 93 per cent of all employers in Canada. A sample of about 86,000 accounts is systematically selected from the 800,000 PD7 accounts. Each month, Revenue Canada transfers information on the number of employees and the total gross monthly payroll for these accounts. The sample is longitudinal in that it is always made up of the same units except when new businesses are created and out-of-business establishments are identified. No sample rotation takes place. The proportion of small businesses sampled by region is: North (100 per cent), Atlantic (20 per cent), Prairies (20 per cent) and the three largest provinces (10 per cent). The Small Business Payroll Survey (SBPS) provides the basis for the estimation of total hours and the allocation of hours, earnings and employment to categories of employees. Sample allocation is oriented towards maximizing the accuracy of the estimates of total hours and average weekly or hourly earnings. The SBPS draws a small rotating sub-sample of 7,500 selected from those 86,000 payroll deduction accounts which are potentially in business and classified for both industry and province. This sub-sample is used to collect the data for all SEPH variables, including total employees and total payrolls for the month. The administrative sub-sample is stratified using combinations of province and major industry groups. There is no stratification by employment size group. Respondents to the SBPS are divided into three groups, with each group being contacted only four times a year. The Establishment Survey accounts for about 70 per cent of the total payroll employment, while the remaining portion is accounted for by the Administrative Survey. Information for general government services is provided by the Public Institutions Division of Statistics Canada.

Field work

Data collection

A combination of methods is used: mail, telephone, computer reports and administrative records are all employed. For mail units, questionnaires are mailed to the payroll office of employers each month. Telephone interviews are used for respondents who express a preference for being surveyed by telephone. Computer print-outs are supplied by those respondents who prefer to report their data through their computerized payroll system. Administrative information for total gross monthly payrolls and the total number of employees are obtained from payroll deduction accounts maintained by Revenue Canada. The SBPS is conducted monthly by CATI (Computer Assisted Telephone Interviews) collection methodology and collects data on variables not included on the administrative forms. Finally, the Public Institutions Division of Statistics Canada provides information for general government services.

Survey questionnaire

It consists of a bilingual two-page form which collects information for the last pay period(s) of the month, and for each employee category, on:

the beginning and end dates of the last pay period(s),
the number of employees,
their regular gross pay (including overtime pay) and overtime pay,
their hours of work;

In addition, it collects data for the entire month on:

irregular payments, by employee category and period covered,
total payroll for the survey month and since the beginning of the year,
the number of employees absent without pay,
the reasons for large variations in employment, payrolls or hours from the previous month.

The questionnaire is accompanied by a four-page leaflet containing general information on the survey purpose and confidentiality and instructions to fill in the form.

Substitution of sampling units

In case of total non-response, sampling units are not replaced and imputation is used (see below, under "Adjustments").

Data processing and editing

For the Establishment Survey, all questionnaires are screened for completeness by the interviewers of Statistics Canada regional offices. Extensive edit procedures check for accuracy and consistency. As a result of this editing, the respondent may be contacted to confirm or explain the data. The edit procedures include the comparison of the values of key characteristics (e.g. number of employees, hours worked, salary per employee) reported for consecutive months by the same unit to detect large changes or errors in reporting. Whenever there is a large change or the data appear inconsistent, the record is either manually corrected by editors, imputed or confirmed with the respondent. For the Administrative Survey, there are edit and verification procedures at the data capture stage to ensure that the data is of the best quality possible. In addition, on-line edit procedures are in place to check for extreme values. Reporting units which are non-respondents to the initial mailing are followed up by telephone by the staff of the regional offices. If units continue to be non-respondents even after follow-up, then data are imputed for these units. All the records with imputed values are subject to the edits mentioned above. The SEPH programme has a head-office Quality Assurance unit that is mandated to reduce recording, coding and processing errors and ensure that concepts and definitions are respected.

Types of estimates

Employment: totals
Earnings and hours of work: totals and averages.

