4.0 Options

Archived information

Archived information 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.

The development and evaluation of alternative options involved several steps:

A review team was developed consisting of a group of Agriculture Division managers, methodologists from Business Surveys Methods Division and a representative from AAFC.
The key overarching considerations to be taken into account during the assessment of the alternative options were developed and agreed upon by the Review Team and Senior Management at Statistics Canada.
Three alternative options were developed based on the priority data requirements for the Canadian program, coupled with the international review.
Detailed criteria were developed to evaluate the alternative models. (These criteria are presented in Table 2 on the following page.)
The three alternative options were subsequently evaluated against the current Canadian program. The essential conditions and investments required for implementation of the models were determined along with each option's strengths, weaknesses and risks.
The most attractive attributes of the three alternative options within the Canadian context were combined to develop two hybrid options, which were subsequently evaluated.

4.1 Evaluation of the options

The following table lists the criteria that were used to evaluate the options. A total of 32 evaluation criteria were identified and organized into 10 categories. They include Statistics Canada's six elements of quality (relevance, accuracy, timeliness, coherence, interpretability and accessibility) as well as a number of other categories that merit special consideration in the context of the agriculture statistics program (cost, response burden, operational feasibility and acceptability).

**Table 2: Option evaluation criteria**
Quality	Evaluation criteria
Relevance:	Content Frequency Target population Small area data needs
Accuracy: (Reliability)	CVs of important survey variables Bias of important survey variables Available information to identify the target population and associate it with the data Quality of 'take-none' modelling (defining the survey population) Accuracy of the data sources
Coherence:(Comparability)	Coherence of important survey variable time series Coherence of data between sources
Timeliness:	Impact of the data source on the timeliness
Interpretability:	Details available on administrative files
Accessibility:	Data suppression Availability of supplementary information
Respondent Burden:	Number of contacts per unit Interviewing time per unit Sensitive content Burden on people other than survey respondents Burden placed on respondents by entities other than Statistics Canada
Cost:	Collection costs Post-collection costs Development costs Cost sustainability Costs to other organizations in the system Compliance costs to farmers
Operations:	Ability to react quickly to new needs Ability to conduct large occasional surveys Timing of and time necessary for implementation Statistics Canada Corporate Business Architecture compliance
Acceptability:	Acceptance in the data user community Acceptance in the respondent community

The key features, strengths, weaknesses, risks and investments of the current Canadian program are presented next, followed by the evaluation of the alternative models that were explored.

4.2 Baseline option: The current Canadian program

The current Canadian program is a highly integrated system. The production data from both the crops and livestock sections combined with prices from both survey and administrative sources generate the farm cash receipts. Expenses, derived largely from administrative sources, serve to generate the net farm income estimates. As well, data from CEAG flow into the commodity programs, while data from the commodity, farm income and prices programs are used to validate the CEAG data. Survey frame updates flow from the survey programs into the Farm Register,²⁰ and, subsequently, into the CEAG, and vice versa. The integrated nature of the program requires that for any proposed change to part of the program, impacts on the other components of the program must be assessed.

Description for figure 1

Figure 1: The Current Agriculture Statistics Program

4.2.1 Key features

A CEAG is conducted nationally every five years in years ending in "1" and "6." Response burden is minimized during the years that the CEAG is conducted. For the 2011 CEAG, follow-up calls were eliminated or co-ordinated for the FFS sample, some surveys were cancelled, the cap-on-calls for the majority of surveys was reduced and the sample size for the July livestock survey was reduced significantly.
The CEAG is linked to the CEPOP/NHS in years ending in "1" and "6" to provide socioeconomic data.
The target population for both the CEAG and surveys includes all farms with the intention to sell agricultural products. This definition provides comprehensive coverage to users.
The survey population varies by survey; some survey samples exclude operations based on a minimum size threshold or for specific farm types. (For example, the FFS excludes operations with complex structures, farms on First Nations Reserves, community pastures and farms with less than $10,000 in gross sales.)
A frame maintenance program includes information from both administrative sources and a short survey to update and maintain the Farm Register.
The survey program is commodity specific and comprises field crop, horticulture, livestock and financial surveys.
Remote sensing delivers the CCAP, which combines earth observation, geographic information systems (GIS) and the Internet to provide near-real time information on crop and pasture/rangeland conditions using a mapping application for agricultural land.
Administrative data are an integral part of the program (approximately 140 different sources are incorporated) including tax data, marketings, prices, imports, exports, production, debt, inspections data, etc. These data are provided by Canada Revenue Agency, provincial administrations, national producer organizations, AAFC and Statistics Canada's International Trade Division.

4.2.2 Strengths

The five-year interval between CEAGs maintains the relevance and usefulness of the data to users. The CEAG data used for policy development and evaluation, program monitoring, benchmarking, measuring industry structural changes, supporting legislative and regulatory instruments and for trade purposes are perceived to be sufficiently frequent.
Although some data gaps exist, this model meets the majority of user requirements for small area data, benchmarking and critical survey frame information.
This model has the advantage of reliability and predictability since the program has been running successfully for a long period of time.

4.2.3 Weaknesses

Response burden is a concern in an environment where the Government of Canada is firmly committed to reducing red tape.
The cost of the program is a concern in an environment of deficit reduction and increased efficiency.
Despite some very rapidly produced statistics, there are some concerns with the timeliness of some of the statistics.

4.2.4 Essential conditions

The current program has developed over time with the funding, technology and infrastructure required, so the essential conditions for this option are in place. However, fiscal pressures and commitments to reduce response burden are raising uncertainty as to the sustainability of this model.
The Corporate Business Architecture (CBA) is transforming the way that Statistics Canada collects and compiles data. The entire agriculture statistics program will complete the transition to the CBA in 2014-15. The CBA is expected to increase the overall efficiency of the agriculture program.
The transition from the Farm Register to the Business Register in 2012 is also expected to reduce the cost of the frame and will allow the Division to measure and manage response burden in a more global manner.

