Sexual Corruption, Entrepreneurship Motivation and Migration Intentions: Evidence from a Survey Experiment in Burkina Faso

Respondents are randomly assigned with equal probability to one of two information conditions.

Treatment condition:
Treatment respondents receive an information about the objective prevalence of sexual corruption amongst women in Burkina Faso, as estimated in Borcan et al (2025).

Control condition
Control respondents receive neutral demographic information about the share of women in Burkina Faso as per the 2019 census.


Borcan, Oana and Gillanders, Robert and Aja-Eke, Doris and Maiga, Eugenie, The Threat of Sexual Corruption to Female Entrepreneurs: Evidence from a Pilot Study in Burkina Faso (December 03, 2025). Available at SSRN: https://ssrn.com/abstract=5851802 or http://dx.doi.org/10.2139/ssrn.5851802

Randomization is at the individual respondent level, with approximately 50% of respondents assigned to each condition.

Treatment condition
Treatment respondents receive the following information:
“A recently conducted survey in the 2 biggest cities in Burkina Faso found that 14 out of 100 women have experienced situations in which a public official says that they will give a government benefit (such as access to services or resources that they are entitled to, quicker service, approval of documents, a job or promotion , or avoiding a fine or imprisonment) in exchange for sexual favours such as sexual activity, inappropriate touching, exposing body parts, or posing for sexual photos.”
This figure was estimated from a survey on women entrepreneurs in Burkina Faso in Borcan et al (2026). The treatment therefore provides information about the prevalence of sexual corruption among women.
Control condition
Control respondents receive neutral demographic information:
“51.7% of Burkina Faso’s population is made up of women.”

2026-07-01

2026-08-31

Outcome 1: Entrepreneurial motivations
Outcome 2: Migration intentions

Entrepreneurial motivation is measured using four statements, each rated from 1 (“Strongly disagree”) to 7 (“Strongly agree”):
1. “I am ready to do anything to be an entrepreneur.”
2. “My professional goal is to become an entrepreneur.”
3. “I will make every effort to start and run my own firm.”
4. “I am determined to create a business venture in the futur[e].”
The primary entrepreneurial motivation index will be the arithmetic mean of these four items, standardized mean 0 and standard deviation 1 relative to the control group mean.
Higher values indicate greater entrepreneurial motivation.

International migration intention is measured using one question, which asks:
“How much, if at all, have you considered moving to another country to live?”
Responses are coded:
• 0 = Not at all
• 1 = A little bit
• 2 = Somewhat
• 3 = A lot.
This variable is the primary outcome for Hypothesis 2.
Higher values indicate stronger consideration of international migration.

International Migration Preparation and Domestic Migration intentions are exploratory outcomes

International migration preparation
Among respondents reporting some consideration of international migration, a further question measures the extent of planning or preparation:
1. no specific plans or preparations;
2. planning to move in the next year or two but not yet preparing;
3. currently making preparations, such as obtaining a visa.
Treatment effects on binary measures defined as some plans / no plans and concrete plans / no concrete plans will be reported as exploratory.
Domestic migration
One further question measures consideration of moving to another city or village within Burkina Faso on the same 0–3 scale as international migration intention. A linked question asked conditional on non-zero domestic migration consideration, measures planning or preparation.

The phone survey opens with a module on demographic and economic characteristics, followed by a module including questions concerning experiences and perceptions of sexual corruption. Thereafter, respondents are randomly assigned with equal probability to one of two information conditions.
Randomization is at the individual respondent level, with approximately 50% of respondents assigned to each condition.
The estimand is therefore the causal effect of receiving information about the prevalence of sexual corruption relative to receiving neutral demographic information. We do not interpret the experiment as identifying the effect of information relative to no information.
An open comprehension question is administered following the information statement. Respondents who do not understand the information are given the information again. Comprehension will not be used to define the primary analysis sample.
The primary analysis will therefore be a treatment effect analysis based on randomized assignment.

To assess the sensitivity of the responses to question eliciting the prevalence of sexual corruption experiences to privacy concerns, we also apply a garbling technique for 50% of all respondents (randomly assigned) who are also informed about this privacy procedure.

For a randomly selected 50% of respondents, responses are garbled as follows:
• if the true response indicates an experience of sexual corruption — Rarely, Sometimes, Often or Very Often — it is recorded as “Never” with probability 0.10;
• if the true response is “Never”, it is recorded as one of the four non-Never categories with probability 0.10, with the particular category selected at random.

We will retrieve the true estimate from this distribution and compare it to the prevalence estimate in the subsample without garbling, and to the prevalence to the questions about peers' experiences with sexual corruption and broader perceptions of the incidence of sexual corruption.

Randomisation done by a computer.

Individual

No

0

600 adult urban women

300 adult urban women

Pre-Analysis Plan - Sexual Corruption and Women's Entrepreneurship Motivation and Migration Intentions