Abstract: Forecasting Risk and the Risk of Forecasting in the American Cotton Market, 1865-1905
The primary way of knowing risk in the late nineteenth-century cotton market was through the production of agricultural statistics—by estimating present acreage and condition and forecasting yield. Beginning in 1863, the U.S. Department of Agriculture defined the purpose of its systematic crop reporting as stabilizing agricultural commodity markets, whereas private forecasters were commonly accused of doing the opposite. The USDA' s competition with private forecasters inspired fervent debates over whether objectivity was possible in the production of cotton statistics. This paper traces such debates over objectivity in cotton forecasting in the late nineteenth- and early twentieth-century United States and illustrates the economic and epistemological challenges of knowing risk in the postbellum cotton economy. A multiplicity of public and private cotton forecasts had the ironic and unintended consequence of producing a perception of cotton statistics as a mechanism that exacerbated rather than mitigated risk in a volatile early twentieth-century cotton economy. After the events of the 1905 cotton season made visible the economic risks of forecasting, the USDA's Bureau of Statistics began to redefine its work by shifting away from its professed statistical objectivity and toward an acceptance of producerist bias in government agricultural statistics.