Experimental Design
Participants are small scale businesses located across markets in Lusaka, Zambia.
After a short survey, participants are assigned to participate in a trust game designed to simulate real-world economic interactions. In this game, a “sender” decides how much money to send to a “receiver” for a business opportunity framed as a partnership, knowing that the amount sent will be multiplied by three. The receiver then decides how much of the multiplied amount to return to the sender. To incorporate institutional support, senders are given the option to complain to the their market chief if they think the amount returned back by the receiver is not fair. The market chief acts as an adjudicator and decides on the final allocation of tokens (that represent real money). There are two experimental treatments: 1) in the gender-blind treatment, the sender can complain to the market chief with a complaint form that only shows to the chief the choices of the two players, nothing else about their identity, 2) in the gender-revealed treatment, the sender can complain to the market chief using a complaint form that also shows the sender’s gender, in addition to the choices made by the sender and the receiver in the game. The receiver knows that the sender can complain and that complaints keep both players’ identities anonymous, but does not know about the existence of two possible complaint forms.
The sampling strategy aims to achieve balance in gender and industry (manufacturing vs non manufacturing). Our main target industries for senders are manufacturers, hairdressers/barbers and restaurants. For receivers, we target all the remaining industries.
We complement this main experiment with three additional activities:
- At the end of the experiment, we elicit incentivized beliefs about the chief’s choices when s/he has to arbitrate complaints coming from men or women. The assignment of complaints brought by men or women will be randomized between subjects, stratifying by the main strata and treatment assignment. This exercises will give us a proxy of player’s perceptions of the chief’s gender bias in the game.
- We conduct a survey with the chiefs that are involved in the trust game which asks about their behavior in arbitrating inter-gender disputes, and also how they would arbitrate in different types of complaints coming from female or male player in the trust game. Chiefs will answer about a set of trust game complaints for both female and male players, so we can compare differences in their arbitration behavior between the two genders. The survey questions and the complaints’ questions will give us measures of the chief’s gender bias. For the survey questions, we will build an index of the chief’s gender bias as in Ashraf, Delfino & Glaeser (2022).
We will perform treatment heterogeneity analysis on measures of these perceived and actual chief’s gender bias measures, by gender.
- In the pre-games survey, we will ask several questions about the players’ experiences with their market chief, other institutions, disputes with other businesses and harassment in business. This will allow us to understand more broadly how past experiences may shape behavior in the game. We will do heterogeneity analysis also on this aspect.
The main hypothesis is that trust by women in the gender-revealed arm will depend on the perceived and/or actual gender bias. Women may send more (less) in the gender-revealed treatment than the gender-blind one if they think the chief is biased in favor of (against) women.