REWARDING HONEST TAXPAYERS: AN EXPERIMENTAL ASSESSMENT.

• Author: Satterthwaite, Emily A.

INTRODUCTION I. RELATED LITERATURE A. Rewards Generally B. Tax Contexts 1. Lotteries a. Laboratory Experiments b. Field Experiments 2. Monetary Rewards II. DESCRIPTION OF EXPERIMENT A. Experimental Design 1. Eligibility 2. Set-Up a. Audit Rule Conditions b. Reward Rule Conditions c. Other Parameters: Instruction Screens B. Rational Expected Utility Model Predictions III. RESULTS A. Summary of Data B. Effect of Audits and Rewards on Average Compliance C. Effect of Audits and Rewards on Perfect Post-Audit Compliance D. Effect of Audits and Rewards on Consistent Compliers IV. CONCLUSION APPENDIX INTRODUCTION

In recent years, U.S. federal tax administration and enforcement resources have been drastically curtailed. (1) At the same time, complex and hastily adopted new tax legislation, (2) coupled with rapidly changing income-earning patterns of the gig economy, (3) have put the enforcement capacities of the Internal Revenue Service ("IRS") under increasing strain. (4) To respond to these developments and to deter evasion among taxpayers who may be emboldened by news reports of an agency stretched thin, interest in developing innovative policy interventions that can bolster voluntary tax compliance at low cost has grown. For example, a series of U.S-based field experiments that have focused on the design details and efficacy of audit threat letters, which are inexpensive to deploy and simple to adjust in response to research findings, address this demand for cost-effective new enforcement tools. (5)

This laboratory-style study was conceived of and designed in a similar spirit. It draws conceptual inspiration from proposals by academics to compensate taxpayers for the cost of being subjected to random audits (6) and empirical inspiration from recent policy initiatives outside the United States, where taxing authorities may face similar, or more severe, constraints. Hilke Brockmann and co-authors stated in 2016: "Pressed by high spending requirements and high political obstacles to tax increases, some governments have recently experimented with recovering some of these losses through a new rewards approach to tax compliance. The idea is to curb tax evasion by providing positive rewards for individual tax compliance." (7)

Prizes, lotteries, tax holidays, and other incentives have been the subject of field experiments conducted by governments in a number of countries, including Argentina, (8) Germany, (9) and Uruguay. (10) In the context of value-added taxes ("VATs"), evidence is emerging that receipt-based lotteries (11) have been effective in incentivizing consumers to ask for VAT receipts and sellers to correctly report their sales for VAT purposes. (12)

However, research on rewarding taxpayers for voluntary compliance is still in its early stages. (13) As Leandra Lederman states in a recent paper that reviews empirical studies of tax enforcement and voluntary compliance: "There are few studies of what effects rewards would have in the context of tax compliance. ..." (14) And, despite suggestions that experimenting with reward programs may be on the IRS's radar, no U.S.-based field experiments involving rewards appear to have been announced to date. (15)

This Article uses low-cost experimental methods to contribute to the literature on rewarding taxpayers for voluntarily complying with their tax reporting obligations. Specifically, it evaluates the effectiveness of providing honest taxpayers with nominal rewards following an audit in an online experimental setting. Might such reward programs represent an underappreciated enforcement tool and, as such, merit field experimentation?

From a theoretical perspective, two related but separate debates in the vast academic literature on tax compliance animate this objective. (16) These debates feature the same two contestants in rotating roles: the carrot and the stick. (17) In the particular context of voluntary tax compliance, the carrot represents a subsidy: a reward, prize, bonus, chance to win a lottery, or other positive inducement for honest reporting of income. The stick represents a sanction. The most common forms of sanctions are audits and penalties, where penalties are assessed as a percentage of unreported income detected upon audit.

