Targeting in Tax Compliance Interventions: Experimental Evidence from Honduras

This experiment will send email communications to approximately 30,000 at-risk taxpayers in Honduras before the income tax filing deadline for FY2019.

The control group (49% of the sample) will receive a simple reminder about the deadline to file their taxes and the importance to file truthful information.

The treatment group (51% of the sample) will receive an e-mail containing the same information as the control group and, in addition, will be told about specific information the tax authority possess about their transactions during FY2019. The emails are personalized for each taxpayer and include either the sources of third-party information the authority possess on their revenue (sales to other taxpayers, debit/credit card sales, exports or sales to the government) or indicators about their operations flagged by the authority (repeated reported losses, financial transactions inconsistent with reported revenue or too low revenues compared to peers) for taxpayers with no third-party information available.

Among the treated sample, taxpayers were further randomized into receiving three slightly different messages in addition to the information available to the TA. One-third of the treatment group (17% of total study sample) will see a message reminding them that reporting incorrect information makes them liable for fines and sanctions according to the law. One-third will see a message reminding them about other administrative sanctions for incorrect filings, such as the denial of documents necessary for firms' operations. Finally a last group will receive a call to "moral duty", reminding them that taxes are used to finance public goods such as health and education.

2020-03-11

2020-03-20

1) Probability of filing Income taxes
2) Declared gross revenue
3) Declared deductions
4) Declared taxable income

1) Timeliness of FY 2019 declaration
2) Amount of income taxes paid
3) FY 2020 sales tax declarations
4) Rectification of previous years income and sales taxes.

The experimental sample consists of at-risk taxpayers, as identified by the risk-model used by the TA, with at least one valid email address. We implement a stratified randomization using 60 strata defined by whether third-party information was available or not; whether the taxpayer was a corporation; municipality of operations defined as Distrito Central (capital Tegucigalpa), San Pedro Sula (second largest city) or other; and five risk levels as defined by the tax authority (2*2*3*5 = 60 strata).

In each strata we allocated 49% of taxpayers to the control group and the remaining 51% in three equally sized treatment arms (fines, administrative sanctions and moral duty). Following Bruhn and McKenzie (2009) we deal with "misfits" (remaining taxpayers in each strata) by randomly assigning them to one of the four groups (control + 3 treatments) using the above weights as assignment probabilities.

Randomization done in office by a computer (Stata).

Individual taxpayer

No

This is not a cluster randomized trial, so the number of clusters is equal to the number of observations: 31,396 taxpayers.

This is not a cluster randomized trial, so the number of clusters is equal to the number of observations: 31,396 taxpayers.

There are 15,399 taxpayers in the control group and 15,997 taxpayers in the treatment group divided in three arms: threat of fines (5,306 taxpayers), threat of administrative sanctions (5,332) and call to moral duty (5,359).

Minimum detectable effects with 80% power, 95% confidence level and 60% compliance for treatment group (estimate of share of taxpayers who will open the emails, which we measure) for our primary outcomes are: 1) Probability of filing: 2 percentage points. 2) Declared gross revenue: L 71,000 (USD 2,800); 5.5% of the mean; 0.03 s.d. 3) Declared deductions: L77,000 (USD 3,000); 6.2% of the mean; 0.03 s.d. 4) Declared taxable income: L 8,800 (USD 350); 8.8% of the mean; 0.04 s.d.