22nd Meeting of the Wiesbaden Group on Business Registers
– International Roundtable on Business Survey Frames
Tallinn, 27 – 30 September 2010
René BEAUDOIN
Statistics Canada
Canada
Session No 5
Register quality as a common task: cooperation of the register department and survey departments
Business Register Coverage Improvements : Adjustments for Activity based surveys


Table of Contents

1 Introduction 2

1.1 Background 2

1.2 Vision and Objectives 2

2 Strategies 2

2.1 Simplification 2

2.1.1 Integrated Business Structure - Reconciliation of concepts 2

2.1.2 Generic Frame Files supporting survey processes 2

2.1.3 Generic Survey Frame Assessment Tool 2

2.2 Completeness 2

2.2.1 Improve timeliness 2

2.3 Completeness - Extend the Coverage 2

2.3.1 Survey Specific Fields (SSF) 2

2.3.2 Activity codes & associated size measures 2

2.4 Accessibility 2

2.4.1 Allowing users to perform updates 2

2.4.2 Integration with the collection system 2

3 Opportunities 2

3.1 To improve response burden management 2

3.2 To improve respondent relations 2

3.3 To improve coherence between various survey results 2

3.4 To create analytical potential for business demographics 2

4 Current Developments 2

4.1 Integration of Farm Register 2

4.2 Hook-up of more than 100 additional surveys - including all Activity based surveys 2

4.3 Integration of collection system 2

4.4 Improvement of geographical coverage by using additional Administrative information 2

5 Conclusion – Vision of the future 2

1  Introduction

The following paper will touch on the strategies being developed to ensure the realisation of a unique central frame that will satisfy the needs of all economic and institutional surveys, including activity based surveys, and thus, portraying the complete coverage of all business activities in Canada. In addition, this paper will outline the challenges and opportunities created when developing and maintaining a central frame.

1.1  Background

Like many statistical agencies around the world, Statistics Canada has been under financial pressure. For this reason, Statistics Canada is reviewing the way it is conducting its business. To continue delivering the same products and services, with the same level of quality, Statistics Canada must find ways to be more efficient in producing and delivering these products and services. One of the many ways for Statistics Canada to become more efficient will be to reduce the duplication of systems supporting the same activity, and to maximize the usage of already existing central corporate services.

Previously, many systems performing similar functions were developed to satisfy the needs of specific programs or surveys. For example, even though the Business Register (BR) existed as the central frame for most of Statistics Canada‘s economic surveys, some economic surveys continued to maintain their own survey frames.

In April 2005, Statistics Canada began an ambitious project to completely redesign the BR environment. The main objectives of this redesign were to simplify the concepts and operating procedures, eliminate outdated mainframe technology and significantly reduce operating costs. The redesign project resulted in a completely new environment for the users. All information concerning a business structure and its characteristics as well as the tools that support operations and the analytical work are available within the same environment. This is a major improvement from the old system. Every module was developed with the objective that each could easily be linked to the other. This has resulted in a truly integrated environment for users of the BR.

1.2  Vision and Objectives

Within the current context, one objective is to work towards the BR as the one and only frame, for all economic and institutional surveys in order to maximize efficiency and reduce frame maintenance costs (i.e.: to reduce/eliminate the number of survey frames and all satellite systems which duplicate functions such as: sample selection, collection, allocation and estimation). For this to be realized, the central frame must accommodate the needs of both the Industry and Activity based surveys.

Our vision of a Central BR is based on the following principles: the system must be simple, flexible and accessible by all.

Having centralized control over the maintenance of the BR will also allow us to improve the frame quality as we can maximize the usage of administrative data and survey feedback to update the frame. In addition, a formal training program and standard update procedures for all BR users will help in ensuring the coherence and consistency of all updates.

2  Strategies

This section of the paper provides an overview of the strategies developed to capitalize on the efficiencies that can be generated from the use of a unique central frame, as well as the strategies developed to ensure that the central frame will satisfy the needs of all economic and institutional surveys, including activity based surveys, and thus, portraying the complete coverage of all business activities in Canada.

2.1  Simplification

2.1.1  Integrated Business Structure - Reconciliation of concepts

It is very important to determine clear concepts for the central frame. The new BR has simplified the way it represents business structures. First, we delineated all Legal Entities within any Ownership Structures, and second, we created and joined all of the Operating Entities corresponding to each of the Legal Entities within the structure. The result is an integrated structure of the Enterprise, or Enterprise Group, represented by all Legal and Operating Entities.

Following the creation or update of a structure, a fairly simple algorithm assigns the Statistical Indicators to all the entities in the structure. Based on the reporting capabilities characteristics assigned to each entity by the business updaters, the algorithm determines which entities within the structure represent the Enterprise(s), Company(ies), Establishment(s) and Location(s).

This new model of integrated structures, whereby Legal and Operating Entities are inter-related, helps maximize the usage of the administrative information received at the legal level and simplifies the frame maintenance processes.

2.1.2  Generic Frame Files supporting survey processes

In order simplify the usage of the BR and to generate additional efficiencies, it was decided from the very beginning that we would develop standard, generic files format and tools. Generic file format and tools will become standards across surveys, facilitating the understanding of the BR. By having these common tools, the users will get familiar with the BR overtime, regardless if they work on various surveys during the course of their career at Statistics Canada.

For every survey hooked to the BR, the Survey Universe File (SUF), the In Sample File (ISF) and the Survey Interface File (SIF) are required. They are used by people across the organization to obtain their targeted universe, identify their sampled units and produce their collection entities. (i.e.: they contain all information required to support the collection process).

