Perceptions of Corruption and Tax Compliance: A Survey Experiment in Honduras

We will employ various messaging interventions to assess their impact on taxpayer perceptions and willingness to pay or evade taxes within the context of the Honduran tax administration (SAR, as per its Spanish acronym). We have four treatment arms:
• The control group, which received basic information about SAR's role in tax collection, together with an image of the locations of the SAR offices in Honduras.
• The "perception" treatment group, which is going to receive basic information about SAR's role in tax collection and information about SAR's reputation as one of the most honest tax authorities in Latin America, backed by a graph using survey data indicating it as the second most honest tax authority institution in the Latin American region.
• The "purge" treatment group, which is going to receive basic information about SAR's role in tax collection and information about SAR's recognition of corruption as a significant issue and its recent efforts to combat it, including a graph highlighting the personnel renewal done by the institution.
• The "combined" treatment group, which is going to receive the information from the "perception" treatment first, followed by the "purge" treatment information.

2023-10-02

2023-10-09

SAR Transparency, Probability SAR Agent Bribed, SAR's Capability, Willingness to Bribe SAR Agent, Discount Offer Rejected, No Lie Dice Game.

We divide our outcomes in two different groups. The first one is referred to the beliefs and perceptions group, which includes:
1. SAR Transparency: This metric comprises a self-reported outcome in which respondents rate, on a scale from 0 (completely lacking transparency) to 10 (very transparent), their perception of the transparency of SAR officials. We are also going to include as an outcome the difference between this variable and the perceptions of transparency of the public administration in general (explained in the Experiment Design section).
2. Probability SAR Agent Bribed: This metric comprises respondents' beliefs regarding the probability that an SAR official might request gifts or money in exchange for resolving or preventing tax-related issues for taxpayers (e.g., allowing tax evasion, avoiding penalties, or expediting processes). Respondents rate this likelihood on a scale from 0 (not at all likely) to 100 (highly likely). For analysis purposes, we will invert the index so that a higher score indicates a lower probability of being bribed.
3. SAR's Capability: This metric comprises a self-reported outcome where respondents are asked to indicate their perception of the SAR's capability in collecting the amounts owed by individuals with tax obligations. Possible responses include "Not capable at all", "Slightly capable", "Moderately capable", "Capable" and "Highly capable". This outcome is incorporated for capturing mechanisms. We want to understand if, thanks to the treatments, respondents are more willing to pay taxes because now they trust the institution responsible for collecting them.
Considering that the first two outcomes have the same objective, which is to analyze the impact on citizens' perceptions of the tax authority, we are going to combine the first two in different indexes that allow us to reflect the change in the perceptions of individuals being treated.
The second group of outcomes, on the other side, is the group tax morale and proxies of willingness to pay or evade taxes outcomes, which includes:
1. Willingness to Bribe SAR Agent: This outcome evaluates respondents' agreement with the statement: "I would be willing to bribe an SAR official to avoid paying the taxes I owe." Respondents rate their level of agreement on a scale from 0 (completely disagree) to 10 (completely agree). For analysis purposes, we will invert the index so that a higher score reflects a lower willingness to engage in bribery.
2. Discount Offer Rejected: This measure presents respondents with a scenario in which a store owner offers a 10% discount in exchange for not providing a receipt, invoice, or proof of purchase during a shopping transaction. Respondents indicate whether they would accept or reject this discount (accepting the discount implies being willing to evade taxes). The outcome will be equal to 1 if the respondent says that they will not accept the offer.
3. No Lie Dice Game: This binary outcome originates from an implemented Dice Game (see below in the Experimental Design Section) and indicates whether an individual provided truthful information in their self-report regarding the outcome of the dice roll.
Considering that all these outcomes have the same objective, which is to analyze the impact on citizens' willingness to pay taxes and on tax morale outcomes, we are going to be combining them in different indexes that allow us to reflect the change in the willingness to pay or in the tax morale of the individuals.

