The IRS’s audit enforcement policy is based on the principle that audits deter taxpayers from cheating and therefore increase voluntary tax compliance. For more than 40 years, however, there has been considerable debate over whether higher audit rates actually contribute to increased voluntary tax compliance.
Many scholars today believe that this “deterrence theory” is incorrect since the overall data does not comport with the theory.
Taxpayers, therefore, must voluntarily comply with their tax obligations for reasons other than an increased probability of getting audited. While this conclusion might be true for individual income tax returns overall, it appears there are certain sub-categories of tax returns for which audits actually do matter. For these groups, audits might affect voluntary tax compliance levels more directly.
The IRS publishes two core strategic goals:
(1) Improve service to make voluntary compliance easier; and
(2) Enforce the law to ensure everyone meets their obligations to pay taxes.1 The IRS taxpayer service program supports the first strategic goal, and is committed to helping taxpayers understand and meet their tax obligations.2
The IRS enforcement program supports the second objective and is based on a long-standing conviction that audits serve as deterrents for taxpayers otherwise inclined to evade their tax responsibilities.3
Therefore, both goals have a common aim: to increase voluntary tax compliance.
From these two strategic goals, one can conclude that a strong audit program is necessary for voluntary compliance for those groups of taxpayers who underreport tax intentionally.
For taxpayers who underreport unintentionally, increased audits might have little effect on future voluntary compliance. Instead, a strong taxpayer service and education program is perhaps a better way to improve voluntary compliance for taxpayers in this group since their underreporting is most likely a result of mistake or ignorance of complex tax rules.
If one could calculate the underreporting rates for the various groups of individual income tax returns, one could then calculate the correlation between the published audit rates and the underreporting rates for each group of returns. Positive correlations – i.e., as audits go up so does underreporting – suggest underreporting is most likely unintentional. Negative correlations – i.e., as audits go up underreporting goes down – suggest underreporting is most likely intentional.
These “intentional” groups perhaps have a hypersensitivity to audit enforcement levels, whereas for the “unintentional” groups, audit rates do not appear to matter to underreporting. The problem is, with no understanding of how many returns contain underreported tax, and therefore no way to calculate the respective underreporting rates, it becomes impossible to determine for which groups audits affect voluntary compliance.
A recent theory, developed elsewhere, allows for the approximation of underreporting rates for both the overall individual income tax return population and its disaggregated groups.4
This theory purports to approximate the underreporting rate for any individual income tax return group for which there exist published enforcement data. The results allow for a correlation analysis between published audit and predicted underreporting rates.
The IRS’s Statistics of Income report both aggregate and disaggregate data for individual income tax returns. For taxable years 2005 through 2011, the disaggregate data is similar enough to use for a year-by-year comparison. The audit rate is the proportion of returns selected for audit from all of the returns filed. Of those returns selected for audit, the proportion containing underreported tax is called the detection rate. The predicted underreporting rate, not reported by the IRS, is the proportion of returns containing underreported tax from all of the returns filed.
Table 1 lists the categories of individual income tax returns for taxable years 2005 to 2011 along with the mean audit (α) and detection (δ) rates. The mean predicted underreporting rate (υ) is calculated from the theory previously mentioned.
Although audit and detection rate data are published for total individual income tax returns for other taxable years, the disaggregated categories of returns are reported differently prior to taxable year 2005, thereby making comparisons problematic.
Each taxable year in Table 1 also includes the correlation coefficient (r) between the reported audit and predicted underreporting rates for the 2005 to 2011 period. The correlation coefficient is a normalized measure of how the audit and underreporting rates are linearly related. Correlation coefficients close to unity (1) indicate a strong positive linear relationship. Correlation coefficients close to negative unity (-1) indicate a strong negative linear relationship, whereas correlation coefficients close to zero (0) indicate a weak linear relationship between audits and underreporting. Negative correlations are displayed in parentheses. See table 1
|Aggregate and disaggregate data for individual income tax returns
from TY 2005 through TY 2011
|Individual Income Tax Returns||α||δ||υ||r|
|All returns (aggregate roll up)||0.0103||0.8269||0.3507||0.71|
|Returns with total income < $200,000|
|Nonbusiness returns without EITC:
without Schedules C, E, F, or Form 2106
with Schedule E or Form 2106
|Business returns without EITC:
Nonfarm business returns
|$25,000 < $100,000||0.0228||0.8562||0.3376||0.24|
|$100,000 < $200,000||0.0425||0.8624||0.3412||(0.12)|
| Business and nonbusiness returns with EITC:
|Returns with total income from
$200,000 to $1,000,000
|Returns with total income ≥ $1,000,000||0.0895||0.5763||0.4392||(0.70)|
Before discussing the data, a walk-through of a category in table 1 might be helpful; for example, “Returns with total income ≥ $1,000,000.” Of all the individual returns filed with $1 million or more of income, 8.95 percent were selected for audit (α).