4.2.5 Required investments

Regular maintenance costs and post-censal redesign for an existing survey system.
Investments required for each CEAG cycle through Treasury Board submissions.

4.2.6 Risks

Should the 2016 CEAG be cancelled by Order in Council, there would not be time to fully replace it. There would therefore be significant data gaps, particularly related to benchmark data, small area data and frame update information.

4.3 Option 1: The modified British Model

4.3.1 Key features

A CEAG would be conducted every 10 years (in years ending in "1").
The CEAG would be linked to the CEPOP/NHS in years ending in "1" to provide socioeconomic data.
Two annual modular surveys would replace 12 of the current commodity-specific surveys conducted throughout the year. These surveys would be conducted in June and December. Different commodities would be collected together but subsequently processed and disseminated separately. Overlap between commodities would be controlled to reduce burden for diversified farms. (For example, a mixed livestock and crops farm may only receive the crops module for one survey occasion and the livestock module on another survey occasion. The operation would not automatically receive both modules on every selected survey occasion.) The number of survey occasions per year would be reduced for field crop, horticulture and livestock surveys.
The sample size and content of the June Modular Survey would be expanded in the years ending in "5" and "8" to compensate for some of the data loss due to the absence of a CEAG in years ending in "6." In these two years, comprehensive survey modules would be integrated, so that analysis can be conducted at the whole farm level as is presently the case with the CEAG. (For example, a mixed livestock and crops farm will receive both the crops and the livestock modules in these years.)
Tax data would be used to replace all comparable financial questions on the CEAG and on surveys.
The target population for both the CEAG and surveys would exclude smaller farms under a specified production threshold, (for example, an amount of cultivated land, livestock, other criteria or combination thereof), for reasons of burden and cost.
The survey population would be equivalent to the target population.
A regular frame maintenance program would include a short survey to complete missing information from new farm tax filer records and those operations not recently surveyed to update and maintain the agriculture frame on the Business Register. (It should be noted that some of these activities are already carried out in the current program.)
A small number of commodity-specific surveys (Greenhouse, Sod and Nursery Survey, Mushroom, Maple, etc.) would continue to exist because of their unique requirements.

4.3.2 Strengths

This option mitigates some of the risk of data loss by providing a subset of data requirements should the CEAG in years ending in "6" be cancelled by Order in Council.
Structural changes, new production and trends would be captured by the CEAG in years ending in "1" and partially captured with the two expanded occasions of the June Modular Survey in years ending in "5" and "8."
The new annual modular surveys provide a flexible, regular vehicle to identify and address emerging issues.

4.3.3 Weaknesses

This option's content and sample size increases in years ending in "5" and "8" do not meet user needs for small area and custom geographic data, for provincial benchmarking data or for the enumeration of rare or emerging commodities.
This option provides a reduced level of survey frame information, even with an increase in sample size in years ending in "5" and "8." This would lead to frame deterioration and a related decrease in data accuracy from the survey program over the intercensal period.
The ten year gap between CEAGs would reduce the relevance and usefulness of the data to users. The CEAG data used for policy development and evaluation, support of legislative and regulatory instruments and for trade purposes is likely to become out of date before the next CEAG is conducted.
The generalized survey design for integrated surveys would not be ideal for some commodities.
The timeliness of data releases would be affected.
This option does not allow the entire population to be measured. The program will no longer cover 100% of agriculture activity in Canada. This loss of coherence and comparability of the data would require transition data (back-casting) and technical assistance to data users to make adjustments for changes to the coverage of the target population and the availability and frequency of data.

4.3.4 Essential conditions

An Order in Council would be required to cancel a CEAG in years ending in "6."
Technology and procedures would have to exist to deliver the modular surveys in an intelligent manner, so that response burden and collection costs could be controlled. To minimize burden and collection costs, delivery and collection of the appropriate modules (crops, livestock, financial, other) would need to be established prior to collection. The appropriate module would be determined from CEAG information and frame and survey update information.
Due to the lengthy gap between CEAGs, an enhanced coverage program would have to be implemented to maintain the data required to determine whether an operation should be included in the target population based on the predetermined threshold. This coverage program could include access to AAFC administrative program data such as crop insurance, AgriInvest and AgriStability, supplemented with a frame update survey. Both the access to and processing of the administrative data would require development.

4.3.5 Required investments

The new modular surveys would have to be designed, tested, developed and implemented.
Historical data would have to be adjusted to match the new target population definition. This includes the development of user training material to avoid data misuse and misinterpretation and to clarify the impacts of the change to the target population.
Alternative sources of commodity data would have to be developed at the micro level (for example, program data such as crop insurance, AgriInvest and AgriStability). This is necessary to establish and maintain threshold information on the agriculture frame on the Business Register between CEAGs and the large survey years.

4.3.6 Risks

There could be negative reaction from data users regarding
- changes to the target population
- the loss of small area data
- the loss of provincial benchmarking data
- the increase in reaction time to capture new trends and industry structural changes
- the timeliness of specific annual commodity data that would be included in the two integrated surveys.
The two very large and comprehensive surveys in years ending in "5" and "8" may result in as much response burden as the CEAG in years ending in "6".

