Background/Question/Methods Agricultural production has increased dramatically in the past 50 years, supported, in part, by the simplification of agricultural landscapes. While the benefits of increased food production are difficult to dispute, simplification, both at the local and landscape level, has fueled declines in biodiversity and ecosystem services. In addition to the concerns that this loss of complexity necessitates higher levels of pesticide use in general, local and landscape simplification may also increase pest outbreaks and consequently, infrequent but particularly high pesticide use with potentially damaging consequences for the environment and human health. This contribution pairs intuition from ecology with methods from agricultural economics to derive unique insight into how crop diversity, surrounding cropland extent and field size affects the stability of agricultural pest control. To do so, we combine refined, field-level data on crops and insect pest control for 150,000 field-year observations from Kern County, CA with panel data approaches. Results/Conclusions We find increasing cropland in the landscape and larger fields generally increases the level and variability of pesticides, while crop diversity has the opposite effect, as predicted by ecological theory. In all cases, accounting for non-random planting decisions and farmer-specific behavior strongly influences the magnitude of the estimated statistical relationships. This suggests that while complexity increases stability and reduces high deviations of insecticide use, accounting for crop and farmer-specific characteristics is crucial for statistical inference and sound scientific understanding.