Addressing the Enigma of the Gift: The Role of Social Relations in Gift Exchange

2016-09-23

2016-12-30

Study 1: Workers’ productivity (total number of characters, total number of characters inputted, number of references, number of typos). Study 2: Worker's productivity (Number of audios and minutes evaluated, actual time worked).

STUDY 1: Following the standard field-experiment design in gift exchange, we recruit subjects to build the electronic library of a professor in a Chilean University. Subjects older than 18 from 13 different high-education institutions were recruited as workers for a 6-hour work of inputting academic cites on a software tracking their performance without their knowledge. The job consisted of an initial training session and four work shifts of 50 minutes each with three coffee breaks and a half-an-hour lunch. The task was advertised a one time job offering the market wage.

Workers were randomly assigned to one of six treatments varying whether a gift was granted or not and the existence and strength of the social relationship with the employer. In particular, we have a 3x2 design, with and without a gift and three different levels of social relation: No-Social relation, Weak-Social relation and Strong-Social relation.

STUDY2: Similar to study 1, we recruit workers to perform a one time job (this time, evaluating audio files from a call center). Workers are recruited though a social media platform that advertises part-time jobs (students and non-students). We recruit them at the market wage and instruct them that this is a one-time 6-hour job, which can be executed from whichever computer they desire. They are only required to attend a training session. Their 6 hours of work are due three days after they complete the training. Payment occurs at the end of the training, before the job takes place.

Workers are randomly assigned to one of five treatments varying whether a gift was granted or not, the existence and strength of the social relationship with the employer and whether the employer explicitly states that she holds the expectation that workers will reciprocate the gift with higher effort.

This second study differs from the first in three fundamental aspects. First, by allowing subjects to work form home, it provides workers with enough room to reciprocate without fatigue or time restriction concerns. Second, it strengths the social relation treatments by manipulating not only workers interaction with the principal but the interaction with other relevant people involved (RAs). Third, it explicitly manipulates workers' second order beliefs about the principal's expectations on their effort response to the monetary gift.

To understand and build the social relation between the parties, we follow the anthropological and psychological literatures. Since Mauss (1954), anthropologists have claimed that recipients will not interpret all monetary transfers as a gift. For a monetary transfer to be perceived as gift, it must be the case that the social relation between the parties is close and intimate. For a working definition of closeness and intimacy, we follow the psychological model by Reis and Shaver (1988), defining intimacy as a process that ``starts when a person expresses personal feelings or information to another […] to what the listener responds supportively and empathetically”. Based on this idea, we create a strict protocol that tightly controls the interaction between the principal and the workers. The No-Social relation treatment replicates the standard gift-exchange design were the employer commissioning is not present in any moment (STUDY 1) and training occurs through and online platform (STUDY 2). In the Weak-Social relation the professor is present at the start of the training period and limits the interaction to subjects to being introduced to workers by the RA conducting the training. In the Strong-Social relation the employer builds a social relationship with the workers by engaging in conversation with them following a protocol tailored by Reis and Shaver (1988)’s definition of intimacy (STUDY 1 and 2). The protocol was carefully designed to avoid workers making inferences about their productivity from their interaction with the principal and RAs. In STUDY 1, the employer-employee interaction takes place in the first and second work shifts, the latter taking place half an hour before the research assistant grants the gift in the Gift treatments. In STUDY 2, interaction occurs at the beginning and end of the training session.

Because social relationships are the core of this paper null hypothesis, In STUDY 1 research assistants were not assigned by rooms, but rather by task, in order to make sure that all other (potentially relevant) social relationships are held constant across treatments. In STUDY 2, the same set of RAs interacted with subjects in all conditions.

Once all the data is collected, in STUDY 1 we will contact subjects by email for them to participate in a survey measuring their satisfaction with the task, payment and gift (if correspond), their level of perceived closeness to the employer (following the closeness measures validated in Gatcher, Starmer and Tufano (2015)) and their social preferences. In STUDY 2, closeness measures are obtained before the gift is granted in the training sessions by implementing a short online survey.

In both studies randomization takes place at the time of recruitment. As interested students contacted the research assistant in charge of recruiting, they were randomly offered two possibles dates to work, each with two possible time schedules. In STUDY 1, we did not randomly assign people to a give date and time as that affects the external validity of the study. In STUDY 2 we did random assigned workers to shifts as the pool of applicants is extremely large and thus it does not affect the external validity as groups of friends are unlikely.

Worker level (STUDY 1 and 2).

No

In STUDY 1 workers were randomly assigned to two time schedules as they applied. In STUDY 2 we had a large number of simultaneous applicants, which allowed us to construct strata by gender, occupation, and age to randomise within the resulting clusters (36 clusters)

Around 200-250 workers in both studies (actual number will depend on the amount of no shows).

40 to 50 workers by each treatment.

Field studies on gift exchange are extremely disperse in their results. Thus, it is impossible to do power calculations without cherry pick.