4.4 Option 2: The modified Australian/American Model

4.4.1 Key features

A CEAG would be conducted every 10 years (in years ending in "1"). (It should be noted, however, that both Australia and the US conduct censuses of agriculture every five years.)
The CEAG would be linked to the CEPOP/NHS in years ending in "1," to provide socioeconomic data.
One new large survey would be conducted in years ending in "6" to compensate for some of the data loss due to the lack of a CEAG in those years.
Tax data would be used to replace all comparable financial questions on the CEAG and surveys.
The target population for both the CEAG and surveys would exclude farms under an estimated value of agricultural operation. (For example, US = $1,000 USD; Australia = $5,000 AUD.)
The survey population would be equivalent to the target population.
A regular frame maintenance program would include a short survey to complete missing information from new farm tax filer records and those operations not recently surveyed to update and maintain the agriculture frame on the Business Register.
The survey program would remain commodity specific much like the current program. However, the number of survey occasions per year would be reduced for some crop, horticulture and livestock surveys.
Remote sensing would play an increasingly important role. This technology would be integrated into the agriculture statistics program as it becomes mature. The initial focus would be on replacing the Potato Area Survey and the July and September Field Crop Reporting Surveys in the Prairie provinces.
Administrative data would play an increasingly important role. These data would be integrated into the agriculture statistics program as they become available.
This option expands on current partnerships and promotes new partnerships with federal, provincial and industry stakeholders. These partnerships would be necessary to share responsibility for the development, collection and compilation of administrative data (such as AAFC's AgriInvest and AgriStability programs and livestock traceability data).

4.4.2 Strengths

This option would achieve cost savings and reduce response burden by conducting a new large survey in years ending in "6" instead of conducting a CEAG.
This option mitigates some of the risk of data loss by providing a subset of data requirements should the CEAG in years ending in "6" be cancelled by Order in Council.
Since the annual agriculture surveys remain relatively similar, this option would be expected to have little impact on data users in terms of timeliness and survey content.

4.4.3 Weaknesses

Despite the content and sample size increases of the new large survey in years ending in "6," this option does not meet user needs for small area data, custom geographic data, provincial benchmarking data or for the enumeration of rare or emerging commodities.
This option provides a reduced level of survey frame information, even with an increase in sample size in years ending in "6." This would lead to frame deterioration and a related decrease in data accuracy from the survey program over the intercensal period.
The ten year gap between CEAGs would reduce the relevance and usefulness of the data to users. The CEAG data used for policy development and evaluation, support of legislative and regulatory instruments and for trade purposes is likely to become out of date before the next CEAG is conducted.
This option does not allow the entire population to be measured. The program will no longer cover 100% of agriculture activity in Canada. This loss of coherence and comparability of the data would require transition data (back-casting) and technical assistance to data users to make adjustments for changes to the coverage of the target population and the availability and frequency of data.

4.4.4 Essential conditions

An Order in Council would be required to cancel a CEAG in years ending in "6."
To increase the use of administrative data, it would be necessary to renegotiate existing partnerships or develop new ones with federal, provincial and industry stakeholders. These agreements would establish protocols for data sharing, confidentiality and protection. The collaboration of multiple players over several jurisdictions would have to be established and maintained. This commitment must begin at the highest levels in the participating organizations and extend to the working level.
Federal, provincial and industry data holders would need to include a declaration to their data providers (farm operators) regarding the provision of data for statistical purposes. There may be a need to change legislation.
Agriculture respondents would have to be aware of and support the increasing use of administrative data, being aware of the associated benefits and risks.
A feasibility study would be required to fully evaluate the costs, benefits, risks and potential timeframes for incorporating administrative data sources and increased use of technology (such as remote sensing) into the program.
A methodologically sound and realistic framework through which new sources of administrative data could be identified, evaluated, incorporated and operationalized in the program must be developed to reduce the risk of errors.

4.4.5 Required investments

Historical data would have to be adjusted to match the new target population definition. This includes the development of user training material to avoid data misuse and misinterpretation and to clarify the impacts of the change to the target population.
Alternative sources of commodity data would have to be developed at the micro level (for example, program data such as crop insurance, AgriInvest and AgriStability) for statistical purposes and to establish and maintain threshold information on the agriculture frame on the Business Register between CEAGs and the new large survey years.
Remote sensing would have to be developed to completely or partially replace traditional field crop surveys. A Land Area Survey would need to be developed. The data from this survey combined with administrative data (e.g., crop insurance data) would be used to calibrate remote sensing results.
A new large survey to replace the CEAG in years ending in "6" would have to be developed and implemented.

4.4.6 Risks

There could be negative reaction from data users regarding
- changes to the target population
- the loss of small area data
- the loss of provincial benchmarking data
- the increase in reaction time to capture new trends and industry structural changes.
Increased reliance on administrative data sources may put the coherence, comparability and sustainability of the data at risk due to changes in programs, regulations or provider partners over time.

4.5 Option 3: The modified Scandinavian Model

4.5.1 Key features

Administrative data form the basis of Option 3. There would be no traditional CEAG. (It should be noted, however, that all Scandinavian countries conduct a traditional CEAG every 10 years.) An administratively based CEAG could potentially be conducted on an annual basis if sufficient information existed.
Linking of the CEAG to the CEPOP/NHS should be possible for the years that the CEPOP/NHS are conducted to provide socioeconomic data.
Existing administrative data sources would be expanded to include new sources, as they become available. For example,
- crop insurance
- AgriInvest, AgriStability and Business Risk Management programs
- CFIA data
- national producer organizations (NPOs)
- livestock traceability systems.
A farm structure survey would be conducted every three years to address administrative data gaps, monitor changes, measure emerging trends and perform frame updates and maintenance.
Tax data would be used to replace all comparable financial questions on the CEAG and surveys.
The target population for both the CEAG and surveys would exclude farms under a specific sales threshold.
The survey population would be equivalent to the target population.
A regular frame maintenance program would include a short survey to complete missing information from farm tax filer or administrative data records to update and maintain the agriculture frame on the Business Register.
A small program of other surveys would be run each year, if necessary, to cover data requirements not covered by administrative data or farm structure survey data (on specific commodities such as fur production). As more administrative sources become available, more survey data would be replaced by administrative data.
This option depends on reliable, complete, timely, stable and accessible administrative information covering the target population.