Total payroll employment estimates are derived by adding up the employment data obtained from the Establishment and Administrative portions of the survey. Average weekly earnings (including overtime) represent gross taxable payrolls divided by the number of employees. It is calculated for the total number of employees, for the hourly-rated employees, for the salaried employees and for the other types of employees. Average weekly overtime earnings are derived by dividing the overtime weekly payrolls by the number of employees. They are calculated for the total number of employees, hourly-rated employees and salaried employees. Average hourly earnings (including overtime) are obtained by dividing the total weekly payrolls by the total weekly number of hours (the total weekly number of hours reported for hourly-rated employees, while, for salaried employees, they are derived by multiplying the average standard workweek hours by the number of salaried employees). Average hourly earnings are calculated for hourly-rated employees and for salaried employees. Average hourly overtime earnings are calculated for hourly-rated employees by dividing the overtime weekly payrolls by the weekly number of overtime hours reported for this group of employees. Average hourly earnings (excluding overtime) are obtained by dividing the weekly payrolls excluding overtime, by the weekly number of hours excluding overtime. This estimate is calculated for hourly-rated employees and salaried employees. Average weekly overtime hours are derived by dividing the weekly number of overtime hours by the number of employees. This average is calculated for hourly-rated employees.

Construction of indices

A set of fixed-weighted average hourly and weekly earnings indices have been constructed to deal with the impact of changes in the paid hours and employment mix between industries, provinces or territories and salaried and hourly employees. This index provides an indicator of the underlying trends in wage rates. Overtime is excluded and the base year for the fixed-weighted earnings is 1986=100.

Weighting of sample results

Estimation of SEPH totals for the Establishment Survey uses the expansion type estimator. When estimates are required for domains or parts of strata, a simple domain estimator is used. Simple random samples are selected independently for each cell (industry group x province x size) and weights are computed excluding outliers detected by the system. Dead units are set to zero. Universe estimation of total is obtained by summing all cells. Estimates for payroll employment and gross monthly payrolls from the Administrative Survey are produced using the expansion estimator, while estimates for the other variables from the SBPS are derived via a regression estimation procedure.

Adjustments

Non-response

In the Establishment Survey: for non-respondent units which have been in the survey for more than one month, data are imputed by calculating a ratio from units that did respond and applying this ratio to the non-respondent's previous month's data. For non-responding units that are new to the survey, averages of the responding businesses in the same industry division, province and size, are substituted for the non-responding businesses. For establishments with more than 300 employees, their previous months' reports are carried forward. In the Administrative Survey: these are four methods of imputation:

imputing zero where the firm has indicated to Revenue Canada that there are no employees in the month;
imputing previous month information with a month-to-month change ratio where there are indications that the units have activity in the current month;
imputing based on current month stratum averages in some cases where only one value has been reported;
reweighting when an in-sample unit is suspected of being late in reporting.

Other bias

Each month, SEPH releases preliminary estimates for the current month, and revised estimates for the previous month. Preliminary data are revised in the following month. On an annual basis, seasonally-adjusted and raw data are revised with the release of the revised data for December for the current and previous three years.

Use of benchmark data

Not relevant.

Seasonal variations

Seasonally adjusted estimates of employment and average weekly earnings, for all employees, for selected industries, Canada, provinces and territories, are produced each month, using the X-11 Variant of the United States Bureau of the Census Method II Seasonal Adjustment Computer Programme.

Indicators of reliability of the estimates

Coverage of the sampling frame

The Establishment Survey frame derived from the Business Register is always several months out-of-date relative to the reference month of the survey. However, some 97 per cent of the units are covered, which represents almost 100 per cent of employment. Two sources of errors associated with the survey frame may occur, which relate to duplication and undercoverage. Duplication occurs on the frame when payroll deduction records are unable to be linked to their associated businesses residing on the frame. Undercoverage arises when new businesses are not classified on the Business Register. Separate estimates are calculated for the "unclassified businesses" for overall employment by province. The Administrative Survey covers Payroll Deduction accounts where the data source for employment and gross monthly payrolls for smaller businesses are available and usable. The Establishment Survey covers establishments where this data source are either not available or not usable. The frame of small businesses used for the Administrative Survey is continuously updated for the most recent new businesses and is never frozen.

