Experimental Design
The experiment is based upon Joyce Berg’s "investment game” (1995). An even set of subjects is randomly allocated to two rooms, A and B, half of the group in each. Each session requires six, eight, or ten subjects. Each subject is allocated a tablet, a set of printed instructions, some paper, and a pencil. Subjects cannot communicate with each other. During the experiment, subjects in room A will be framed as “investors”, and agents in room B “entrepreneurs”. Sanitary precautions will be taken in consideration of the pandemic, including the use of surgical masks, hand sanitizer, room ventilation, and social distancing. Tablets and desks will be sanitized before each new game, and paper will be properly discraded.
The game has a duration of twelve rounds. Four rounds are control rounds, and the last eight rounds are treatment rounds. Before starting the game, three practice rounds take place. Each subject will have a constant endowment per round of 40 bolivianos ($us 5.80), so that we can prevent effects related to inequality aversion. At the beginning of each round, each investor is randomly and anonymously paired with an entrepreneur. Each investor can invest any amount, including zero, from her 40 bolivianos in multiples of five, by sending them to the paired entrepreneur. Upon receiving it, the investment sent to the entrepreneur is multiplied by three. Hence, the maximum amount that can be received by the entrepreneur is 120 bolivianos. The entrepreneur can then choose to send back to the investor any amount from the resulting gains, again in multiples of five. Once the entrepreneur decides the amount to send back, the round ends. This procedure is repeated in the remaining rounds.
Before the treatments start, some general rules are laid out and are enforced by the experimenter. The main rules are:
• Pairs are anonymous, meaning that investors do not know with which entrepreneurs they are randomly matched.
• Players have no information on the preceding behavior of the group or of specific individuals
• The experimenter does not intervene or regulate the actions of the players beyond defining the strategy space and explaining the rules
• The experimenter guarantees the fulfillment of each agent’s payment. Payments correspond to the outcome of a randomly selected round using a fair twelve-sided die.
During the experiment, we intend to identify ways to increase investment and retribution by exposing the agents to different sets of rules, i.e. different institutions, which will vary the information available and the expected payoffs. We use a between-subject design, assigning participants to one treatment only. However, we exploit within-subject variation by using the first four rounds in the game to establish a baseline of interpersonal trust. In these rounds, no innovation is introduced and all participants in a group start with the setting described previously. From round 5 on, participants are assigned randomly to one of four treatments. The treatments are as follows.
Treatment 1 (Social Reputation): Starting at round five and for each following round, participants are informed of how many entrepreneurs reimburse less than the investment in the previous round. This information intends to create incentives to conform to a social norm.
Treatment 2 (Individual Reputation): Starting at round five and for each following round, investors are informed of the history of repayments by the entrepreneur with whom they are matched at each round, providing an opportunity for the entrepreneur to build up a favorable reputation.
Treatment 3 (Imperfect institution): Starting at round five and for each following round, participants are told that if entrepreneurs reimburse less than what has been invested in any round, they might be punished with a certain probability. If punished, they will only receive the show-up fee at the end of the game. Participants are issued a warning at the end of the round. The warning states that the experimenter has been notified of the scam and that there is a 5% chance that the entrepreneur will only receive the show-up fee at the end of the game, excluding therefore any gains made during the game. For each additional “scam” move, there is an additional 5% chance of facing charges, therefore a maximum of 40% probability of being sanctioned if she scams at every treatment round. The investor is informed of these rules.
Treatment 4 (Perfect Institution): Starting at round five and for each following round, participants are told that if an entrepreneur reimburses less than what has been invested in at least one round, then he o she will be punished and will only receive the show-up fee at the end of the game.
Concerning the implementation of the experiment, it will be programmed using Google sheets, benefiting from the cloud computing possibilities which allows the use of tablets in a mobile lab, facilitating sanitary precautions (papers are not exchanged, the set-up and duration of the sessions are relatively short), and can permit a sufficiently simple and familiar interface to accommodate the sampled population.
The experiment will be conducted at the facilities of Universidad Privada de El Alto, a public university located near informal commercial zones in the city of El Alto and in the facilities of Universidad Privada Boliviana in La Paz. Recruitment will be made in different popular markets, informal trading zones, and small businesses of the cities of La Paz and El Alto. Inside the university buildings, we will adequate two classrooms to conduct the experiment and participants will be distributed in such a way that they won’t be able to see the strategies of one another. A bio-safety protocol will be implemented and all material should be prepared well in advance and ready to start fieldwork as soon as contagion rates are sufficiently low.