4.5.2 Strengths

This option achieves significant cost savings in the long term by cancelling the CEAG in years ending in "1" and "6."
Because it uses data already collected for administrative purposes, the marginal costs of producing statistics are generally much less than for a traditional CEAG or commodity-specific survey (once the databases, systems, and data-sharing and protection protocols are in place).
This option has the potential to reduce survey response burden by replacing the traditional CEAG and survey program with an administratively based CEAG and survey program that uses data already collected for other purposes.
Like the traditional CEAG, the administratively based CEAG can meet the objectives of the FAO features of a CEAG, which are to provide data on the structure of agriculture (from small administrative units) that enable detailed cross-tabulations to use as benchmarks for current agriculture statistics and frames for agricultural sample surveys.²¹
An administratively based CEAG may be able to produce data on a yearly basis, compared to every five or ten years for a traditional CEAG.
The administratively based CEAG may be used to identify subgroups for surveys, if needed, depending on the variables available.

4.5.3 Weaknesses

Few of the essential conditions currently exist for this option to be successfully implemented in the short or medium term. Development of an administratively based CEAG would be a longer term process, requiring several years or even decades.
A significant front-end investment would be required to implement this option. In addition, negotiating agreements among many players and across multiple levels of government and non-governmental organizations would be necessary. Maintaining systems, definitions, concepts, as well as ongoing oversight would require additional resources and funding.
The program content would initially be limited to the data variables already available in the administrative databases. Over time, the required variables could be added to the administrative requirements of the programs, so that they could be collected for statistical purposes. This would likely require additional legislation and funding. It may also require enforcement strategies to ensure compliance with the statistical requirements and data-sharing agreements.
The concepts and definitions that apply to data in the administrative databases may not correspond to those desired for statistical purposes. Linkage of different administrative databases for the same unit may result in data inconsistencies that may be difficult to resolve without significant investments. Changes to and differences in concepts, definitions, target populations, etc., of administrative sources across jurisdictions and over time may limit the data availability, comparability and accuracy of the data for statistical purposes.
Unlike the traditional CEAG, the administratively based model cannot provide a snapshot of the entire country at one point in time during a census year. Data from multiple administrative sources are unlikely to reference one date.
There would likely be an increase in overall response burden due to the fact that every agriculture producer would be required to provide administrative data to fill statistical requirements whereas a survey approach requires only a sample of operators to provide such data.
There would be increased burden placed on the providers of administrative data to meet the requirements of the national statistical agency.
Developing and expanding partnerships with federal, provincial and industry stakeholders as well as with academia will require an investment of time and resources.

4.5.4 Essential conditions

The Statistics Act may need to be revised to cancel the CEAG in years ending in "1" and "6." (An investigation would need to determine if an administratively based CEAG would meet the legal requirements for a CEAG.)
Legislation to provide a stronger regulatory framework to develop, collect and acquire administrative data sources would be necessary. The legislation would also have to provide a detailed definition of data protection; for example, it should specify that the statistical data produced by the linkage process cannot be fed back to the administrative databases (known as the "one-way traffic" principle). In other words, the Statistics Act allows for Statistics Canada to procure data, but prevents Statistics Canada from feeding any data back to the source (for example, the Canada Revenue Agency) as stipulated in the Statistics Act.
A strong infrastructure covering legislative, regulatory and operational requirements, along with inter-agency cooperation across jurisdictions would be necessary. This would require the adoption of a "clearing house" approach to ensure that the same data are not collected more than once by different organizations.
To increase the use of administrative data, it would be necessary to renegotiate existing partnerships or develop new ones with federal, provincial and industry stakeholders. These agreements would establish protocols for data sharing, confidentiality and protection. The collaboration of multiple players over several jurisdictions would have to be established and maintained. This commitment must begin at the highest levels in the participating organizations and extend to the working level.
Federal, provincial and industry data holders would need to include a declaration to their data providers (farm operators) regarding the provision of data for statistical purposes. There may be a need to change legislation.
There would have to be a unique common identifier for all agricultural operations. This identifier should be used to conduct virtually all transactions with government (at all levels) thus enabling its use to link administrative data across all government sources.
Agriculture respondents would have to be aware of and support the increasing use of administrative data, being aware of the associated benefits and risks.
It would be necessary to develop a good set of register systems that fulfill administrative needs and that also contain data covering the most important subject areas for the statistical system. The coverage of these databases and the quality of the data contained within them would have to be sufficient to be useful for statistical purposes.
There would have to be incentives, such as legal requirements, for the target farm population to register and to inform authorities of changes or events (for example, changes of address, operator or ownership, bankruptcies). This documentation would have to be reliably recorded and with minimal delay.
It would be necessary to have a reliable method of assigning units to a detailed geographic level (geocoding) to produce small-area detail (for example, assigning owners, operators or establishments to specific geocodes).
In the absence of a CEAG, the frame maintenance would depend entirely on administrative data rather than drawing from a CEAG (and other sources).
Highly skilled, professional staff and training would be required to maintain this program due to the complexity that arises when data are procured from many different sources for many different programs. It would be necessary for analysts to be able to interpret the differences in concepts, definitions, scope and history of the administrative sources of data, particularly if attempting to conduct analysis in an integrated approach using data from different administrative sources. In addition, it would be necessary to educate users to accurately interpret the data due to the complexity of this option, both in terms of its operations and the resulting data.

4.5.5 Required investments

Historical data would have to be adjusted to match the new target population definition. This includes the development of user training material to avoid data misuse and misinterpretation and to clarify the impacts of the change to the target population.
Alternative sources of commodity data would have to be developed at the micro level (for example, program data such as crop insurance, AgriInvest and AgriStability) for statistical purposes and to establish and maintain threshold information on the agriculture frame on the Business Register.
Developmental costs may be substantial in the short and medium term. The cost of developing and maintaining these data and the systems required may be shifted from Statistics Canada to the providers of the administrative data. There would likely be costs related to the cleaning of the data and ensuring coherence among the sources. As such, Statistics Canada may be required to share these costs with the data holders.