In the first debate, the carrot and the stick are potential substitutes that may have differing capacities to promote tax compliance. (18) Although there is a body of research in psychology that has emphasized the efficacy of carrots in inducing desired behavioral outcomes, (19) the standard expected utility model of income tax compliance and most empirical research on its predictions features sticks. (20) The model conceptualizes individual income tax compliance as a choice between a risk-free (reported, after-tax income) asset and a risky (unreported, untaxed income) asset. (21) An increase in the probability of audit, or in the penalty for detected evasion, causes taxpayers to shift towards the risk-free asset of reported income. (22) This means that--all else equal--more sticks (in the case of audits), or bigger sticks (in the case of penalties), should increase tax compliance. (23) According to this conceptual approach, carrots in the form of rewards for compliant taxpayers can have the same effect: they change the price of compliance relative to evasion. (24)

In the second debate, the carrot and the stick are lumped together as a single potential villain. At issue is the efficacy of using prices in the form of carrots or sticks, as opposed to relying on the "intrinsic motivations" of individuals to pay their taxes honestly. (25) In this debate, the use of either a carrot or a stick risks crowding out taxpayers' commitments to voluntary tax compliance. (26) The crowd-out problem rears its head when a sanction or subsidy chills individuals' willingness to voluntarily comply, thereby unintentionally increasing rather than decreasing tax evasion. (27) Crowding-out might occur in a variety of circumstances, including when a specific individual decreases her compliance after directly experiencing an audit. (28) Or it might occur in a more generalized way, when the use of sanctions or subsidies results in decreased average compliance rates within a particular community or across a broader population. (29)

Evidence of the former circumstance--individual-level crowding-out--has been observed in laboratory settings (30) and has been investigated in several field experiments. (31) Some of these studies have documented a "bomb crater effect" following audits, which is a particular form of crowding-out that is specific to the taxpayer who has been audited. (32) It refers to a taxpayer's reduced compliance (relative to her pre-audit compliance level) after experiencing an audit. (33) As explained by Francesco Guala and Luigi Mittone, the name evokes the following scenario:

They say that troops under heavy enemy fire hide in the craters of recent explosions, for they believe it is highly unlikely that two bombs will fall exactly in the same spot within a short time period. Something similar seems to happen in the tax experiments: immediately after each audit, tax payments fall sharply (i.e. evasion increases). (34) A recent experimental study found a bomb crater effect that was sensitive to the audit rule used by the tax authority: it appeared in response to random audits, but it did not appear in response to audits that were conditioned on prior detected evasion ("endogenous audits") and announced as such to taxpayers. (35) The bomb crater studies raise the question of whether measures used in combination with random audits might ameliorate their unintended post-audit consequences. Given the well-documented centrality of random audits to effective tax enforcement, (36) such measures could represent important enhancements to self-assessment tax systems. (37)

Taxpayer reward programs are natural candidates. The most recent and comprehensive empirical study (to the best of this author's knowledge) of the effectiveness of taxpayer reward programs found that their success is highly sensitive to their design. (38) As a result, the treatment experimental conditions explored in this Article were structured to test the effectiveness of the simplest and most easily-executed form of taxpayer compliance rewards: token monetary transfers in consideration for full compliance. (39)

In particular, this Article experimentally investigates the interaction between audits, rewards, and voluntary compliance over time in a U.S. taxpaying context. (40) It reports the results of a multi-period online tax reporting survey administered in May 2017 to approximately 400 U.S. participants on the Amazon Mechanical Turk ("MTurk") task completion platform. (41) The survey evaluated the performance of rewards in combination with both random and endogenous audits. (42)

Analysis of the data showed that the reward treatment was ineffective in improving average post-audit compliance rates under either audit rule. Rewards also failed to increase the likelihood that audits--of either type--were followed by perfect post-audit compliance in the period immediately following the audit. However, the reward treatment's performance stood out with regard to a specific metric relating to the durability of taxpayers' voluntary compliance over time: when used in combination with random audits, the reward increased the likelihood that participants in the experiment would fully report all of their income in each of the 60 periods of the experiment ("consistent compliance") by 89% as compared to the no-reward condition (statistically significant at the 5% level).

Average compliance rates exhibit deterioration over time in most experimental dynamic audit studies; they slope downward over time. (43) Therefore, one interpretation of the result regarding consistent compliance is that using rewards in combination with random audits may prevent the progressive erosion of taxpayers'...