A note on the role of survey data and expert opinion in constructing input-output tables

This note addresses the role of survey information and expert opinion in constructing input-output tables. In contrast to assuming polar positions between tables based on expert opinion and tables based on empirical observation, we argue that the distinction is not well-defined. We provide, as evidence, a comparison of U.S. Bureau of Economic Analysis (BEA) data on the use of commodities by industries with corresponding Bureau of the Census (BOC) Census of Manufacturers data on selected use of materials by industry (published as Census Table 7). Results show that changes in BEA data cannot be explained by corresponding changes in the Census data on which they are founded.