4.5.6 Risks

There could be negative reaction from data users regarding
- changes to the target population
- the increase in reaction time to capture new trends and industry structural changes
- the timeliness of specific commodity statistics.
The potential exists to increase response burden by requiring all producers to provide information currently collected by a sample of the population. For example, data currently collected from a comparatively small sample of survey respondents represents the larger target population; however, if these same data were required on an administrative form, all program participants would be required to provide this information thereby significantly increasing response burden. (An example of this would be adding a data variable, currently collected on a sample survey, to a tax form, which all farm tax filers would be required to provide.)
Increased reliance on administrative data sources may put coherence, comparability and sustainability of the data at risk due to changes in programs, regulations or provider partners over time.
The perception of intrusiveness and loss of privacy may lead to a loss of cooperation by the agriculture community.

4.6 Summary of the alternative options

To better assess the potential of these three options to revamp the current Canadian agriculture statistics program, the advantages and disadvantages of each need to be compared and contrasted against the others. The following presents a summary of that evaluation.

Summary of Options 1, 2 and 3

Options 1 and 2 (modified British and Australian/American Models) both feature a CEAG conducted every ten years. Their main difference is in how they go about filling the intercensal data gaps; each model takes a different approach.

Option 1 (Modified British Model) involves a CEAG conducted every 10 years. It incorporates two modular surveys that would be conducted in June and December each year. A range of commodities would be collected together but subsequently processed and disseminated separately to maintain relevance to data users. Overlap between commodities would be controlled to reduce response burden for diversified farms.

A small number of commodity specific surveys would continue to exist because of their unique requirements. The sample size and content of the June Modular Survey would be expanded in the years ending in "5" and "8" to compensate for the absence of a CEAG in years ending in "6." In these two years, comprehensive survey modules would be integrated so that analysis could be conducted at the whole farm level as is presently possible with the current CEAG. The target population for both the CEAG and the surveys would exclude smaller farms under a specified production threshold. The survey population would be equivalent to the target population.

Option 2 (Modified Australian/American Model) involves a CEAG conducted every 10 years (although both countries conduct a CEAG every five years). It would comprise a commodity specific intercensal survey program much like the current Canadian program. One new large survey would be conducted to compensate for the loss of some of the data due to the absence of the CEAG in years ending in "6." The target population for both the CEAG and the surveys excludes farms under an estimated value of agricultural operation. For example, US = $1,000 USD; Australia = $5,000 AUD.) The survey population would be equivalent to the target population as in Option 1. However Option 2 features increased incorporation of remote sensing technology and administrative data compared with Option 1.

Option 3 (the Modified Scandinavian Model) is largely based on administrative data and therefore no traditional CEAG would be required (although it should be noted that all Scandinavian countries conduct a traditional CEAG every ten years). A farm structure survey would be conducted every three years to address administrative data gaps, monitor changes, measure emerging trends and perform frame updates and maintenance. If necessary, a small number of special surveys (for example, on specific commodities such as fur production) would be run each year to meet data requirements not covered by administrative data or by the farm structure survey. The target population for both the administratively based CEAG and the special surveys would exclude farms under a specific sales threshold. The survey population would be equivalent to the target population. This option would largely eliminate the need to conduct many agriculture surveys presently necessary with the current Canadian model, but is only possible when comprehensive databases and administrative data sources are available.

Advantages of Options 1, 2 and 3

The examination of these options led to the identification of their key strengths and weaknesses as well as to the investments that would be required for implementation in Canada. The most promising features that emerged from this evaluation include the modular survey approach of Option 1, the similarities with the current Canadian program of Option 2 and the incorporation of administrative data of Option 3.

Both Options 1 and 2 mitigate some of the risk of data loss by providing a subset of data requirements should the CEAG in years ending in "6" be cancelled. In their own way, each of these options would realize cost efficiencies and reduce response burden in the years ending in "6." With both options, structural changes, new production and trends would be captured by the CEAG in years ending in "1" and partially captured intercensally.

In addition, Option 2 provides the benefit of being relatively similar to the current survey program and therefore would be expected to have less impact on data users in terms of timeliness and survey content.

Option 3 achieves cost savings by cancelling the CEAG in years ending in "1" and "6." For the data already collected for existing administrative purposes, the marginal costs of producing statistics would be generally much less than for a traditional CEAG or commodity-specific survey (once the databases, systems, and data-sharing and protection protocols are in place). This model has the potential to reduce survey response burden by replacing traditional surveys with administrative data.

Like the traditional CEAG, the administratively based CEAG can meet the objectives of the FAO features of a CEAG, which are to provide data on the structure of agriculture (from small administrative units) that enable detailed cross-tabulations to use as benchmarks for current agriculture statistics and frames for agricultural sample surveys.²² An administratively based CEAG may be able to produce data on a yearly basis, compared to every five or ten years with a traditional CEAG.

Disadvantages of Options 1, 2 and 3

The weaknesses of these options were determined to be sufficiently significant that none of them could be adapted in their entirety to the Canadian context. These options are unable to adequately fill the data needs to replace the quinquennial CEAG, particularly when it comes to the need for benchmarking and small area data.

Option 1 would require significant restructuring of the current program including design, development, testing and implementation of the two new integrated modular surveys. As well, the integrated survey approach would adversely affect the timeliness for some crop and livestock estimates. In spite of this option's survey content and sample size increases in years ending in "5" and "8," this strategy would not meet user needs for small area and custom geographic data, for provincial benchmarking data and for the enumeration of rare or emerging commodities that only a CEAG based on complete enumeration can give.

The key disadvantages would be losses to coherence, data gaps and relevance related to

changes to the target population
the loss of small area data
the loss of provincial benchmarking data
the increased delay in capturing new trends and structural changes in the industry.

