Survey on Financing and Growth of Small and Medium Enterprises

Detailed information for 2017

Status:

Active

Frequency:

Occasional

Record number:

2941

The survey is designed to find out what kinds of financing small and medium enterprises are using, and to collect information on recent attempts to obtain new financing. The survey also collects information on barriers to growth, the financial position of small and medium enterprises, the characteristics of ownership and the extent to which the enterprise is involved in innovation and intellectual property.

Data release - November 16, 2018

Description

The objective of this survey is to collect general characteristics on small- and medium-sized businesses and their financing activities. It collects information on the types of debt, lease and equity financing that small and medium enterprises (SMEs) rely on. Furthermore, it collects information on any recent attempts to obtain new financing. It also collects additional information about circumstances that affect the way these businesses operate.

Statistics Canada is conducting this survey on behalf of a consortium led by Innovation, Science and Economic Development Canada. The data obtained from this survey will be used by both the public and private sectors. Innovation, Science and Economic Development Canada will use this information to study the availability of financing to SMEs and to recommend policy changes to assist businesses. Other government departments will use this information to develop national and regional programs and for policy planning. Businesses will use this information for market analysis or to compare the performance of their firm with the performance of firms of a similar size within the same industry. Industry associations will use the information for industry performance measurement and for market development, and suppliers of financing for SMEs will use the information to determine gaps in their services.

Subjects

  • Business performance and ownership
  • Small and medium-sized businesses

Data sources and methodology

Target population

The target population is derived from Statistics Canada's Business Register (BR). The BR is an information database on the Canadian business population and serves as a frame for all Statistics Canada business surveys. It is a structured list of businesses engaged in the production of goods and services in Canada. The following enterprises are excluded to arrive at the target population:

1- Enterprises with 0 employees or with 500 or more employees
2- Enterprises with less than $30,000 in gross revenue
3- Non-profit organizations
4- Joint ventures
5- Government agencies
6- Enterprises in specific industries, identified by the North American Industry Classification System. These industries are: utilities (22); finance and insurance (52); management of companies and enterprises (55); educational services (61); public administration (91); automotive equipment rental and leasing (5321); commercial and industrial machinery and equipment rental and leasing (5324); out-patient care centres (6214); medical and diagnostic laboratories (6215); other ambulatory health care services (6219); general medical and surgical hospitals (6221); psychiatric and substance abuse hospitals (6222); specialty (except psychiatric and substance abuse) hospitals (6223); community food and housing, and emergency and other relief services (6242); and private households (814110).

Instrument design

The survey questionnaire was designed by Statistics Canada in collaboration with Innovation, Science and Economic Development Canada and its consortium of partners. The Questionnaire Design Resource Centre of Statistics Canada made suggestions to improve the questionnaire and then field-tested the questionnaire with approximately twenty-five different businesses. The results of those tests were used to further improve the questionnaire.

Sampling

This is a sample survey with a cross-sectional design.

The Business Register (BR) was the starting point for creating the survey frame. The latter was constructed by selecting from the Register all enterprises that met the criteria which defined the target population. The number of enterprises within the target population was 840,989.

The sampling unit was the enterprise. The main population was stratified by age of business, enterprise size, industry and geography. The regions were the Atlantic Canada, Quebec, Ontario, Manitoba and Saskatchewan, Alberta, and British Columbia together with the territories. The industries were based on the 2017 North American Industry Classification System (NAICS), and the size was based on the number of employees of the enterprise. For enterprises less than two years old (start-ups), there were two size categories: 1 to 4 employees and 5 or more employees. For enterprises two years and older, there were four categories: 1 to 4 employees; 5 to 19 employees; 20 to 99 employees and 100 or more employees.

The size of the sample selected was determined based on the following:
- The most important variables of interest were proportions (assuming a maximum proportion of 50% for each variable of interest).
- a minimum stratum sample size requirement of 10 units for province level strata and 5 units for strata defined at the sub-province level within Atlantic Canada and Ontario was imposed. For a few strata, this increase is not possible because it would exceed the population size and, therefore, we take all the units.
- Precision targets (standard errors) were specified for certain groupings of strata.
- The sample selection method was simple random sampling without replacement.
- A minimum response rate of 35% for start-ups and 40% for the general population was expected.
The base sample size was 17,323 out of a total of 840,989 units in the survey frame.

In addition to estimates for the base population, estimates for certain special populations were needed. These populations came from lists supplied to Statistics Canada containing units that represented the target population, except for Information and Communication Technologies, where the population came from a pre-specified list of NAICS codes. The sample sizes for each of these sub-populations were:
- Cooperatives, 617 units selected
- Users of the Canadian Small Business Financing Program (CSBFP), 1,334 units selected
- Enterprises that had contracts with Public Services and Procurement Canada (PSPC), 675 units selected
- Information and Communication Technologies, 640 units selected
- Enterprises that used Business Development Bank of Canada (BDC) services, 1,354 units selected
- Enterprises that were involved in producing or providing clean technology products or services, 622 units selected
- Registered charities and non-profits that were identified by Employment and Social Development Canada (ESDC) to be Social Enterprises, 1,146 units selected

The supplementary lists were not stratified, with the exception of the CSBFP list. This list was stratified such that strata with a greater probability of containing units within the scope of the survey were more heavily sampled. In each list, units were selected randomly using the method of simple random sampling without replacement.

