Gardner begins with technology and productivity improvements, using the introduction of tractors and hybrid corn as case studies. He covers changes in farm numbers and sizes, and in farm incomes and distribution. And he considers broader aspects of farm communities, such as age distribution, schooling, out-migration and off-farm work, integration with towns, and quality of life.
A long chapter on markets focuses on commodities but also considers markets in inputs, in credit and labor. Two chapters explore the role of government: the first covering infrastructure, loan subsidies, and agricultural research and the second commodity prices, subsidies, stockpiling, and insurance.
Looking for explanations, Gardner surveys the different attempts to explain the standout feature of all this data, the rise in productivity and output that started between 1935 and 1940 and has continued pretty much since. These theories invoke research and education, the New Deal programs, specialisation and scale, and other factors.
"It is unlikely that any one of the causal factors that have been considered can qualify as the explanation for the takeoff in productivity, or for the longer-term history of technological change in U.S. agriculture."
Moving beyond case studies and narrowly restricted analysis to a "systematic statistical investigation of interactions among the hypothetically related variables" has not been possible, as "the time series data are too dominated by trends, with a few structural changes, that result in too few independent annual observations to identify the relationships among them".
So Gardner turns to state and county level data, focusing on the same trends, the growth in productivity and output and the growth in household income. The data here still do not allow "a definitive contest between different rival hypotheses for the acceleration of productivity growth in the late 1930s". Nor do they provide a definitive evaluation of the effects of commodity programs. But the growth of farm incomes is more amenable to explanation:
"the most important factors in the growth of farmers' incomes are not specifically agricultural at all. They key is rather the economic adjustments of farm people to the opportunities afforded by the nonfarm economy."
What little technical apparatus Gardner uses is explained as he goes. So the first chapter considers the problems of constructing aggregate indices, the choice of end-point, and related issues, and includes a numerical worked example. The idea of a deadweight loss is explained in the chapter on government commodity subsidies. And so forth. One exception to this is that Gardner makes no attempt to explain regressions, which get a workout in the chapters analysing state and county level data.
The data are linked to social history and politics, but only superficially. And Gardner only touches narrowly on environmental costs, with nothing at all about greenhouse emissions, to take one topical example. But the biggest limitation is perhaps the restriction to those topics where good time series data is available, either throughout the 20th century or in some cases for its second half.
By avoiding technical details, American Agriculture in the Twentieth Century may frustrate economists. Its narrow focus may limit its appeal to students or historians of agriculture, but it will be ideal for those wondering what can be done with the long time scale quantitative data.