Monday, November 16, 2009

Compensating audited taxpayers: an idea whose time has come?

Ian Ayres and Barry Nalebuff, two Yale economists who write for Forbes, have proposed that the government pay taxpayers whose returns are selected for audit, especially taxpayers subjected to random exhaustive audits the IRS may conduct purely for research purposes.

Peter Pappas doesn't think much of the idea:

I’ll research this more, but here’s my initial response to Ayres and Nalebuff: What the hell have you two been smoking?

First, the IRS can simply pass these audit compensation fees back to the taxpayer by more aggressively disallowing tax deductions. Second, because IRS auditors and their group managers have discretion as to whether or not to assess penalties against taxpayers it’s reasonable to assume that they they will be more aggressive in assessing those penalties and denying reasonable cause abatements so as to offset the cost of compensating the auditees.

I hope American taxpayers are not stupid enough to trust a program that purports to compensate them for being audited when the compensator, the IRS, is the entity responsible for determining the amount of their indebtedness to the government.

Am I missing something here?

Ahem, well Peter, it wouldn't be the first time you've been missing something.

Economists have been talking about the idea of compensating audited taxpayers who are found to be fully compliant for their time and trouble for decades, probably since before you were born. I first encountered the idea in the 1970s, and it was not a new idea even then. I will admit to some mixed feelings about the idea, but I don't think you have to be "smoking something" to think it's an idea that deserves some consideration.

I don't know Ian Ayres (who happens to be a lawyer as well as an economist), but I spent a fair amount of time talking to Barry Nalebuff back in the 1980s. He is not your typical ivory tower economist. He knows a lot of mathematical economics (game theory and that sort of thing--he was a Rhodes Scholar and a former Harvard Junior Fellow, both very big deals), but he also knows and cares a lot about the real world. He runs a sideline business, which he and a former student founded in 1998. I understand it's quite a successful business: ten years after startup, Coca Cola bought a 40% share of the business for $43 million. That kind of real-world savvy alone sets him apart from most academic economists. You should read his books--I'm not saying that I agree with every single one of his offbeat ideas, but there's a lot of food for thought in them.

Paying audited taxpayers is clearly a controversial idea, and I'm not saying I'm ready to jump on the bandwagon, but I do think it's an idea that merits serious consideration. Your arguments, based on your extensive experience dealing with IRS auditors, are certainly worth taking into consideration, as well.

I don't know if you are old enough to have been practicing back when the IRS had its old Taxpayer Compliance Monitoring Program (TCMP) audits, but they seemed like thoroughly unpleasant affairs. As I understand it, they constituted a very major imposition on the time and energies of a hapless group of taxpayers selected entirely at random, in a stratified sample selected for research purposes.

In other words, the TCMP audits were exhaustive audits conducted on a random group of taxpayers about whom the IRS had absolutely no reason to suspect were any more guilty of non-compliance than the next guy. Why? Because the IRS believed it needed baseline data for research purposes in order to construct statistical models that would allow it to target all the rest of its audits more effectively.

That meant auditing a random cross-section of taxpayers, including some who were fully compliant, indeed, those TCMP audits even turned up a significant number of taxpayers who had overpaid their taxes. The IRS believes its duty is to enforce the tax law and to collect the correct amount of taxes from taxpayers, but not more than the government's due, so such taxpayers received refunds.

So what was a TCMP audit like? As I understand it from IRS researchers with whom I spoke back in the 1980s when they were still conducting such audits, if your return was selected for a TCMP audit, you were required to substantiate every line on your tax return, producing your marriage certificate if you were filing jointly, birth certificates for your children if you claimed them, evidence that you met the support tests and other tests to claim those dependents, and so on, line by line through every item of your return.

TCMP audits were understandably very unpopular with most taxpayers, although I understand that the IRS Commissioner kept a drawerful of letters from a minority of grateful taxpayers who found out from their TCMP audits that the government actually owed them money. They were the minority, of course, and a political clamor arose to discontinue those research audits, even though the IRS insisted that it needed that information to construct and updated accurate models for selecting the lion's share of its audits through the so-called Discriminant Function (DIF) model.

Congress listened to the aggrieved taxpayers and ordered the IRS to abandon the TCMP audits, and it's been almost two decades since there have been any TCMP audits. Without those TCMP audits, some statisticians believe the current DIF model used by the IRS is seriously out of date.

I'm not privy to the details of the DIF, nor am I privy to the classified data which causes those statisticians to claim that the IRS needs new research audit data to update its statistical models.

But, for the sake of argument, let's say those statisticians are correct, that the IRS does need to collect some audit data from a stratified random cross-section of taxpayers about whom the IRS has no particular reason to suspect non-compliance.

It doesn't seem entirely unreasonable to me that such taxpayers ought to be compensated for their time and trouble, if they are found to be compliant. In many jurisdictions, taxpayers are paid if they are selected for jury duty, albeit not very generously. In some states, jurors receive no pay for short trials (less than three days), but they do receive pay for serving on a long and complicated case. When I was growing up, young men drafted into the armed services were still paid for their work, again, not all that generously.

Random selection for a research audit is a burden more onerous than most jury duty assignments, and less burdensome than most military duty. There is a case to be made that a taxpayer whose return was randomly selected for a research audit has been "conscripted" into an important civic duty, and deserves at least some compensation from the government for his time.

There are ways to structure the incentives for the auditors so that those who conduct the research audits are rewarded for audits that are as accurate as possible, rather than audits that significantly favor the government. The old TCMP auditors were told not to worry about the dollar totals their audits collected, but rather to worry about making the audits as accurate as possible so the IRS would have the best possible data on which to build its DIF model.

By similar reasoning, accounting for the funds the government uses to compensate taxpayers could be done in a way that does not create perverse short-term incentives to maximize direct revenue collection from these exhaustive research audits. Such funds could come from a separate budget outside the direct control of the IRS. To serve their intended purpose, research audits need to focus on collecting the most accurate possible data, not the most revenue from the particular taxpayer who is audited. In other words, you don't want the taxpayer just throwing up his hands and saying, "Just get out of my hair. I'll pay whatever you tell me to pay." For a genuinely useful research audit, you want the taxpayer providing all the information needed to assess his tax bill accurately.

One could also, for example, ask the National Taxpayer Advocate's office to be involved in "auditing the research auditors" to make sure that they are conducting those very burdensome audits in an even-handed manner.

That said, I'm not necessarily willing to jump on the Nalebuff-Ayres bandwagon.

I still have some reservations about the idea of paying audited taxpayers. Paying people for complying with civic duties can sometimes be a two-edged sword. It might encourage so much cynicism and negativity that it could undermine what voluntary compliance remains among American taxpayers.

But I don't think you have to be "smoking something" to propose the idea for public consideration.


  1. Are you suggesting that I've been wrong before?

  2. What an insightful post! I appreciate and enjoy reading the background you provide. I would so enjoy being in one of your classes!