A unified framework for calculating aggregate commodity prices from a census dataset

Economic data collection from commodities producers in the United States typically consists of revenues and quantities. While the data collected in some sectors such as fisheries are a census of the population, features of the population such as prices, must be calculated. Unit values are widely used as a price measure to impose a single price in place of dispersed ratios of revenue to quantity from individual producers but alternatives exist. In this paper, different linear aggregation procedures are used to calculate price measures, such as ratio-based calculations (e.g., ratio-of-means, mean-of-ratios), or estimation by ordinary least squares. There are non-trivial differences in the prices calculated depending on the procedure. This paper proposes a unified framework, including Bayesian estimation, for considering the tradeoffs inherent in the different methods commonly employed.