What is a Farmer and How Does the Government Count Them? The High Politics of a Seemingly Mundane Statistical Task

Sign up!

Email sent! Check your inbox to complete your signup.

Silvia Secchi is a natural resources economist and a Full Professor in the Department of Geographical and Sustainability Sciences at the University of Iowa.

Editor's Note: Dear readers, Nathan here. A few months ago I came across a paper by professor Secchi on social media which was just published in the journal “Agriculture and Human Values”. I thought it was quite interesting and clearly written and wanted an even more popularized version for Notes on the Crises. I’m very happy that this guest piece came together. At a time when the Trump administration seems poised to hand out even more funds to farmers while it further guts our statistical agencies, I think this topic is more important than ever. It is also a great exercise in thinking critically about the processes that go underneath the statistics we throw around. Enjoy!

The way we collect and present statistics on economic activities is always a political choice. In the US, historically, data has generally been properly gathered according to statistical principles. But the questions asked, the categories chosen to group information. and who (and what) is included (or omitted) have always been decisions with political implications. These methods are driven by political forces, rather than neutral “objective” choices (whatever people may mean when they say that). Researchers are rarely truly impartial, whatever their best intentions. One of the best ways I can think of illustrating this is the 1890 US Census map that shows Indian Territory as having no population. And if you think that type of omission is in the past, you may be surprised to find that the Census of Agriculture did not count Native American farmers in reservations until 2007. Prior to that point, collectively-owned reservation farms were considered “abnormal” (which in and of itself is a rather loaded term!)  

As my example clearly indicates, farm statistics specifically are exceptionally political dynamite in the United States. Let’s consider how the category “farmer” interacts with other key categories (gender and ethnicity). Some readers might have the kneejerk reaction that I’m “making” farming about gender or ethnicity. Why should it matter whether the farmer counted is a woman or not? Well, as it turns out, we need to highlight gender and ethnicity to make sure these farmers are counted at all. Women farmers were not counted in the Census of Agriculture until 1978, which was also the first time that “Spanish origin” was considered separately from race. 

The implicit gender and ethnic assumptions about who counted as farmers are quite easy to observe. The “ideal” farmer, or at least the easiest one to count, is a white guy- not according to me, but according to more than a century of USDA practice. Yet equally, we have evidence of other modes of farming being “overcounted”, generating a falsely rosy picture of diversity breakthroughs in rural regions. Though the Census has become more inclusive in terms of gender and race, agriculture remains a largely white and male dominated profession. A recent paper estimates that in 2012-2014 white people owned 98% of all farmland and operated 94% of it, and women are much less likely to run an operation on their own than men. 

In fact, one could say that today the Census of Agriculture has become too inclusive: since 1974, it considers a farm any operation that has the potential to produce at least $1,000 in sales value, and since the last census it counts up to four operators per farm. 

The result of this combination of choices is that agriculture appears more diverse (and proportionally composed of small family farms) than it actually is. In the current political climate, diversity may no longer be valued (USDA has already announced it will no longer consider categories of socially disadvantaged farmers in making policy decisions). But nevertheless inflating the number of farms and farmers is a longstanding bipartisan choice unlikely to be challenged. That’s thanks to the power of the agricultural lobby and longstanding narratives glorifying US farmers as the stewards of the land feeding the world. These pleasant sounding fables harken back all the way to manifest destiny, and the Jeffersonian ideal of the yeoman farmer. The reality is considerably more exploitative, and concentrated in the hands of relatively few larger landowners.

Favorable tax provisions allow farm losses to be offset against off-farm income and inheritance tax laws shield capital gains made on owned farmland if the land is passed to heirs. Through these mechanisms we have made the basic reality of farming in the 21st century United States imperceptible- at least in the headline statistics. We count as small family farmers people who (totally legally, I want to make clear) personally benefit from being considered farmers, but who have no interest in operating a commercial enterprise. Their inclusion in the Census of Agriculture masks the level of consolidation in US agriculture, by lowering the value of aggregate statistics, and bolsters the number of small farmers. The last census of agriculture gave another assist to the agricultural lobby by eliminating the correspondence between one primary operator and a farm (that had existed since the 19^th century). That means from now on we will likely see increases in the number of operators, even as the number of farms decreases.  

To give a sense of the dimensions of this problem, in the latest Census of Agriculture 20% of farms sold less than $1,000 worth of agricultural products, and another 20% sold between $1,000 and $4,999. A full 65% of US farms sell less than $25,000. More than half the farms in Colorado, Connecticut, Florida, New Jersey, New Hampshire, New Mexico, Oregon, South Carolina, Tennessee, Texas, and West Virginia sell less than $2,500. In Arizona, a whopping 70% of farms sell less than $2,500 worth of agricultural products. This is clear evidence that ranchette owners are largely counted as farmers: 55% of Arizona farms are below nine acres. In contrast, the state’s crop extension budgets are based on a 1,500-acre farm with 1,000 acres in crops. In the Census of Agriculture, there are only 230 farms that would meet that criteria, just 1.4% of the total. 

When I started working on this issue I was driven by my observation that the general public, many policy makers and other stakeholders are not aware of the misleading issue associated with agricultural statistic averages. As I dug into the literature, I confess I was rather surprised to see that, while there had been very lively and spirited discussions among agricultural economists about how to count farms until the 1960s, the literature largely stopped with Prof. Richard Foote. Foote wrote a thoroughly researched report for USDA about the farm classification issue in 1970. His study included a terrific literature review and included some still valid suggestions, such as eliminating the use of the “family farm” in classifying farms (another rather loaded term with limited usefulness). 

