Spatial Yield Risk Across Region, Crop and Aggregation Method

Abstract

A researcher interested in crop yield risk analysis often has to contend with a lack of field‐ or farm‐level data. While spatially aggregated yield data are often readily available from various agencies, aggregation distortions for farm‐level analysis may exist. This paper addresses how much aggregation distortion might be expected and whether findings are robust across wheat, canola and flax grown in two central Canadian production regions, differing mainly by rainfall, frost‐free growing days and soil type. Using Manitoba Crop Insurance Corporation data from 1980 to 1990, this research, regardless of crop or region analyzed, indicates that (i) spatial patterns in risk are absent; (ii) use of aggregate data overwhelmingly under‐estimates field‐level yield risk; and (iii) use of a relative risk measure compared to an absolute risk measure leads to slightly less aggregation distortion. Analysts interested in conducting farm‐level analysis using aggregate data are offered a range of adjustment factors to adjust for potential bias.