Data Used in General Revenue Sharing

Abstract

The GRS allocation formulas involve a large number of data elements (DE’s). This chapter develops error models for each of the data elements. These error models will be used in chapter 6 to calculate the expected losses in GRS allocations, but the models are also of independent interest because the methodology extends naturally to other kinds of data as well. Construction of error models is a difficult undertaking because the mechanisms giving rise to errors in data are complicated and not generally understood. Of these mechanisms random sampling is typically the simplest and often the least important. To derive the error model for a data element we proceed in five steps: (1) notation; (2) definition and data sources; (3) data quality: evidence; (4) data quality: assumptions and conclusions; and (5) relevant distributions and approximations.