Tax Compliance of SMEs
How to cite
Schmitz, Jan; Doerrenberg, Philipp. 2017. "Tax Compliance of SMEs." AEA RCT Registry. AEARCTR-0002390. https://doi.org/10.1257/rct.2390-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
We present the design and pre-analysis plan for a randomized control trial evaluating the effects of randomized audit threats and moral appeals on tax compliance of small and medium sized firms in Bulgaria. True audit probabilities are randomly varied between treatments and communicated to the firms. In addition, we employ treatments in which firms receive mailings highlighting the importance of taxes and specifying the purpose of tax money. Similar to the treatments with varying audit probabilities, we also manipulate the strength of the moral appeals to taxpayers. We benchmark our findings in the different treatments against a control group which receives a neutral mail from the tax authorities.
Intervention
The study will be conducted between July and December 2017 (or January/February 2018--depending on the capabilities of the NRA) as a randomized control trial (RCT) in collaboration with the Bulgarian tax authorities. 172.172 SMEs and self-employed taxpayers have been randomly divided into an untreated control group and three main treatment groups. In the first treatment group - the treated control group -, taxpayers receive a neutral mail informing about the website of the Bulgarian tax authorities which contains valuable instruction on how to file taxes correctly. In the second treatment group - the audit group -, the probability with which a taxpayer will be subject to an audit in the upcoming month is communicated to the taxpayer by mail. In the third treatment group - the moral appeals group -, taxpayers receive a mail informing them about the moral obligation to pay taxes as well as what taxes are used for. In all treatments, taxpayers receive the same information as in the treated control group but the content of the mailings is extended to either include information about the true audit probability or information about the moral obligation to pay taxes and what taxes are used for (content invoking reciprocity). The two main treatment groups - audit group and moral appeals group - consist of several subgroups in which the audit probability and the strength of the morale appeal are varied; see below for detailed descriptions of all treatment groups.
Because tax evasion is a major problem -- not only in Bulgaria -- , the tax authorities and we are interested in identifying the optimal and most cost effective audit probability (from the whole possible set between 0\% and 100\%). Specifically, in six sub-treatment groups we introduce individual firm specific audit probabilities of 0\%, 1\%, 10\%, 40\%, 60\%, 100\% and inform taxpayers about their individual probability to be audited. We further introduce one treatment with an ambiguous positive audit probability (of 1\%) where taxpayers are not informed about the exact probability and only that the probability is positive. Additionally, since audits are costly, the tax authorities are also interested in how effective moral appeals to pay taxes work in comparison to an officially announced probability of receiving an audit. Therefore, we want to use tools from behavioral economics that have proven to effectively increase cooperation such as, e.g., inducing an increased feeling of reciprocity through moral messages. Precisely, we have four sub-treatments using moral appeals. In one treatment, we invoke weak reciprocity. In another treatment, we invoke strong reciprocity indicating what taxes are used for. In the third moral appeal treatment we directly address the taxpayer and mention that the taxpayer benefits from services and infrastructure paid for by tax money. In the fourth treatment using moral appeals the mail is accompanied by a graphical example (i.e., a picture of a playground for children) of tax spending. We then aim at testing how well these moral messages work in increasing tax compliance compared to an announced audit.
All interventions have been prepared in collaboration with the tax authorities. The interventions take into account the capabilities and interests of the NRA. The NRA, however, will communicate and interact with the taxpayers. The researchers will not have any contact with the taxpayers. All mailings that taxpayers receive will be sent by the tax authorities.
Taxpayers in the untreated control group will not be subject to any intervention. However, some taxpayers in the this group will be invited to voluntarily fill in a questionnaire and state their perceived probability of receiving an audit and how immoral they think tax evasion is. This survey will - depending on the response rate - inform us about the underlying belief about the probability of receiving an audit and the underlying immorality of cheating on tax returns in the population of SMEs in Bulgaria. However, the survey is mainly independent from the experiment presented here. Therefore, we will not specify the details of the survey in more detail here. Firms in the survey, however, will be excluded from the sample as they maybe behave differently because of filling in a tax (and honesty) related questionnaire.
Experimental design
The study will be conducted between July and December 2017 (or January/February 2018--depending on the capabilities of the NRA) as a randomized control trial (RCT) in collaboration with the Bulgarian tax authorities. 172.172 SMEs and self-employed taxpayers have been randomly divided into an untreated control group and three main treatment groups. In the first treatment group - the treated control group -, taxpayers receive a neutral mail informing about the website of the Bulgarian tax authorities which contains valuable instruction on how to file taxes correctly. In the second treatment group - the audit group -, the probability with which a taxpayer will be subject to an audit in the upcoming month is communicated to the taxpayer by mail. In the third treatment group - the moral appeals group -, taxpayers receive a mail informing them about the moral obligation to pay taxes as well as what taxes are used for. In all treatments, taxpayers receive the same information as in the treated control group but the content of the mailings is extended to either include information about the true audit probability or information about the moral obligation to pay taxes and what taxes are used for (content invoking reciprocity). The two main treatment groups - audit group and moral appeals group - consist of several subgroups in which the audit probability and the strength of the morale appeal are varied; see below for detailed descriptions of all treatment groups.
Because tax evasion is a major problem -- not only in Bulgaria -- , the tax authorities and we are interested in identifying the optimal and most cost effective audit probability (from the whole possible set between 0\% and 100\%). Specifically, in six sub-treatment groups we introduce individual firm specific audit probabilities of 0\%, 1\%, 10\%, 40\%, 60\%, 100\% and inform taxpayers about their individual probability to be audited. We further introduce one treatment with an ambiguous positive audit probability (of 1\%) where taxpayers are not informed about the exact probability and only that the probability is positive. Additionally, since audits are costly, the tax authorities are also interested in how effective moral appeals to pay taxes work in comparison to an officially announced probability of receiving an audit. Therefore, we want to use tools from behavioral economics that have proven to effectively increase cooperation such as, e.g., inducing an increased feeling of reciprocity through moral messages. Precisely, we have four sub-treatments using moral appeals. In one treatment, we invoke weak reciprocity. In another treatment, we invoke strong reciprocity indicating what taxes are used for. In the third moral appeal treatment we directly address the taxpayer and mention that the taxpayer benefits from services and infrastructure paid for by tax money. In the fourth treatment using moral appeals the mail is accompanied by a graphical example (i.e., a picture of a playground for children) of tax spending. We then aim at testing how well these moral messages work in increasing tax compliance compared to an announced audit.
All interventions have been prepared in collaboration with the tax authorities. The interventions take into account the capabilities and interests of the NRA. The NRA, however, will communicate and interact with the taxpayers. The researchers will not have any contact with the taxpayers. All mailings that taxpayers receive will be sent by the tax authorities.
Taxpayers in the untreated control group will not be subject to any intervention. However, some taxpayers in the this group will be invited to voluntarily fill in a questionnaire and state their perceived probability of receiving an audit and how immoral they think tax evasion is. This survey will - depending on the response rate - inform us about the underlying belief about the probability of receiving an audit and the underlying immorality of cheating on tax returns in the population of SMEs in Bulgaria. However, the survey is mainly independent from the experiment presented here. Therefore, we will not specify the details of the survey in more detail here. Firms in the survey, however, will be excluded from the sample as they maybe behave differently because of filling in a tax (and honesty) related questionnaire.
Design details
Cited by
No citation data available for this registration yet.