Rebuilding Trust, Social Cohesion and Democratic Values

2021-05-05

2021-06-30

• Administrative data on turnout (and for the urban subset, surveyed voting choices) at the June 2021 election.

• Preference for democracy using questions adapted from Latinobarometer. We will create an index with the following 3 questions (standardizing each and summing them and standardizing the sum), but also show effects separately for each.
a. “In general, would you say you are very satisfied, somewhat satisfied, a little satisfied or not satisfied with the democracy in Mexico?”
b. “Do you consider it essential to live in a country where governors are elected by the majority’s vote?”
c. Please, state if you strongly agree, agree, disagree or strongly disagree with the following statement: “Democracy may have its problems, but it is still the best form of government”

• Stealing from /donating to the municipality: using the Resource Allocation Game --- following Lowes et al (2017, ECMA), and from neighbors. We conjecture the stealing from politicians will be larger and more precise than stealing from neighbors, as the videos are about politicians. However, there may still be (small) spillovers to neighbors.
a. We conjecture also that people will tend to keep for themselves higher denomination coins more than lower denomination ones.

• Survey measures of trust in others (from the community and outside, including politicians of different parties). Note that corruption may erode social capital and trust not only in politicians, but in a broader cross section of groups of institutions or people. These are the primary questions on trust we will use.
a. Please state, for each one of the groups, institutions or people mentioned in the following list, how much trust do you have in them? A lot, some, a little or none?
i. Political parties
ii. Political parties
iii. MORENA, PRI or PAN politicians
iv. In a person from same the community
v. In a person from other community
vi. In neighbors
vii. In rich people (relevant mostly for the Mexico video arm)
viii. In non-indigenous (relevant mostly for the Mexico video arm)
ix. Media
x. Army

b. Speaking in general terms, would you say most people can be trusted, or do you believe one is never careful enough when dealing with other people?

• Perceptions about the prevalence of corruption. We expect that the videos will cause people to update upwards on average on the level of corruption in the country. We will create an index with the following 4 questions (standardizing each and summing them and standardizing the sum), but also show effects separately for each.
a. How much progress do you think has been made in reducing corruption within State institutions in these past 2 years?
b. Please state if you strongly agree, agree, disagree or strongly disagree with the following statement: “All politicians are corrupt”
c. What percentage of politicians in Mexico do you think are involved in acts of corruption?
d. Out of every 100 pesos from tax revenues, how much do you think is stolen by politicians?

• Survey measures of voting
a. If federal elections were to be hold this Sunday, which party would you vote for? We conjecture that the incumbent video will reduce the intention to vote for the incumbent, and the analogous will happen for the non-incumbent parties.
b. How likely are you to go to vote in the 2021 election?

We will measure these a few minutes after treatment, and also 1-2 months later. The baseline surveys are done in person, but the follow up survey could be done in person, by email, or by phone.

Secondary outcomes gather conjectures which are not as important to us, or those for which we think our measures are weaker. We list them here for completeness only.

• Revealed preference willingness to participate as an electoral observer in the June 2021 election or to donate money to for electoral observers. This will only be measured for surveys conducted before the election.

• Hypothetical questions about tradeoffs between freedom and surveillance.
a. Would you agree to have the army run checkpoints in the streets to maintain order?
b. Would you support a government controlling the media and bugging phones to fight corruption?

• Propensity to cheat and to help: Hypothetical questions on paying taxes, donating blood, and skipping the order to get vaccinated for COVID even if its not your turn.
a. If you were asked to donate blood to a complete stranger (with no economic remuneration), would you accept?
b. On a scale from 1 to 10, where “1” is “not justifiable” and “10” is “completely justifiable”, how justifiable do you think it is to Avoid taxes.
c. On a scale from 1 to 10, where “1” is “not justifiable” and “10” is “completely justifiable”, how justifiable do you think it is to: Get the COVID-19 vaccine before your turn

• A trust game: amount sent in a trust game either to a member of congress (“Diputado”) or a member of the community (“Encuestado en tu municipio”). We attempted an implementation akin to that of Alan et al (2021, QJE), however this was done in a one-shot manner socially distant due to Covid, with the receivers and senders that appeared abstract to the participants. Pilots revealed that our implementation was hard to understand for both surveyors and for participants. Nonetheless we attempted this with the full trial. To address potential lack of understanding we will implement a procedure to over-weight participants whose characteristics match those for whom the correlation between the self-reported trust measure and choice in the trust game is strongest in the control.

• We will also randomize who gets the Resource Allocation Game, which allows us to measure its effect on the later trust game, and on the finance arm.

We will have 5 main arms:
1. Incumbent corruption video: A video showing official information about how prevalent and costly to the pockets of citizen certain kinds of corruption are using data from INEGI, and also showing them footage of corruption scandals of major politicians in Mexico in the last 10 years of the current ruling party (Incumbent corruption)
2. Opposition corruption video: A second video like the first, but with corruption scandals of members of the opposition.
3. Economics Video. A third video summarizes some of the (unfortunately mostly negative) economic results for Mexico in recent years (economic growth, poverty, and the handling of the COVID19 crisis, compared to other Latin American countries).
4. Mexico video. Common identity video and perspective-taking: a motivational video that primes individuals to recall many aspects of Mexican history and culture that forge its common identity, and including a common past with indigenous people, and how it took centuries to have a democracy with power alternating parties. It also discusses the importance of getting involved to shape a common future, and the importance of protecting democracy. We combine the video with a perspective taking on why people should vote and participate in civic life.
5. Control group (no video)

Some of these groups are cross-randomized with the “Stock treatment”. We will give a subset of people financial assets (stock) that tracks the Mexican stock exchange (200 pesos of stock), which is about 1.5 days of salary for them.

Randomization done in office by a computer.

Randomization at the individual level.

No

4000 individually randomized adults drawn from 66 stratified municipalities.

4000 subjects: adults with a valid voting ID.

On the main arms we have about 600-1000 observations per arm. See the respective table in the pre-analysis plan.

• Incumbent corruption video G1 650
• Opposition corruption video G2 650
• Economics video G3 300
• Mexico video G4 1000
• No video (Control group) G5 1400

These are cross-randomized with a financial treatment assigned to 1000 in the urban sub-sample.

Given individual level randomization, 95% confidence, and 80% power, we have power to detect the following effects approximately: • Trusting: 3 percentage points in trust comparing any pair of the groups. • Reciprocity: 3 percentage points in reciprocity comparing any pair of the groups. • Voting turnout: 5 percentage points in a comparison of corruption: Corruption (G1+G2) vs Control. • Preference for democracy: we can detect an effect of 0.2 points in the classic 4-points Likert “satisfaction with democracy” question. • Perceptions about corruption: For the question “Out of 100 how many politicians are corrupt” we can detect an increase of 3pp.