Is Your Agency Too Conservative? Deriving More Reliable Confidence Intervals

• Published in: The journal of government financial management Vol. 54; no. 2; p. 30
• Main Authors: Kvanli, Alan H, Schauer, Robert
• Format: Journal Article
• Language: English
• Published: Alexandria Association of Government Accountants 01.07.2005
• Subjects: Accounting records; Audits; Confidence intervals; Conservatism; Councils; Financial management; Government agencies; Governmental accounting; Tax underpayment; Taxes; United States > US
• ISSN: 1533-1385

With the advent of the computer, many theoretical advances in statistical analysis have become practically feasible. These developments, when viewed in the backdrop of statistical study, are emerging at an amazing pace. However, in our opinion, few government agencies have used these new approaches in evaluating random samples in such areas as projecting total tax underpayments or projecting monetary recovery amounts. This article sheds some light on these new approaches, which are more efficient and could potentially raise tax revenue. Accountants, tax audit agencies, and other related individuals or groups tend to be conservative. This is generally for good reason. Nevertheless, excessive conservatism can be problematic. As we have demonstrated, being overly conservative can mean that tax revenues are needlessly reduced. The newer methods offer more realistic lower limits and higher tax revenues. Any agency relying on lower limits should take a serious look at the Bootstrap procedure, and even more so, the Empirical Likelihood procedure introduced in this article.