Farmland and Cropland Concentration Measures: Data Documentation and Methods

According to recent Census of Agriculture data, the number of large farms and very small farms has increased, while the number of small to midsized farms has declined. The changing size distribution of farms makes it difficult to capture trends in the concentration of production using common measures, such as the average or median farm size. This data product provides a size measure (the acre-weighted median for cropland and for farmland) that reflects the increasing concentration of production on large farms. The acre-weighted median, here provided at the county level, is calculated by ordering farms from smallest to largest and picking the farm size at the middle acre (the standard median focuses on the middle farm). Half of all land is on farms smaller than the acre-weighted median, and half of land is on bigger farms. A county's acre-weighted median better reflects the size of operations where most production occurs.

Here we document two measures of concentration, constructed for each county from individual records for each of five Censuses of Agriculture: 1982, 1987, 1992, 1997, and 2002. For each county and year, farmland concentration is defined as the acre-weighted median of total land in farms. Intuitively, this measure is the farm size (in acres) for which half the farmland area in the county occurs on larger farms and half occurs on smaller farms. The measure is constructed by ordering farms from smallest to largest, finding the cumulative sum of acres along this ordering, and selecting the farm size at the point where the cumulative sum equals half the total number of farmland acres in the county. Observations are not included for counties where half or more of the total land in farms occurs on a single farm, or if the Census of Agriculture did not disclose total land in farms for other reasons.

Cropland concentration is defined analogously to farmland concentration, except that only cropland acres are considered (farmland also includes pasture, range, woodland, etc.). The measure of cropland used here equals cropland acres harvested plus cropland on which all crops failed. Thus, cropland concentration is the size (measured in acres) for which half the county's cropland occurs on farms with more cropland acres and half occurs on farms with less cropland. Observations are not included for counties where half or more of the cropland occurs on a single farm, or if the Census of Agriculture did not disclose total cropland harvested for other reasons.

Nonresponse Weights and Changes in 2002

For each census year and county, nonresponse weights are used to adjust Census statistics for farms that did not return a census form. Beginning in 2002, the sample frame and non-response weights used in the Census of Agriculture changed significantly. This change increased the total farm count as reported by the Census of Agriculture. To a lesser degree, the change also influences our measures of concentration. The data on concentration measures for 2002 are provided two ways, using the old (pre-2002) and new (2002) nonresponse weights. When making comparisons with earlier years, it is better to use the pre-2002 weighted measures.

For more information about ERS research using these measures of concentration in research, see the Amber Waves data feature, Measures of Trends in Farm Size Tell Differing Stories, and the article, Cropland Concentrating Faster Where Payments Are Higher.