Abstract:
|
Abstract
The Census Bureau obtains administrative receipts data from the Internal Revenue Service (IRS) for corporations, partnerships, and sole proprietorships through the Business Register (BR). Economic Surveys utilize administrative receipts to substitute for non-respondent survey receipts or to supplement survey-collected receipts. Unedited BR administrative receipts are, at times, a poor substitute for survey-collected receipts . This paper describes the methodology devised to edit administrative receipts for single-location (single-unit) businesses. We used the Hidiroglou-Bertholet (HB) edit to flag units with inconsistent receipts to payroll ratios based on a standardized HB "score" statistic. In order to define boundaries for HB score cutoffs, we devised a logistic regression model that divided the data into three categories: 1.Use for tabulation and imputation 2. Use for tabulation but not for imputation 3. Do not use for tabulation or for imputation.
|