For Option 1, the two large and comprehensive surveys in years ending in "5" and "8" may result in as much response burden as the CEAG in years ending in "6" that they replace, without providing the benefits of complete enumeration at one point in time.

For both Options 1 and 2, the ten year gap between CEAGs would reduce the relevance and usefulness of the data to users. The CEAG data used for policy development and evaluation, support of legislative and regulatory instruments and for trade purposes would become out of date before the next CEAG is conducted. Like Option 1, Option 2 does not meet the user needs for small area data, provincial benchmarking data and for the enumeration of rare or emerging commodities.

In addition, these two options do not allow the entire population to be measured. The program will no longer cover 100% of agriculture activity in Canada. This would cause a loss of coherence and comparability of the data, which would require transition data (back-casting) and technical assistance to data users to adjust for changes to the target population and the availability and frequency of data. This work would also require the development of user training material to avoid data misuse and misinterpretation and to clarify the impacts of the change to the target population.

Option 3 would require an extensive administrative framework that does not currently exist in Canada and that would require significant time and investment to establish. As was discovered in the CEPOP review,²³ a common unique identifier permitting efficient linkages of multiple datasets would be required for such an administrative model to function. Such an identifier does not currently exist in Canada.

Additionally, response burden would likely be increased in such a model due to the fact that every agriculture producer would be required to provide administrative data to fill statistical requirements, whereas the present survey approach requires only a sample of operators to provide such data. Administrative concepts would have to be aligned with statistical concepts to ensure coherence. Privacy and confidentiality aspects of a program based on administrative data would have to be evaluated.

The evaluation determined that none of these options on their own would be an adequate replacement for the current agriculture statistics program. The investments required coupled with the data losses and compromises to the quality, timeliness, relevance and coherence of the data are not outweighed by the reduction in response burden and costs.

However, specific components of these alternative options were identified as being productive and efficient.

4.7 Refining the options for further consideration

The evaluation of the three options led to the identification of their most attractive features and their major weaknesses when considered in the Canadian context. Consequently, two hybrids of these options were constructed that incorporate these advantages while minimizing the disadvantages.

A description of these two hybrid options follows.

4.8 Option 4: Hybrid A

4.8.1 Key features

Hybrid A features a full decennial CEAG with increases in content and sample sizes of commodity-specific surveys in years ending in "6," coupled with increased use of administrative data and remote sensing.

More specifically:

A CEAG would be conducted every 10 years (in years ending in "1"). The questionnaire content would be similar to the current Canadian option, with the following distinctions:
- the detailed expense questions would be replaced with taxation data (i.e., the CEAG would exclude these questions)
- any questions that could be replaced with comparable and available administrative data would be excluded.
The CEAG would be linked to the CEPOP/NHS in years ending in "1" to provide socioeconomic data.
There would be an increase in content and sample size in the existing commodity-specific surveys in years ending in "6" to compensate for some of the data loss due to the absence of a quinquennial CEAG.
The survey program would remain commodity specific much like the current program. The number of survey occasions per year would be reduced for some crop, horticulture and livestock surveys.
Tax data would be used to replace all comparable financial questions on the CEAG and in surveys.
The target population would remain the same as for the current program (i.e., the target population includes all farms that produce agricultural products intended for sale).
The survey population would continue to exclude smaller farms under a specified threshold for reasons of burden and cost. The option to raise the threshold for specific surveys needs to be investigated further. The non-surveyed population would continue to be estimated (using statistical models) and included in published estimates.
An annual rolling frame update program would provide frame maintenance on a continuous basis for frame update and sampling efficiency purposes that would have been provided by a quinquennial CEAG. The program would include a short annual survey to a rotating percentage of the target population to complete missing information for new farm tax filers and operations not recently surveyed, so that the agriculture frame on the Business Register can be updated and maintained.
Remote sensing would play an increasingly important role. This technology would be integrated into the agriculture statistics program as it becomes mature. The initial focus would be on replacing the Potato Area Survey and the July and September Field Crop Reporting Surveys in the Prairie provinces.
Administrative data would play an increasingly important role. These data would be integrated into the agriculture statistics program, replacing content on surveys and censuses as they become available.
This option expands on current partnerships and promotes new partnerships with federal, provincial and industry stakeholders. These partnerships would be necessary to share responsibility for the development, collection, and compilation of administrative data and for Statistics Canada to obtain access (such as to AAFC's AgriInvest and AgriStability programs and livestock traceability data).

4.8.2 Strengths

This option provides an evolutionary approach to change within the agriculture statistics program, reducing risks to the relevance, coherence and accuracy of the program.
This option would achieve cost savings and reduce response burden by replacing the CEAG in years ending in "6." These reductions would be partially offset with increases in sample size and content for the main annual surveys in years ending in "6" and an increased frame update survey.
This option mitigates some of the risk of data loss by providing a subset of data requirements should the CEAG in years ending in "6" be cancelled by Order in Council.
The current target population definition would remain unchanged and therefore the coherence of agriculture data would not be affected. No investment would be required to adjust historical data for a new target population definition. Similarly, there would be no investment required to develop training material for users to avoid data misuse and misinterpretation and to clarify impacts of changes to the target population.
Since the annual agriculture surveys remain relatively similar to the current program, this option would be expected to have little impact on data users in terms of timeliness and survey content.

4.8.3 Weaknesses

This option's survey content and sample size increases in years ending in "6" do not meet user needs for small area and custom geographic data, for provincial benchmarking data or for the enumeration of rare or emerging commodities.
This option provides a reduced level of survey frame information, even with a rolling frame update, leading to frame deterioration and a related decrease in data accuracy from the survey program over the intercensal period.
The ten year gap between CEAGs would reduce the relevance and usefulness of the data to users. The CEAG data used for policy development and evaluation, support of legislative and regulatory instruments and for trade purposes is likely to become out of date before the next CEAG is conducted.
The loss of the quinquennial CEAG would impact the ability to model for the non-surveyed portion of the population in the survey program.