Sampling error / sampling variance

The standard error and the relative standard error or coefficient of variation are computed each month. Sample allocation is determined via an approach which initially specified a target coefficient of variation for estimated total employment for each industry division. These targets vary between 0.5 and 1.1 per cent, with more precision being required in industries with greater employment. The target coefficient of variation for estimated total employment for Canada is 0.2 per cent. The standard error for employment averages 4 per cent at the industry division x province level. For average weekly earnings, it averages 2 per cent at the same level.

Non-response rate

At the Canada level, less than 1 per cent of firms surveyed refuse to report data. About 10 per cent of firms report too late to be included in the estimation process at all. Late reporters tend to be large firms and those using the computer reporting service for data collection. Administrative data are received for about 60 per cent of the PD accounts; an additional 20 per cent are businesses without any employment or payrolls.

Non-sampling errors

SEPH has a statistical quality control programme at the data capture, business structure updating and data editing stages to minimize non-sampling errors. These programmes monitor and control the completeness, accuracy and consistency of the reported survey data. Procedures are in place to follow-up for non-response and to impute for those late responses.

Conformity with other sources

SEPH payroll employment estimates are frequently compared to the employment estimates provided by the Labour Force Survey (LFS). Some differences can be quantified (e.g. addition or subtraction of groups of workers), and others not (e.g. errors in SIC coding, differences in reference periods). A reconciliation process is also conducted between SEPH data and public administration data to ensure that they are compatible.

Available series

The following tables are published each month:

seasonally adjusted estimates of employment and average weekly earnings, for all employees, for selected industries, Canada, provinces and territories;
fixed-weight indexes (1986=100) of average hourly earnings, for all employees, for selected industries, Canada, provinces and territories;
number of industries showing year-over-year increases in employment, for selected industries, Canada;
estimates of employment, payrolls and average weekly earnings, for all employees, by industry, Canada, provinces and territories;
estimates of employment, average hourly earnings and hours, for employees paid by the hour, and for salaried employees, by industry, Canada, provinces and territories;
historical estimates of employment and average weekly earnings, for all employees, industrial aggregate, Canada, provinces and territories.

History of the survey

The SEPH was introduced in 1983. From time to time, a historical revision of the estimates is necessary for changes related to new data sources and revised industry classifications, and frame changes. Historical revisions covering the period January 1983 to December 1991 were produced in 1992, which included:

movement to the Business Register as the SEPH survey frame in 1987,
conversion to a "new" redesigned and updated Business Register in October 1990,
replacement of the 1970 SIC with the 1980 SIC resulting in a more accurate and up-to-date industry classification.

Documentation

Statistics Canada: Employment, Earnings and Hours - Catalogue No. 72-002 (monthly; Ottawa). idem: Canadian Economic Observer (monthly; ibid.) In addition to publications, both standard and special tabulations are offered. Data are available on CD, diskette, computer printouts, microfiche and microfilm, and magnetic tapes. Direct on-line access to aggregated information is possible through CANSIM, Statistics Canada's machine readable database and retrieval system. Information on services to the public is available in "Employment, Earnings and Hours".

Confidentiality / Reliability criteria

Data are collected under the authority of the Statistics Act, Revised Statutes of Canada, 1985, Chapter S19. Any aggregated employment estimate that could reveal information about a specific respondent is declared confidential and consequently not published.

Other information

Data supplied to the ILO for publication

The following statistical series are published in the Year Book of Labour Statistics:

Annual average of paid employment (all employees) - general level, non-agricultural activities and specific industries - in Tables 3A, and 4 to 8;
Average hours paid for of hourly paid employees, in Tables 11 to 15;
Average hourly earnings of hourly paid employees and average weekly earnings of employees, in Tables 16 to 21.

The corresponding monthly series are published in Tables 3 (paid employment in manufacturing), 5, 6, 7 and 8 (respectively, hours paid for and average hourly and weekly earnings in non-agricultural activities and in manufacturing) of the Bulletin of Labour Statistics.