Commentary: Why Congress Should Empower the IRS to Certify Tax Preparers: The Debate Continues
The libertarian Institute for Justice's Dan Alban recently testified before the Senate Finance Committee that Congress should not grant the IRS the authority to require commercial tax preparers to pass a test of basic competence in tax rules and stay current with changes in those rules. I followed up with a blog post in which I contested a key argument that Mr. Alban made and the evidence he cited for it. He responded with a series of comments and, as part of them, suggested we have a substantive debate on this matter. In that spirit, I'm following up with this commentary.
For starters, I withdraw any suggestion in my blog post that Mr. Alban may have knowingly misled the Finance Committee when he testified that "IRS trained-and-certified preparers in the VITA volunteer program were found by the Treasury Inspector General for Tax Administration (TIGTA) to have a 61% error rate in 2011." But while he evidently believes that's a valid statement, it surely is not. In fact, it's substantively indefensible. Mr. Alban's very source, the Treasury Inspector General for Tax Administration (TIGTA), said that no one should draw the very conclusion that Mr. Alban does. Consider:
- The TIGTA study on which Mr. Alban bases his claim of a 61 percent error rate considered just 36 hypothetical tax returns for the entire country -- far too few for a statistically valid sample. Moreover, TIGTA says that these 36 returns, which reflected three tax-return scenarios that TIGTA designed, were tested at "judgmentally selected" VITA sites -- not randomly selected sites -- which also is incompatible with basic statistical procedures to determine a valid national error rate.
- Mr. Alban tries to dispute the IRS observation that the 36 tax returns on which TIGTA tested VITA preparers involved uncommon, complicated, more error-prone tax scenarios that do not represent the universe of returns that VITA sites handle. He cites TIGTA's statement that the tax-return scenarios it tested reflected types of returns that VITA sites handle. Yes, VITA sites do handle some complicated scenarios like the hypothetical ones that TIGTA constructed. But these scenarios are uncommon and are not representative of the tax returns that VITA sites prepare. For example, the IRS notes that tax returns of self-employed individuals who operate a small business and file a Schedule C -- a category of returns with a notoriously high error rate -- account for 4 percent of all returns that VITA sites handle, but 33 percent of TIGTA's 36 hypothetical returns. Similarly, another 33 percent of the TIGTA returns involved a taxpayer who paid child support and incurred dependent care expenses and contributed to a 401(k) plan; but the percentage of returns that VITA sites handle with all three of these elements is far below 33 percent; and the other 33 percent of TIGTA's returns all included contributions to a Health Savings Account, which few low-income households make.
- TIGTA's review of VITA sites of one year earlier (in 2010) found mistakes in only 10 percent of TIGTA's tax returns. TIGTA subsequently changed the three scenarios for its 36 hypothetical tax returns and, consequently, the share of returns on which VITA sites made mistakes shot way up.
Clearly, it's statistically and analytically invalid to claim, as Mr. Alban does, that TIGTA's exercise with these 36 atypical tax returns show that VITA sites nationally have a 61 percent error rate. But, don't take my word for it. The IRS and TIGTA itself made the same point. In its 2011 report, TIGTA wrote, "the accuracy rate should not be projected to the entire population of tax returns prepared at the Volunteer Program sites." TIGTA reiterated this warning in its 2013 report (which used the same methodology), cautioning, "Our results cannot be used to project to the population of tax returns prepared at VITA sites because the tax returns we had prepared were included as part of a judgmental sample. A judgmental sample is a nonstatistical sample, the results of which cannot be used to project to the population." Similarly, the IRS said, "the 36 cases that TIGTA included in their 2011 shopping review only reflect the quality of service provided for those 36 cases."
Mr. Alban also asserts that TIGTA's method is much like what the Government Accountability Office (GAO) used for its recent examination of commercial tax preparers, and that we at CBPP cited the GAO study in a blog post. He's off base here as well. In its second paragraph, GAO's report states, "GAO found significant preparer errors during undercover site visits to 19 randomly selected preparers -- a sample which cannot be generalized" (emphasis added). Nether GAO, nor we in referencing the GAO report, ever used that report to construct a national error rate for paid preparers. GAO and we recognize that doing so would be invalid.
Mr. Alban further argues that, to project a national error rate for VITA sites, TIGTA's study is better than the IRS' Quality Statistical Sample (QSS) reviews. Yet the QSS review involves a much larger, statistically valid sample of actual randomly selected tax returns from randomly selected VITA sites, with the sampling methodology designed by IRS' highly professional Statistics of Income division.
To be sure, Mr. Alban makes a valid point that VITA volunteers know when IRS reviewers are present and that may be a factor in the QSS's low error rates for VITA sites -- a 13 percent error rate in 2011 and 9 percent in 2012. VITA volunteers may be more careful when IRS reviewers are present. (Having said that, the VITA tax preparers are unpaid volunteers whose jobs and pay do not depend on what the IRS reviewers find and who are accustomed to routine review of their work by experienced preparers at their VITA site).
Nevertheless, in contrast to TIGTA's tiny sample of 36 unrepresentative tax returns, the QSS uses a statistically valid approach to derive its results. Moreover, the QSS findings undermine Mr. Alban's contention that the IRS training and certification that VITA tax preparers receive is essentially worthless. They show that VITA volunteers were trained well enough to achieve roughly a 90 percent accuracy rate when they were being most careful (i.e., when IRS reviewers were present).
So, here's the bottom line:
We know that the Treasury and taxpayers are hurt by errors on tax returns, including errors in the Earned Income Tax Credit (EITC). We also know from other IRS research that error rates are especially high among paid tax preparers who are neither certified (they're not lawyers, CPAs, enrolled agents, or the like) nor affiliated with a national tax preparation firm (they often work alone without supervision): the IRS' National Research Project's statistically valid audit of EITC returns (from which widely cited error rates for the EITC are drawn) found a 49 percent error rate and a 33 percent EITC overpayment rate among the tax returns that these preparers handled. To improve accuracy and avert billions of dollars in losses, the IRS sought to require commercial preparers to pass a certification test showing basic competence and to stay current with tax law changes through annual continuing education courses. Mr. Alban brought a lawsuit to block these measures and prevailed, with the courts ruling that the IRS must first get the requisite legal authority from Congress.
The question now thus is whether Congress should provide such authority. Mr. Alban emphatically says no. The IRS National Taxpayer Advocate Nina Olson, however, notes that even hairdressers generally must receive training and certification to engage in their work. With billions of dollars on the line, is it too much to ask that the IRS ensure that commercial tax preparers have basic competence in the tax rules?