Medium to large firms are typically audited on three-year cycles for sales and use tax compliance by states collecting sales tax. Audits of purchases frequently involve populations containing several hundred thousand transactions. Audits of sales can involve millions of transactions, especially for large retail corporations. Large size sample audits require considerable expenditures of financial resources and often produce audits that require more than a year to complete. Consequently, using efficient sampling designs and optimal estimation methods are of paramount importance from a cost/benefit viewpoint.
This paper presents methods for using Bayesian estimation techniques to improve the sampling efficiency and estimation precision in current sample audits by incorporating the results from the previous audit cycle. Using theoretical analysis and Monte Carlo simulation, significant gains in precision are shown to be achievable under a broad range of population characteristics and factors. These gains promise to produce substantial cost savings for the jurisdictions and for the firms undergoing the sample audits.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.