The Government Accountability Office has released IRS Could Significantly Increase Revenues by Better Targeting Enforcement Resources:(GAO-13-151):
The IRS spends most of its enforcement resources on examinations. Correspondence exams of individual tax returns, which target fewer and simpler compliance issues, are significantly less costly on average than the broader and more complex field exams. GAO estimated that the average cost (including overhead) of correspondence exams opened in 2007 and 2008 was $274, compared to an average of $2,278 for field exams. IRS spent almost 20 percent of the $1.6 billion per year that it devoted to exams on returns from taxpayers with positive income of at least $200,000, even though such returns accounted for only 3 percent of the 136 million individual returns filed per year. (Positive income, a measure that IRS uses to classify returns for exam planning purposes, disregards losses that may offset this income).
GAO estimated that, for the 2 years of cases reviewed, correspondence exams were significantly more productive in terms of direct revenue produced per dollar of cost than field exams. Both types of exams of taxpayers with positive incomes of at least $200,000 were significantly more productive than exams of lower-income taxpayers.
GAO demonstrated how these estimates could be used to inform resource allocation decisions. For example, a hypothetical shift of a small share of resources (about $124 million) from exams of tax returns in less productive groups shown in the figure to exams in the more productive groups could have increased direct revenue by $1 billion over the $5.5 billion per year IRS actually collected (as long as the average ratio of direct revenue to cost for each category of returns did not change). These gains would recur annually, relative to the revenue that IRS would collect if it did not change its resource allocation. This particular resource shift would not reduce exam coverage rates significantly and, therefore, should have little, if any, negative effect on voluntary compliance.