Climate adjusted productivity on Australian cropping farms

This chapter presents a method of controlling for climate effects on firm-level productivity measures. The approach involves applying non-parametric (machine learning) regression methods to large spatio-temporal data sets where firm productivity is observed for many time periods and locations. This method is applied to the Australian broadacre cropping industry, using Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) farm survey data for the period 1977–78 to 2014–15 and climate data from the Australian Water Availability Project (AWAP). The study demonstrates the importance of controlling for climate variability and change when measuring productivity in agriculture, particularly on cropping farms. The results show a significant deterioration in climate conditions for cropping over the last 15 to 20 years, particularly in southern Australia. After the effects of climate have been removed a clear picture of underlying productivity trends emerges, with a slowdown in productivity growth after 1993–94 and a rebound since 2007–08.