Abstract

Transforming Agriculture Statistics : A Modernization initiative within Statistics Canada aims to use alternative data and advanced technologies to reduce response burden on farmers:

Agriculture plays a critical role in Canada’s economy and food system, with Statistics Canada’s Agriculture Statistics Program producing over 100 annual data releases across 30 active surveys. These estimates inform key questions on food supply, food security, trade, and policy. Yet like many statistical agencies worldwide, Statistics Canada faces growing challenges: declining survey response rates, rising costs, and increasing pressure to minimize respondent burden.

In response, the Agriculture Statistics Program launched AgZero, a modernization initiative that set a goal to move beyond a survey-first approach by replacing agricultural surveys with more innovative data collection methods. The initiative aims to use alternative data sources and advanced technologies, such as Earth Observation data and machine learning, to reduce the response burden on farmers to as close to zero as possible. This modernization initiative is founded on the “collect once and use multiple times” principle and seeks opportunities to respond faster to the emerging needs of data users.

Now several years into implementation, AgZero offers both evidence of success and important lessons learned. This paper will highlight four projects that have moved from concept to production, each deriving statistical estimates with zero contact to farm operators: (1) Agriculture Labour Statistics Program, producing census-level estimates from administrative tax files; (2) Interprovincial Hog Movements, leveraging Canadian Pork Council data to measure hog flows between provinces; (3) Crop Yield Modelling, integrating remote sensing, agroclimatic indicators, administrative data and survey inputs to replace survey-based yield estimates in the summer occasions of the Field Crop Reporting Series; and (4) March and July Ending Stocks Model, combining survey and administrative sources to replace two on-farm stocks survey occasion. Together, these projects have eliminated multiple survey occasions and filled long standing data gaps while sustaining or improving data quality.

The paper will also examine the trade-offs of an administrative data first strategy compared to a traditional survey-first approach. Key lessons will be discussed, including methodological advances, data governance considerations, importance of collaboration with partners, and the operational challenges of running and maintaining model-based production systems.

Looking forward, AgZero is exploring further survey replacements, expanded use of administrative data linkages, and greater integration of alternative methods. At the same time, there are some risks that remain including administrative data access limitations, methodological uncertainty, and the challenge of balancing modernization with the stability required of an official statistical program. By reflecting on both successes and limitations, this paper contributes to the broader international conversation on the future of agricultural statistics and the role of modernization in ensuring sustainable, high-quality, and low-burden data systems.