NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Tax Audits as Scarecrows
Last registered on August 29, 2019


Trial Information
General Information
Tax Audits as Scarecrows
Initial registration date
August 28, 2019
Last updated
August 29, 2019 8:37 AM EDT
Primary Investigator
Other Primary Investigator(s)
Additional Trial Information
Start date
End date
Secondary IDs
The canonical model of Allingham and Sandmo (1972) predicts that firms evade taxes by optimally trading off between the costs and benefits of evasion. However, there is no direct evidence that firms react to audits in this way. We conducted a large-scale field experiment in collaboration with Uruguay’s tax authority to address this question. We sent letters to 20,440 small- and medium-sized firms that collectively paid more than 200 million dollars in taxes per year. Our letters provided exogenous yet nondeceptive signals about key inputs for their evasion decisions, such as audit probabilities and penalty rates. We measured the effect of these signals on their subsequent perceptions about the auditing process, based on survey data, as well as on the actual taxes paid, based on administrative data. We find that providing information about audits had a significant effect on tax compliance but in a manner that was inconsistent with Allingham and Sandmo (1972). Our findings are consistent with an alternative model, risk-as-feelings, in which messages about audits generate fear and induce probability neglect. According to this model, audits may deter tax evasion in the same way that scarecrows frighten off birds.
External Link(s)
Registration Citation
Perez-Truglia, Ricardo. 2019. "Tax Audits as Scarecrows." AEA RCT Registry. August 29. https://doi.org/10.1257/rct.4593-1.0.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
For our study, the IRS mailed four different types of letters with information about audits to the owners of each of these firms. Firms were randomized into four different letter types: baseline, audit-statistics, audit-endogeneity, and public-goods. The baseline letter type included brief and generic tax information that the IRS often includes in its communications with firms. The audit-statistics letter type was identical to the baseline letter, with additional information about the probability of being audited and the penalty rate, based on tax administration statistics. The audit-endogeneity letter provides information about a different feature of the auditing process. The audit-endogeneity letter was identical to the baseline letter, with an additional message about how evading taxes increases the probability of being audited. The last letter type was designed to provide a benchmark for a message that might increase tax compliance but does not involve the tax audits. The public-goods letter was identical to the baseline letter, with an additional message describing the social costs from evasion: all the public goods that could be provided if tax evasion was lower.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
VAT payments: concurrent and backward payments,
Other tax payments
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Our experiment consisted of a mailing campaign from Uruguay’s IRS with multiple treatment arms and sub-treatments. Rather than comparing firms that received a letter to firms that did not, all of our analyses are based on comparisons between firms that received letters with subtle variations in their content.

The letters consisted of a single sheet of paper with the name of the recipient in the header, the official letterhead of the IRS, and the scanned signature of the IRS General Director. These letters were folded, sealed in an envelope with the official letterhead of the IRS on the outside, and sent by certified mail.

The baseline letter contained some information about the goals and responsibilities of the tax authority, which the IRS routinely includes in its communications with firms.

In the audit-statistics letter type, we added a paragraph to the baseline letter, providing firms with information about the audit and penalty rates. Note that we communicated the probability that firms will be audited in at least one of the three following years, because IRS experts stated that this was the relevant probability for firms’ decision-making.

To complement the evidence from the audit-statistics sub-treatment, we implemented an alternative way of randomizing perceptions about audit probabilities using an audit-threat letter. We devised a treatment arm that randomly assigned firms to groups with different probabilities of being audited in the following year. We randomly assigned the firms in this treatment arm to two groups, one with a probability of being audited in the following year of 25% (X=25%) and another with a probability of being audited twice as large (X=50%).

We also devised a treatment arm to provide a benchmark for the effect of messages intended to increase tax compliance without directly mentioning audits. We designed a non-pecuniary message that was expected to be most effective at increasing compliance by the IRS staff and authorities: a message providing information about the cost of evasion in terms of the provision of public goods.

We designed a survey to be conducted with a sample of owners from our main subject pool, months after they received the letters. The IRS, with the support of the Inter-American Center of Tax Administrations and the United Nations, had previously administered a survey on the costs of tax compliance for small- and medium-sized businesses. We collaborated with the tax authority in the design and implementation of a new survey, which included a specific module tailored for our research design.

Our experiment was conducted in collaboration with the IRS of Uruguay. As of May 2015, there were 120,142 firms registered in the agency’s database. A subsample of 4,597 firms, preselected by the IRS, was put aside for the audit-threat sample, which we call the secondary experimental sample. Of the remaining firms, we followed a series of criteria to select our main experimental sample. We first excluded some firms by request of the IRS. For instance, we excluded firms subject to special regimes for VAT payments (very small or very large firms). We also kept in the experimental sample only those firms that made VAT payments in at least three different months in the previous 12-month period and those with a total value of at least USD 1,000.

We sent email invitations to participate in the online survey to 3,845 firms in May 2016, about nine months after our mailing experiment. Our purpose was to elicit the beliefs of firm owners. We did not include email addresses that were repeated more than three times in the full sample, as these most likely corresponded to accounting firms representing multiple small and medium enterprises. Even after applying this criteria, the IRS records could not ensure that the registered email address corresponded to the firms’ owners. We thus asked the survey respondent to self-identify as one of the following five types: owner, internal accountant, external accountant, manager, or other employee.

Experimental Design Details
Randomization Method
Randomization done in office by a computer
Randomization Unit
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
Main sample: 20,000 firms
Secondary sample: 5,000 firms
Sample size: planned number of observations
Main sample: 20,000 firms Secondary sample: 5,000 fims
Sample size (or number of clusters) by treatment arms
Main sample:

Baseline letter: 2500
Audit-statistics letter: 12500
Audit-endogeneity letter: 2500
Public-goods letter: 2500

Secondary sample:

2500 - 25% chances of audit
2500 - 50% chances of audit
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Minimum detectable effect: 5% Standard deviation: 140 Rejection rate (from pilot): 20% Power: 80% Statistical significance: 5%
IRB Name
IRB Approval Date
IRB Approval Number
Post Trial Information
Study Withdrawal
Is the intervention completed?
Intervention Completion Date
September 30, 2015, 12:00 AM +00:00
Is data collection complete?
Data Collection Completion Date
April 11, 2016, 12:00 AM +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
Main Sample: 16,392 firms
Secondary sample: 4,048 firms
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
Main Sample: 16,392 firms
Secondary sample: 4,048 firms
Final Sample Size (or Number of Clusters) by Treatment Arms
Main Sample (After dropping 18.5% of letters returned to the sender): 10,272 firms received audit-statistics; 2,064 firms received baseline; 2,039 firms received audit-endogeneity; 2,017 firms received publicgoods Secondary Sample (After dropping 12% of letters returned to the sender): 2,015 firms in the 25% probability group 2,033 firms in the 50% probability group (total N = 4,048).
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)