Prior to the redesign, different sets of these files were produced for each survey. Although the core of the content was the same for all of them, these files, over time, were modified. This resulted in different record layouts and thus, each file needed to be maintained separately.

During the redesign we reviewed the content of the files for each survey and developed generic record layouts which are now used by all surveys. In addition to reducing the work that BRD performs in order to produce these files, it also helps the users as they don’t need to understand a different record layout if they should start working on a different survey.

Now, the new Generic SUF, portraying the entire frame population, is used for all surveys to identify their population of interest and select their sample. It is produced on the first day of each month, as opposed to different schedules for different surveys. It contains all variables used by all the surveys. Users no longer rely upon information from their own files to supplement that which was not available on the old SUF.

Having a single Generic SUF will simplify the training program and bring additional efficiencies as users will only need to be trained once during their career at Statistics Canada.

2.1.3  Generic Survey Frame Assessment Tool

Historically, each survey area had their own methods or tools to monitor their survey populations. Since every survey area needs to understand the changes happening to their population of interest between survey cycles, it was decided it would be a good investment to develop a generic tool which would allow all BR users to monitor their population of interest.

The Generic Survey Frame Assessment tool (G-SFA) was developed to support all BR users in the monitoring of their population of interest (i.e.: tracking births, deaths, arrival, departures, and other significant changes to their population).

2.2 Completeness

In a perfect world, a survey frame would represent exactly the population of interest, as it exists “today in the real world”, with all the appropriate dimensions.

2.2.1  Improve timeliness

For the BR to have the most representative image of the business population possible, the update processes have to be as timely as possible. Below is a series of strategies recently implemented to improve the timeliness of the information on the BR and to ensure a more complete coverage.

2.2.1.1  Regular frame update processes maximizing the use of administrative information

The BR’s core information is updated on a monthly basis using many administrative data sources, maximizing its representation of real world as much as possible. We continue to improve the data quality of the frame by systematically applying updates which are available from various taxation data files. Businesses are created, activated and inactivated each month. Revenue information from the Goods and Services Tax (GST) file, as well as number of employee from Payroll Deduction files, are applied on a monthly basis. Financial data on corporations’ income tax filings are applied as they are received from Canada Revenue Agency (CRA) as opposed to once a year.

Administrative information such as sales tax remittances, income tax returns, and payroll deductions provide us with clear signals when a business is active. Signals of when a business becomes inactive are a lot less clear.

In the old environment, once a business was considered to be alive, there was no systematic process to inactivate that unit apart from survey feedback, profiling activities or when all the BN accounts had been closed by CRA. Unfortunately, it could take years before CRA actually closed all the accounts of a given business, even though the business was not producing any economic activity. In the new BR, we have developed rules that are applied on a monthly basis to inactivate businesses that are still considered to be ‘alive’ by CRA but are no longer operating. If the business is ‘inactive’ according to our rules, it can later be re-activated if there is evidence of economic activity on the taxation data files.

2.2.1.2  Simplification of updates

The operational procedures of the new BR have been developed to facilitate updates to the business structure and their characteristics. This simplification was designed in order to reduce the time required to apply updates so these changes can be reflected in the survey universe as soon as possible. Simplification also allows subject matter specialists to perform updates themselves thus eliminating the need to send a request for update to the BR staff. Instead, we have granted permissions to subject matter specialists to perform the updates provided they have received the appropriate training. In addition, we are planning to provide collection staff, a direct access to update the frame starting early 2011.

2.3  Completeness - Extend the Coverage

Historically, the central BR main coverage dimensions were the following: Size (mainly revenue, assets and number of employees), Geographical (provinces) and Industry (NAICS; based on the main business activity). One of the main reasons explaining why many surveys did not hook their surveys to the business register in the past was lack of information about the businesses characteristics required by these surveys to either identify or stratify their population. In order to be able to recognize or properly stratify their population of interest on the central frame, many survey areas required the BR to store additional information about the activities of the businesses.

2.3.1  Survey Specific Fields (SSF)

This section briefly describes the key component allowing “non-traditional industry based surveys” to hook to the BR.

The most important generic functionality developed to satisfy the specific needs of survey area is the “Survey Specific Fields (SSF)” table associated to any subset of business entities. The addition of Survey Specific Fields within the BR has opened the door to many surveys previously unable to hook to the BR, to be able to do so now.

The SSF is a new functionality now available to all subject matter divisions. The BR maintains a pre-defined list of stratification variables for every operating entity on the Register. For most surveys, these variables are adequate for selecting their sample for a given population. However, some surveys need additional variables to efficiently identify their survey universe. For example, the Balance of Payments Division uses an additional 15 variables at the enterprise level in order to select their sample for all their surveys. Before the introduction of the new BR, they had to maintain this information separately from the central frame. Now, they can easily attach these additional variables to any given operating entity. Since these variables are stored on the BR, this information is also viewable and accessible to all other users.

Here are some other examples of some variables of interests that are survey specifics: Gross leasable area (Retail survey), number of beds (Hospitals survey), number of trucks (Trucking survey), etc.

This new generic functionality will also allow BR users to target businesses included in different sub-population of cross-cutting surveys, opening the door for new multi-dimensional surveys.

2.3.2  Activity codes & associated size measures

Similar to the needs of surveys requiring survey specific information, activity based surveys also require the central BR to possess additional information about the businesses. For the central BR to support activity based surveys, it is necessary for the BR to store this new coverage dimension as the activity codes.

It was decided to develop standard codes and/or use recognized standard codes, like the NAICS, in order to identify the businesses’ secondary activities. The use of standard codes will facilitate the coherence and understanding of the frame by all its users.