Access to Gig Work and Exposure to Inequality

This trial includes two complementary experimental interventions. First, individuals interested in renting a motorcycle to perform platform-based work will be randomly assigned to receive standard offers or discounted offers. Second, participants who recently acquired a motorcycle and are starting to perform platform-based work will be randomly assigned to receive monetary incentives encouraging them to work for a short period in different types of neighborhoods.

2026-08-24

2026-12-31

For the access component: income, labor supply.

For both: aspirations; worldviews; preferences for redistribution; attitudes towards the rich

Aspirations: index based on endline survey questions on income and education that the participant hopes to reach (and their children to reach) in different time horizons.

Worldviews: endline survey questions on whether participants believe that income differences derive from luck or effort, and whether such differences are fair or unfair.

Preferences for redistribution: index based on questions about particular policies.

Attitudes towards the rich: index based on behavioral games played against someone living in a wealthy neighborhood + questions on "warmth" towards a resident of wealthy neighborhoods.

For the exposure experiment: Psychological Well-Being; Voting Preferences; Political Engagement

Psychological Well-Being: index based on traditional mental health questionnaires (PHQ-8; life-satisfaction index; self-esteem; stress)

Voting Preferences: elicited before and during the 2026 presidential elections in Brazil

Political engagement: index based on endline questions about expected engagement during the elections (e.g., sharing campaign materials with relatives and friends or voluntarily working for a campaign)

This study consists of two individually randomized experiments conducted with potential and new clients of a motorcycle rental and leasing company in Brazil. Most clients use the motorcycles to work as delivery or ride-share drivers through mobile apps.

In the Motorcycle Access Experiment, we will embed randomized commercial offers into the company’s regular registration procedures for approximately 5,000 potential clients. Participants will be invited through the company’s app or WhatsApp and will complete an online baseline survey before receiving their offer. The three study arms are as follows:

- Control: Participants will receive an offer at the prevailing price.
- Discount: Participants will receive an offer with a discount on the initial deposit.
- Surprise Upgrade: Participants will initially receive the prevailing-price offer. Among those who accept, a randomly selected subset will receive an unanticipated upgrade and pay the discounted price.

Outcome data will come from the partner company’s administrative records and an endline survey conducted up to twelve months after recruitment.

In the Neighborhood Exposure Experiment, we will recruit approximately 2,000 new clients who intend to perform app-based delivery or ride-share work. Participants will be invited through the company’s app or WhatsApp and will complete a baseline survey. They will then be independently randomized along two dimensions. First, participants will either receive brief factual information about inequality across neighborhoods in São Paulo or receive no additional information. Second, participants will be assigned to one of the following groups:

- Affluent Neighborhood: Participants will receive a monetary incentive to work for a short period in a highly affluent neighborhood.
- Comparison Neighborhood: Participants will receive a monetary incentive to work for a short period in a comparison neighborhood.
- No Place-Based Incentive: Participants will receive no monetary incentive tied to working in a specific neighborhood.

Work destinations will be selected based on participants’ residential locations and pre-existing neighborhood characteristics, such that treatment and comparison neighborhoods are similar along dimensions other than wealth and income. Outcome data will come primarily from an endline survey conducted after the incentivized work period.

Not available

Randomization done in office by a computer

individuals

No

5000 in the access component; 2000 in the exposure component

5000 in the access component; 2000 in the exposure component

Access Component: 1,000 individuals receive the control offer (full price); 2,000 individuals receive the discounted offer; 2,000 individuals receive the full offer originally and are unexpectedly upgraded after accepting it.

Exposure component: 920 individuals are assigned to treatment (incentives to work in a top-wealth neighborhood); 300 to matched control; 300 to control unmatched in the racial dimension; 300 to control unmatched in the political dimension; 180 to status quo (no place-based incentive)

Access component: Assuming a 90% endline response rate, a 20-percentage-point difference in motorcycle acquisition between the discounted and surprise-upgrade arms and that baseline covariates explain 50% of outcome variation, the experiment can detect an acquisition effect of R$268 per month on earnings (16.2% of the baseline mean) and 4.9 hours of work per week (11.6% of baseline hours) at a 5% level of significance and 80% power. Exposure component: Assuming a 90% endline response rate, a significance level of 5\% and power of 80\%, I would be able to detect an effect of working in a rich neighborhood (versus any control neighborhood) of at least 7.6 percentage points (12\%) on voting for a pro-redistribution candidate, of at least 6.4 percentage points (25\%) on working at a rich neighborhood, and of at least 0.16 standard deviations on an index of attitudes towards the rich or an index of preferences for redistribution.