4.8.4 Essential conditions

An Order in Council would be required to cancel the CEAG in years ending in "6."
To increase the use of administrative data, it would be necessary to renegotiate existing partnerships or develop new ones with federal, provincial and industry stakeholders. These agreements would establish protocols for data sharing, confidentiality and protection. The collaboration of multiple players over several jurisdictions would have to be established and maintained. This commitment must begin at the highest levels in the participating organizations and extend to the working level.
Federal, provincial and industry data holders would need to include a declaration to their data providers (farm operators) regarding the provision of data for statistical purposes. There may be a need to change legislation.
Agriculture respondents would have to be aware of and support the increasing use of administrative data, being aware of the associated benefits and risks.
A feasibility study would be required to fully evaluate the costs, benefits, risks and potential timeframes for incorporating administrative data sources and increased use of technology (such as remote sensing) into the program.
A methodologically sound and realistic framework through which new sources of administrative data could be identified, evaluated, incorporated and operationalized in the program would have to be developed to reduce the risk of error.

4.8.5 Required investments

Alternative sources of commodity data would have to be acquired, adapted and incorporated at the micro level (for example, program data such as crop insurance, AgriInvest and AgriStability).
Remote sensing would have to be developed to completely or partially replace traditional field crop surveys. A Land Area Survey would need to be developed. The data from this survey combined with administrative data (e.g., crop insurance data) would be used to calibrate remote sensing results.

4.8.6 Risks

There could be negative reaction from data users regarding
- the loss of some small area data
- the loss of some provincial benchmarking data
- the increase in reaction time to capture new trends and industry structural changes.
Increased reliance on administrative data sources may put the coherence, comparability and sustainability of the data at risk due to changes in programs, regulations or provider partners over time.

4.9 Option 5: Hybrid B

4.9.1 Key features

Hybrid B features a full decennial CEAG with a reduced quinquennial CEAG, coupled with increased use of administrative data and remote sensing. More specifically:

A CEAG would be conducted every five years:
- In years ending in "1" the questionnaire content would be similar to the current Canadian program, with the following modifications to reduce response burden:
  - The detailed expense questions would be replaced with taxation data (i.e. the CEAG would exclude these questions). Although expense questions represent 7% of the content of the questionnaire, their impact on the level of burden is much greater, due to the need to access reference documents and the potential sensitivity of the questions.
  - Any other questions that could be replaced with comparable and available administrative data would also be excluded.
- For the years ending in "6" a core CEAG would be defined by conducting user consultations and respondent testing. The content of the CEAG in years ending in "6" would be cut to a strict minimum (core) to provide
  - small area data
  - information on the structure of agriculture
  - data to use as benchmarks for required agriculture statistics
  - information required to maintain the frames necessary for the agricultural sample surveys.
    The core content needs to be determined in consultation with key stakeholders to identify priority data requirements; however there is potential to reduce current content by at least 50%. Over time, an increasing amount of content would be obtained through administrative sources rather than from a traditional CEAG.
- Critical data requirements that do not fit the core CEAG criteria could be obtained using a modular approach targeting only a subset of the population, (such as specific farm types or farms located in specific regions), and linking to the fully enumerated results (as recommended by the FAO).
The CEAG would continue to be linked to the CEPOP/NHS in years ending in "1" and "6" to provide socioeconomic data.
The survey program would remain commodity specific much like the current program. The number of survey occasions per year would be reduced for some crop, horticulture and livestock surveys.
Tax data would be used to replace all comparable financial questions on the CEAG and in surveys.
The target population would remain the same as for the current program (i.e. the target population would consist of all farms that produce agricultural products intended for sale). The CEAG would collect data for the entire target population.
The survey population would exclude, to a greater extent than currently, smaller farms under a specified threshold for reasons of burden and cost. To determine the optimal reduction in the survey population further investigation is required. The non-surveyed population would continue to be estimated (using statistical models largely based on CEAG data) and included in published estimates.
A regular frame maintenance program would include a short survey to complete missing information from farm tax filer records and other administrative data sources to update and maintain the agriculture frame on the Business Register.
Remote sensing would play an increasingly important role. This technology would be integrated into the agriculture statistics program as it becomes mature. The initial focus would be on replacing the national Potato Area Survey and the July and September Field Crop Reporting Surveys in the Prairie provinces.
Administrative data would play an increasingly important role. These data would be integrated into the agriculture statistics program, replacing content on surveys and censuses as they become available.
This option expands on current partnerships and promotes new partnerships with federal, provincial and industry stakeholders. These partnerships would be necessary to share responsibility for the development, collection, and compilation of administrative data and for Statistics Canada to obtain access (such as to AAFC's AgriInvest and AgriStability programs and livestock traceability data).

4.9.2 Strengths

This option meets user needs for small area, provincial benchmarking and critical survey frame information. These were identified as a major weakness in the other options.
This option allows for the survey population threshold for specific surveys to be raised because a quinquennial CEAG provides complete and regular data for modelling the non-surveyed population.
This option provides the best coverage of data users' needs, albeit less than the current model. In particular, the requirements for provincial benchmark data, small area data, data on rare and emerging commodities and the ability to perform cross-tabulation analysis are best met with this alternative.
The modular approach for non-core data in years ending in "6" (either concurrently or post-censally) has the flexibility to target only a subset of the population, (such as a specific farm type or farms located in specific regions), as recommended by the FAO.
This option provides an evolutionary approach to change within the agriculture statistics program, reducing risks to the relevance, coherence and accuracy of the program. It could also be implemented in a reasonable timeframe.
This option would achieve cost savings and response burden by replacing some of the content of the decennial CEAG in years ending in "1" and by reducing the content further to the core in CEAG years ending in "6." Research into administrative sources and consultation with users and stakeholders will provide the information required to quantify the savings and response burden reductions to be realized.
The current target population definition remains unchanged and therefore the coherence of the agriculture data over time is not affected. No investment would be required to adjust historical data for a new target population definition. Similarly, there would be no investment required to develop training material for users to avoid data misuse and misinterpretation and to clarify impacts of the change to the target population.
Since the annual agriculture surveys remain similar, this option is expected to have less impact on data users in terms of timeliness and content.