Of the 8.95 percent selected for audit, the IRS determined there to be underreported tax for 57.63 percent of them (δ). The theory discussed previously predicts that about 43.92 percent of returns with $1 million or more of income contain some form of underreported tax (υ). Comparing the reported audit rate and predicted underreporting rate produces a negative correlation coefficient (r) of -0.70.
The overall population of individual income tax returns shows a positive correlation between audit and underreporting rates for taxable years 2005 through 2011. This suggests that audits might make little difference to voluntary compliance levels for the overall individual income tax return population. The same is true for other groups with a positive audit/underreporting correlation.
Since underreporting still occurs for these groups, one conclusion is that most of this underreporting is not intentional, but rather due to some other factor such as mistake, ignorance of a complex tax code and administrative regulations, or even paid return preparer error. For this “unintentional” group of returns, increased enforcement seems to be a waste of government resources. If the underreporting is unintentional to begin with, increased enforcement holds little promise of effecting behavior modifications of the taxpayers within the system.
What might be a better investment, however, is an increase in taxpayer education and awareness of taxpayer obligations under current law, which is in line with the IRS’s first strategic goal: to improve service to make voluntary compliance easier.
This might also support the need for tax simplification overall, especially if future research shows that most of the underreporting on returns is due to ignorance or mistake attributable to the complexity of the tax laws and administrative regulations.
Some of the disaggregated categories, however, show a negative correlation, such as the “Returns with total income ≥ $1,000,000” category discussed earlier. A negative correlation suggests that these categories of returns are perhaps hypersensitive to audit rate changes, and that increased enforcement for these categories might have a more direct effect on voluntary tax compliance. For these groups, audits appear to influence voluntary compliance to some degree.
It is interesting to note in table 1 that some apparent anomalies exist that at first glance seem counterintuitive. For example, tax returns with farm activities have one of the highest predicted underreporting rates, yet they also have one of the highest positive correlations between audits and underreporting, suggesting any underreporting might be unintentional.
Also, tax returns with international and cross-border activities have the lowest predicted underreporting rate, yet they also have one of the highest negative correlations, suggesting any underreporting is likely intentional. Therefore, the degree to which audits affect compliance cannot necessarily be judged from only the levels of predicted underreporting.
It is possible that a category with one of the highest predicted underreporting rates can be least affected by audit levels, and a category with one of the lowest predicted underreporting rates can be most affected by audits.
It is also interesting to note that the categories of returns with the highest negative correlation – the “intentional group” – seem to be those of taxpayers typically regarded as more sophisticated. These include taxpayers who file returns with:
(1) business income of at least $200,000;
(2) international activities or cross-border transactions; and
(3) personal income of at least $1,000,000. On the flip side, the returns with the highest positive correlation, the “unintentional” group, seem to be those of taxpayers typically considered more ingenuous.
These include taxpayers who file returns with:
(1) farm activities;
(2) small businesses with less than $25,000 of income that do not claim the earned income tax credit (EITC); and individual returns that have no business activities and do not claim the EITC. From this, one can make the (over)generalization that to improve voluntary tax compliance, enforcement is perhaps more effective for “sophisticated” taxpayers and education is perhaps more effective for “ingenuous” taxpayers.
Do audits affect voluntary tax compliance? The short answer is that audits appear to affect voluntary compliance for certain categories of tax return filers but not for others. The underreporting estimates suggest that most of the individual income tax underreporting is unaffected by audit rates, and is therefore most likely unintentional. Instead, underreporting for these categories is most likely the effect of mistake or ignorance of a complex tax code and administrative regulations.
Yet if the individual income tax statistics are disaggregated, one finds evidence that suggests tax underreporting at higher income levels on both business and personal returns is most likely intentional. These results suggest that increased enforcement will most likely be fruitful for the “intentional” groups since there appears to be a generally negative correlation between the audit and underreporting rates.
For lower income and non-business categories, however, increased enforcement might not affect the underreporting rates, since the positive correlation between audit and underreporting rates suggests underreporting is most likely unintentional.
For these “unintentional” categories, public investment in taxpayer service and compliance education could promote voluntary compliance more than increased enforcement.
This analysis can hopefully help tax policymakers decide how best to spend limited public resources, especially in today’s austere economic and contentious political environment.
- Office of Management and Budget of the United States Government, Appendix, Fiscal Year 2014 1010.
- See, J. T. Manhire, “Do Audits Matter?: A Parallax Theory of the Relation between Tax Enforcement and Underreporting” http://ssrn.com/abstract=2526417.