The primary goal is to identify effective communication strategies to enhance tax morale within the context of the tax administration in Honduras. Data will be collected through an online survey. Participants will be selected randomly from the general population of Honduras and will be assigned to different communication treatments (explained with more detail in the Intervention section):
1. Control: A neutral message providing information about the role of the Honduran tax authorities.
2. "Perception" Treatment: A message emphasizing the role of the Honduran tax authorities and highlighting the low corruption perceptions of the Honduras tax authorities (based on factual data from a previous regional survey).
3. "Purge" Treatment: A message emphasizing the role of the Honduran tax authorities and highlighting the government's efforts to combat existing corruption (based on a real purge conducted in recent years).
4. "Purge" + "Perception" treatment ("combined" treatment): A combined message in which respondents are going to receive the information from the "perception" treatment first, followed by the "purge" treatment information.
The outcomes of interest include, first, a set of outcomes regarding the perceptions of the tax authority; and second, a set of self-reported outcomes regarding tax morale and proxies of willingness to pay or evade taxes (presented in the Primary Outcomes Section). We will also measure actual tax payment behavior through a Dice Game in which participants have the opportunity to win a prize. This game involves rolling a virtual die which produces a random number. Participants then report this number, and in return, they receive a certain number of tickets based on the reported value. The more tickets they accumulate, the better their chances of winning. Participants also could lie, but we will tell them that if someone is found to have reported the wrong number, they will receive only one ticket for the extra prize draw, regardless of the number they originally reported (although at the end of the day will not penalize those who did not tell the truth). Lying improves chances of winning of the respondents but harms the other participants (we consider this a proxy to the willingness of paying or evading taxes). The dice that we are going to be rolling will be unbalanced, with the idea that in most cases the number shown is one (1).
The hypotheses of this experiment are the following:
1. "Perception" vs. control Hypothesis: Changing individuals' beliefs about corruption by emphasizing lower corruption levels (Perception) will positively affect tax morale or willingness to pay and will positively affect perceptions of the Honduras tax authority compared to a neutral message (Control), assuming that individuals believe the information provided in the "perception" treatment.
2. "Purge" vs. control Hypothesis: Communicating government efforts to combat corruption (Purge) will positively affect tax morale or willingness to pay and will positively affect perceptions of the Honduras tax authority compared to a neutral message (Control). However, as a recent literature showed, it might backfire based on individuals' initial beliefs about corruption (since the purge treatment may reinforce their initial negative perception, Cheeseman and Peiffer, 2022). We thus consider this as a plausible outcome, which would affect also the subsequent hypotheses.
3. "Combined" vs. control Hypothesis: Combining both strategies (Combined), changing beliefs about corruption first and emphasizing government actions, will positively affect tax morale or willingness to pay and will positively affect perceptions of the Honduras tax authority compared to a neutral message (Control). The "combined" effect, however, can backfire if the "purge" treatment is making the problem of corruption more prominent (see Cheeseman and Peiffer, 2022).
4. "Perception" vs. "combined" Hypothesis: Changing perceptions about corruption first will make subsequent messages about government actions (Combined) more effective compared to just changing perceptions (Perception). The "combined" effect, however, can backfire if the "purge" treatment is making the problem of corruption more prominent (see Cheeseman and Peiffer, 2022). In that case, "perception" would be more effective than "combined".
5. "Combined" vs. "purge" Hypothesis: The combined message strategy (Combined) will be more effective than emphasizing government actions alone (Purge).
6. "Perception" vs. "purge" Hypothesis: Changing perceptions about corruption (Perception) will be more effective compared to emphasizing government actions (Purge), assuming that perceptions of transparency of the public administration play a significant role (Cheeseman and Peiffer, 2022) and that individuals believe the information provided in the "perception" treatment.
Given that we consider that there may be heterogeneous effects according to the initial beliefs of corruption of the individuals, we are going to be including pre-treatment transparency variables in order to verify if, for example, having high initial beliefs of corruption can trigger backfire in the "purge" treatment (and, consequently, in the "combined" treatment). One of the variables that we will add for the heterogeneity analysis is the transparency of the general government, which is a metric in which respondents rate, on a scale from 0 (completely lacking transparency) to 10 (very transparent), their perception of the transparency of the public administration in general. This variable will also serve as a control when analyzing the effect of the treatments on our main outcomes as well as to analyze the difference between pre- and post-treatment perceptions of transparency. This possibility arises based on the finding of Cheeseman and Peiffer (2022).
Also, we will ask how much respondents trust, on a scale from 0 (completely unreliable) to 10 (very trustful), in Church, in Congress, in the Executive Branch and in the police. In addition, combining these confidence indexes, or combining them with the transparency of the general government, we will be building different pre-treatment confidence measures that we will use in the analysis of our main outcomes.
Finally, we will include an attention check in our survey, and we will keep only those individuals who have passed it. Another series of baseline characteristics is also going to be collected, such as the age of the respondents, their sex, their ideology, the region in which they live or their education level.

The participants will first answer a baseline survey and then will be randomly assigned to a treatment arm. We are going to randomize at the individual level to our four different treatment arms. The survey software will be in charge of randomizing the individuals into the treatment arms.

Randomization is done at the individual level.

No

Our total sample size will be n=2,400.

Our total sample size will be n=2,400.

A quarter of the participants will be assigned to the "perception" treatment (n=600), a quarter of the participants to the "purge" treatment (n=600), a quarter of the participants to the "combined" treatment (n=600), and the last quarter of the participants to the control group (n=600).

We conducted a preliminary pilot study with a sample size of n=147, during which we observed significant effect sizes concerning our primary outcome, which pertains to the SAR Transparency. Specifically, the effect size approached 0.485 when comparing the "perception" group to the control group. This observed effect is substantially larger than our initial expectations, taking into account that we assume a level of statistical test equal to 0.05. Nevertheless, we plan to employ a larger sample size to determine effects across all our primary outcomes comprehensively. Additionally, this larger sample size will provide us with adequate statistical power to detect effects, considering the multitude of tests we intend to conduct.