4.9.3 Weaknesses

The potential exists to increase response burden if coordination, technology and procedures are not well defined with administrative data providers. For example, data currently collected from a comparatively small sample of survey respondents represents the larger target population; however, if these same data had to be added to an administrative form, all program participants would be required to provide this information, thereby significantly increasing response burden. (An example of this would be in adding a data variable, currently collected on a sample survey, to the tax form where all farm tax filers would be required to provide it.) The goal is to reduce overall burden, not to simply transfer it from one organization to another.
Increased reliance on administrative data sources may put coherence, comparability and sustainability of the data at risk due to changes in programs, concepts, regulations or provider partners over time.

4.9.4 Essential conditions

To increase the use of administrative data, it would be necessary to renegotiate existing partnerships or develop new ones with federal, provincial and industry stakeholders. These agreements would establish protocols for data sharing, confidentiality and protection. The collaboration of multiple players over several jurisdictions will have to be established and maintained. This collaboration must begin at the highest level in the participating organizations and extend to the working level.
Federal, provincial and industry data holders would need to include a declaration to their data providers (farm operators) regarding the provision of data for statistical purposes. There may be a need to change legislation.
Agriculture respondents would have to be aware of and support the increasing use of administrative data, being aware of the associated benefits and risks.
A feasibility study would be required to fully evaluate the costs, benefits, risks and potential timeframes for incorporating administrative data sources and increased use of technology (such as remote sensing) into the program.
A methodologically sound, integrated and realistic framework through which new sources of administrative data could be identified, evaluated, incorporated and operationalized in the program must be developed to reduce the risk of error.

4.9.5 Required investments

Alternative sources of commodity data would have to be acquired, adapted and incorporated at the micro level (for example, program data such as crop insurance, AgriInvest and AgriStability).
Remote sensing would have to be developed to completely or partially replace traditional field crop surveys. A Land-Use Area Frame Survey would need to be developed. The data from this survey combined with administrative data (e.g., crop insurance data) would be used to calibrate remote sensing results.

4.9.6 Risks

There could be negative reaction from data users regarding
- the loss of some small area data
- the loss of some provincial benchmarking data.
This option does not immediately mitigate the risk of data loss should the CEAG be cancelled in years ending in "6" by Order in Council. Over time, this risk would be mitigated with increasing incorporation of data from administrative sources (including taxation data).

4.10 Summary of the hybrid options

The hybrid options were developed to capitalize on the most attractive features of the first three options, while minimizing those aspects that scored the least in the evaluation.

The principal difference between the two hybrid options is the frequency with which the CEAG is conducted. With Hybrid A, there would be no CEAG in years ending in "6," but the current commodity-specific surveys would be increased in content and sample size in those years.

With Hybrid B, a CEAG in years ending in "6" would be conducted, but would be reduced in size to core requirements providing

small area data
information on the structure of agriculture
data used as benchmarks (re-alignment) for current agriculture statistics
information required to maintain the agriculture frames necessary for the agricultural sample surveys.

The two hybrids were evaluated against the current program using the same method as the evaluation of the first three options. This exercise was undertaken primarily to determine whether a blend of the most attractive features of the first three options could adequately compensate for the absence of a quinquennial CEAG. With alternatives in place, would it be possible to conduct a full CEAG every ten years and continue to meet priority data requirements?

The alternatives presented in the first three options were found to be lacking in terms of their ability to adequately compensate for the loss of quinquennial CEAG data. For Hybrid A, another option was evaluated: that of increasing the sample size and content on the entire survey program during the CEAG years ending in "6."

Hybrid A would compensate in part for this loss by increasing the content and sample size of the entire survey program during the years ending in "6." With this option, no new surveys would have to be developed and therefore development costs would be minimized. Limited benchmarking would be possible, and keeping the same target and survey population definitions would provide some continuity to the time series data. Data would be released in the same timeframe as the current program, providing sufficient resources were available to process the larger volume of data.

Hybrid B, on the other hand, would provide comprehensive coverage of the entire population every five years. Hybrid B provides all of the advantages of Hybrid A in addition to providing an answer to the principal problems with the other options, namely:

it provides stability with regard to the target population definition, although the survey population definition should be studied with a view to reducing survey response burden
it provides for small area data at a frequency that users require
it provides provincial benchmarking data at a frequency that users require
it minimizes the delay in capturing new trends and structural changes in the industry
it builds on the already solid foundation of the current program thereby minimizing risks to the quality, relevance, coherence and accuracy of the program's data.

Table 3 summarizes the key features of the options explored and evaluated.

Based on these criteria, the option with the least response burden is Hybrid B. The most burdensome option is the Scandinavian Model because response burden is increased since the entire population of producers is required to provide data that are presently collected from a sample of producers. The burden on the organizations collecting the administrative data is also increased. Therefore, there is some shifting of costs and burden from Statistics Canada to other organizations.

Based on the cost criteria, the least costly option is the present Canadian program largely due to the fact that no additional costs are required to develop alternative collection vehicles. The most expensive option is the Scandinavian Model since aside from lowering Statistics Canada's collection costs this option scored higher for all of the remaining cost criteria.

Table 3: Summary of agriculture statistical program options evaluated

Report a problem or mistake on this page

Date modified:: 2023-01-23