Combining the base sample and the samples from the supplementary lists, the final unduplicated size of the sample was 23,527 enterprises.

Data sources

Data collection for this reference period: 2018-02-05 to 2018-06-08

Responding to this survey is voluntary.

Data are collected directly from survey respondents.

Data collection is conducted by Computer Assisted Telephone Interviewing (CATI) methods.

View the Questionnaire(s) and reporting guide(s) .

Error detection

Error detection is an integral part of both collection and data processing activities. Automated edits were applied to data records during collection to identify capture and reporting errors. Respondents were asked to validate their reported data when these collection edits failed.

Prior to imputation, a series of edits were applied to the collected data to identify errors and inconsistencies. Outlier detection was also performed on select variables to identify improbable or influential values. All outliers were further verified and those deemed to be outliers were imputed along with incoherent and missing values.

Errors and inconsistencies in the data were reviewed and resolved by referring to data for similar units in the survey, data from previous iterations of the survey and information from external sources. If a record could not be resolved, it was flagged for imputation.

Finally, edit rules were incorporated into the imputation system to detect and resolve any remaining errors, as well as to ensure that the imputed data were consistent.

Imputation

After microdata verification, a variable was created for each of the survey variables to identify those that had either failed the verification rules or had missing values. Two classes of units were created: total non-response cases and partial non-response cases. Total non-response units were treated through weighting, as the weights of responding units in the same homogenous class with respect to the propensity to respond were adjusted to represent the non-response units as well. These adjusted weights were calibrated to the strata population counts. Partial non-response units were completed using imputation.

Partial non-response imputation was performed independently for each of the various target populations. For instance, complete data from the sample for cooperatives cannot be used to impute missing data for units in the information and communication technologies sample. The missing variables were imputed using a randomly selected donor or using the nearest neighbour imputation method. For the nearest neighbour approach, the minimax distance function is used to find the closest donor. The minimax distance function determines the closest donor as being the one with the smallest maximum absolute difference between the value of its matching variables and those of the recipient. For most variables, the matching variable used was the revenue figure. Imputation was performed within groups of units called imputation classes. These imputation classes were formed of similar size units (employment), of similar age, in the same geography and in the same industry.

A minimum number of units was required within each imputation class. When imputation classes were too small, larger classes were created by combining several classes together.

To ensure internal consistency (coherence between variables of the same record), missing or incoherent variable values were imputed in sequence in which they appeared on the questionnaire. This allowed for an imputed question at one point in the questionnaire may have been used as a matching variable for a question located further along in the questionnaire.

Most imputation of survey variables was performed in an automated way using BANFF, a generalized system designed by Statistics Canada.

Estimation

A complete file of weighted micro data was created for all sampled enterprises in the survey population for which data were reported or imputed. Weights were adjusted to account for total non-response so that the final estimates would be representative of the entire survey population. Weighted estimates were produced using the Generalized Estimation System (G-Est).

Quality evaluation

Where possible, estimates were compared with similar data from previous iterations of the survey to ensure that the survey results were consistent with historical findings. In addition, subject matter experts from outside Statistics Canada were given an opportunity to review the survey microdata and estimates, as well as provide feedback on their quality prior to their official release.

Disclosure control

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.

In order to prevent any data disclosure, confidentiality checks were performed for direct disclosure as well as for the secondary suppression (residual disclosure). Direct disclosure occurs when the value in a tabulation cell is composed of or dominated by few enterprises. Residual disclosure occurs when confidential information can be derived indirectly by piecing together information from different sources or data series.

Revisions and seasonal adjustment

This methodology type does not apply to this statistical program.

Data accuracy

Since all estimates to the Survey on Financing and Growth of Small and Medium Enterprises are based on sample results, they are subject to sampling error. This error can be expressed as a standard error. For example, the proportion of firms in the target population that would respond YES to a given question is estimated to be 50%, with a standard error of 0.04. In repeated sampling, the estimate would be expected to fall between 42% and 58%, nineteen times out of twenty. The following rules based on the standard error are used to assign a measure of quality to all of the estimates of percentages.

Quality code A - Excellent (Coefficient of variation - Up to 5%; Standard error for percentages - Up to 2.5%)
Quality code B - Good (Coefficient of variation - 5% up to 10%; Standard error for percentages - 2.5% up to 5%)
Quality code C - Average (Coefficient of variation - 10% up to 15%; Standard error for percentages - 5% up to 7.5%)
Quality code D - Mediocre (Coefficient of variation - 15% up to 20%; Standard error for percentages - 7.5% up to 10%)
Quality code E - Poor, use with caution (Coefficient of variation - 20% up to 25%; Standard error for percentages - 10% up to 12.5%
Quality code F - Not reliable, do not use (Coefficient of variation - 25% or higher; Standard error for percentages - 12.5% or higher)

The survey response rate was calculated as the number of respondents divided by the number of estimated in-scope units. The number of in-scope units includes all respondents, in-scope seasonal or part-time operations and an estimate of the number of in-scope units included among non-respondents. For the main population, the response rate was computed as 59.7%.

Date modified: