Online Program

Return to main conference page

All Times ET

Program is Subject to Change

Tuesday, June 15
Tue, Jun 15, 11:30 AM - 1:00 PM
Topics in the Collection, Production, and Estimation of Short Term and other Business Statistics

Transitioning to Model-Based Official Statistics: the Case of Crops County Estimates (308431)

*Lu Chen, NISS 
Nathan Cruze, NASS 
Linda Young, USDA, National Agricultural Statistics Service 

Keywords: Agriculture Surveys; Official Statistics; Small Area Estimation; Auxiliary data; Constraints; Benchmarking

The US Department of Agriculture’s (USDA) National Agricultural Statistics Services (NASS) publishes county-level estimates of planted acreage, harvested acreage, yield, and production annually. These official statistics are important to farmers, ranchers, and other Federal agencies for planning and decision making. Given the importance of the crops county estimates program, NASS engaged a panel of experts under the National Academies of Sciences, Engineering, and Medicine (NASEM) for guidance and recommendations on implementing small area models for integrating multiple sources of information to provide more precise county-level crop estimates with measures of uncertainty. One major challenge of the production of a model-based approach is the ability to provide reliable county-level estimates that satisfy important relationships nested among them. For example, the county-level harvested acreages should not exceed the planted acreages in that county and they must also add up to the state-level estimate. In addition, the yield and production estimates must be provided if the corresponding harvested estimates exist. In this poster, the development and procedure of model-based crops county-level estimates are discussed based on 2015 corn data in Illinois. The example shows that, by implementing three separate hierarchical Bayesian models, all county-level estimates of planted acreage, harvested acreage, yield, and production satisfying important relationships can be provided along with their associated measures of uncertainty.