He also introduced the category “manufacturers of agricultural products”, to better reflect activities such as greenhouses and confined livestock production. The Foote report was not used to modify the definition of farms in the 1974 census, and in fact the $1,000 potential sales threshold is what we still use fifty years later. As concentration has intensified, agricultural economists have remained largely silent about the inadequacy and outdatedness of our classifications—about as silent as they have been on how white-dominated agriculture and farmland ownership are in the United States (this revealing study was written by two Portland State University planners). I found two exceptions: in 2009, USDA’s Economic Research Service produced a report about “Exploring Alternative Farm Definitions”, which was driven by conversations in Congress about Farm Bill eligibility and resulted in no changes, and in 2019 the National Academies of Sciences Engineering and Medicine published a report on “Improving data collection and measurement of complex farms”. 

The 2019 report contained this realpolitik gem: “Even if a farmer earns all of his or her income outside farming and makes a loss running a farm—perhaps to enjoy the living environment, or to benefit from certain social security payments or tax facilities—he or she is still a farmer (producer)”. If we used this kind of definition for other jobs, we would be adding millions of bakers, carpenters and plumbers, let alone daycare workers and home healthcare aids, to the US workforce. It would be a shift on par with Trump’s latest BLS decisions. This loud silence on the increasing outdatedness of our agricultural statistics is also evidenced by the Economic Research Service shift from the use of the term “residential/lifestyle” to “off-farm occupation” farms in 2011. 

The political economy of these choices is clear – after all, the US Census Bureau lost the Census of Agriculture to USDA’s National Agricultural Statistical Service in 1996 because it threatened to increase the $1,000 threshold to $10,000 in response to funding cuts. So USDA got the census and no cuts, the $1,000 threshold is still going strong, and the conversation on these issues at land grant universities is close to non-existent. It may be worthwhile to point out that such universities are beneficiaries of a variety of federal programs for research and extension activities that are partly tied to the number of farms in their state. As the number of farms keeps shrinking, it may not be advantageous to point out that it should be reduced even further (and differentially across states). 

At the same time, we have very little information at the national level on agricultural landowners, the majority of whom are non-operators, that is, not involved in farming. In the United States, agricultural land tenancy is no longer prevalent as it was in the earlier part of the 20^th Century. Full tenants went from 42% of all farms in 1935 to 6% in 2022. Most farms are operated by part-owners who also rent land from non-operators. There are multiple reasons for this, and consolidation is certainly a dominant one. However, our tax laws also play an important role: when farmland is transferred through inheritance, the beneficiaries receive a stepped-up basis, that is, they do not have to pay capital gains on the difference between the current market value of the farm and the value it had when the deceased acquired it. This creates incentives not to sell farmland, particularly since the 1980s farm crisis, as agricultural subsidies have consistently bid up farmland prices. Even though the number of non-operating landowner “entities” (individuals, partnerships, corporations and trusts) at the time of the last survey, 2014, was over 1.8 million, close to the 1.9 million farms in the latest Census of Agriculture, landowners are surveyed every 10 years, and there is much less attention being paid to their characteristics and behaviors, and, for example, to the racialized generational wealth processes that our policy environment is strongly incentivizing.  

The statistics gathered about another critical category of agricultural stakeholders, farmworkers, are also revelatory. The two farm labor surveys only include part of the farmworker population: the Department of Labor’s National Agricultural Workers Survey (NAWS), excludes workers employed on livestock farms and H-2A guest workers on crop farms. USDA’s Agricultural Labor Survey, which is directed at operators, not farm workers, excludes contract laborers. This means that there is no consistent set of statistics on farmworkers and their conditions, which are some of the worst for labor in the US. Besides their longstanding exemption from many forms of protection that goes back to the New Deal and the worsening of their exploitation during covid now we have the administration’s attacks on undocumented people. As a result of this lack of consistent data gathering and the challenges in reaching farmworkers, the estimates of their numbers vary widely. The Census of Agriculture in 2022 estimated almost 2.2 million hired workers, with over a million of them working over 150 days on farms. But USDA’s Economic Research Service estimates the number of farmworkers to be 1.1 million that same year, and the Agricultural Labor Survey reported 680,000 workers. 

The evidence is quite clear. The statistical spotlight is on farmers – it is so large and powerful it captures people that in other sectors would not be counted. And it clearly does so to overestimate the importance of a powerful political constituency. The latest changes in the Census of Agriculture further expand the number of “farmers”, thus masking a reality of consolidation and discrimination. We know far less about landowners—a constituency not less powerful and white-dominated as “farmers”—but one that does not fulfill the same glorified role in the collective imagination and is thus not worthy of extensive study. Farmworkers and their challenges are also largely understudied, and so remain poorly understood. Unfortunately, the Trump administration is accelerating this trend by creating a climate of fear among farmworkers, and withdrawing funds for more innovative and inclusive research. 

Economic statistics in general are under unprecedented threat in the United States. The concern about their integrity goes beyond the economics profession, and academic studies—the dismantling of the mechanisms to produce robust, science-based and independent estimates threatens to undermine the very foundations of the US economy. However, we should not be blind or naïve to the processes that have determined what data we collect (how and why.) In fact, I would argue that a general lack of consideration of political economy issues writ large on the part of academics, experts and the press is part of the reason why we are where we are. In regards to statistics specifically, ignoring the historical context, making generic claims of “neutrality” and “objectivity” and uncritically accepting status quos that have become more and more detached from reality—contributing to a false sense of security and collegiality. We cannot afford silence and passivity any more right now.