Citizen and Politician Surveys on Corruption in Italy, Brazil, Colombia and Pakistan

We elicit respondents' views about corruption and we explore the trade-offs they think politicians should and would make between efficiency and corruption and between equity and corruption. Using two randomized experiments, we study the respondents' willingness to waste money and/or engage in a trivial lie to obtain an ego-rent, and how likely legislators are to change their positions on corruption and pledge their support for anti-corruption policies after being informed of citizens' views on corruption.

We collect a representative citizen sample along the gender, income, and age dimensions in Italy, Brazil, Colombia, and Pakistan. The survey will also be sent to all first-level subnational politicians of these three countries and to all national legislators.

Experiment 1:
The intervention will consist of two treated and one control groups. In the lottery game intervention, participants will be randomized into these three groups. Every respondent will be automatically enrolled in a lottery to win €1,000 in Italy, 2,000 Brazilian Reals in Brazil, and 20,000 Pakistani rupees in Pakistan, to be donated to a charity of their choice (from a list of three in each country). Every respondent will answer if, should they win, they wish to receive a certificate stating that they donated that given amount to charity.
To the respondents in the control group, we state the winner will receive a certificate reporting the €1,000 donation. Respondents in the Treatment 1 will be informed that it will cost €60 to produce the certificate, which will report a donation of €940. Respondents in the Treatment 2 will be instead informed that it will cost €60 to produce the certificate, but it will (misleadingly) report a donation of €1,000.

T1 thus measures the tolerance of respondents for wasting part of the donation to produce a certificate; T2 measures in addition the propensity for lying. The Control group reports a baseline measure of the willingness to receive a certificate
Experiment 2:
The second intervention, only included in the legislators' survey, will consist of one treated and one control group. Participants will be randomized in these two groups.

To respondents in the treated group, we provide information on the aggregate levels of their country citizens’ tolerance for corruption relative to the individual legislator's own answer to the same set of questions. The control group does not receive this information

Subjects are then asked questions on the seriousness of corruption and if they are willing to make anti-corruption pledge, in private and in public. The experiment measures whether making legislators aware of a difference in views on corruption with their citizens will change their perception and willingness to pledge their support for anti-corruption policies.

2021-10-25

2024-05-31

We are interested in i) perceptions about corruption in the respondent's own country across the political spectrum; ii) trade-offs between efficiency and corruption; iii) trade-offs between equity and corruption; iv) willingness to waste money for ego-gratification; v) willingness to lie for ego-gratification; vi) whether respondents who are willing to lie are also more tolerant of corruption; vii) whether proving legislators with information on citizens' views on corruption changes their views on this topic and makes them more likely to pledge their support for anti-corruption policies.

All these outcomes, except the last one, will be collected both for citizens and politicians of the three countries surveyed.

Certificate treatment
We will randomize the information provided to the respondents about the cost of producing a certificate reporting their donation to a charity of their choice. The control group will be informed that the production cost of €60 will be covered by the researchers; therefore their donation will be €1,000 and this is what the certificate will state. Treatment group 1 will be informed that if they ask for the certificate, the production cost of €60 will be covered using part of the donation money. Therefore their donation will be €940 and this is what the certificate will state. Treatment group 2 will be informed that if they ask for the certificate, the production cost of €60 will be covered using part of the donation money. Therefore their donation will be €940 but the certificate will state that they donated €1,000.

---

We also randomize the questions' order. The order randomization will divide respondents in two groups. Respondents assigned to Order 1 will first answer to the vignette questions eliciting trade-offs between equity/efficiency and corruption, then to the questions on redistribution policy preferences, then to the questions on corruption, and finally to the questions on trust. Respondents assigned to Order 2 will start with the question on trust, then the vignettes, then the questions on redistribution policy preferences, and finally the questions on corruption.

---

Information treatment
We will randomize whether legislators receive information about their citizens’ views on corruption. The control group will receive no information about the citizens’ views. The treatment group will be provided with information on the aggregate levels of citizens’ tolerance for corruption relative to legislator's own answers to the same set of questions (the five vignettes that can be read in the attached questionnaire).
We further randomize the kinds of public announcement that we ask legislators to make on social media. Half of the respondents will be asked to publicly announce their support for legislation that strengthens financial disclosure by persons running for elected office. The other half will be asked to disseminate information about the most important piece of legislation they have supported recently.

Randomization done by a computer

Individual

No

0

1,000 citizens in Italy contacted online 1,000 citizens in Brazil contacted online 1,000 citizens in Colombia contacted online 500 citizens in Pakistan contacted online 1,000 citizens in Pakistan contacted in person The survey will be sent to 917 regional politicians in Italy, 1001 state-level politicians in Brazil, and 676 provincial-level politicians in Pakistan. The final number of politicians surveyed will depend on their response rates. A second round of surveying will solicit all national legislators in the lower house in Brazil (513), Pakistan (342), and Italy (630). The final number of politicians surveyed will depend on their response rates.

Italian sample:
167 citizens assigned to Treatment 1, order 1
167 citizens assigned to Treatment 2, order 1
166 citizens assigned to the control group, order 1
167 citizens assigned to Treatment 1, order 2
167 citizens assigned to Treatment 2, order 2
166 citizens assigned to the control group, order 2

Brazilian sample:
167 citizens assigned to Treatment 1, order 1
167 citizens assigned to Treatment 2, order 1
166 citizens assigned to the control group, order 1
167 citizens assigned to Treatment 1, order 2
167 citizens assigned to Treatment 2, order 2
166 citizens assigned to the control group, order 2

Pakistan online sample:
83 citizens assigned to Treatment 1, order 1
83 citizens assigned to Treatment 2, order 1
84 citizens assigned to the control group, order 1
83 citizens assigned to Treatment 1, order 2
83 citizens assigned to Treatment 2, order 2
84 citizens assigned to the control group, order 2

Pakistan in person sample:
167 citizens assigned to Treatment 1, order 1
167 citizens assigned to Treatment 2, order 1
166 citizens assigned to the control group, order 1
167 citizens assigned to Treatment 1, order 2
167 citizens assigned to Treatment 2, order 2
166 citizens assigned to the control group, order 2

Politicians sample (for all countries):
1/6 of politicians assigned to Treatment 1, order 1
1/6 of politicians assigned to Treatment 2, order 1
1/6 of politicians assigned to the control group, order 1
1/6 of politicians assigned to Treatment 1, order 2
1/6 of politicians assigned to Treatment 2, order 2
1/6 of politicians assigned to the control group, order 2

An independent randomization will randomize the legislators in the following groups:
1/4 of politicians will receive the Information Treatment, pubic pledge on anti-corruption
1/4 of politicians will receive the Information Treatment, public announcement of legislation
1/4 of politicians won't receive the Information Treatment, pubic pledge on anti-corruption
1/4 of politicians won't receive the Information Treatment, public announcement of legislation