Experimental Design Details
The entire experiment consists of three stages. In stage 1, participants answer a short questionnaire regarding the attractiveness of multiple (hypothetical) employment opportunities in urban and rural areas. This short survey study is unrelated to the subsequent stages and part of a separate research project. For the completion of stage 1, participants receive a fixed payment of 2 €.
In Stage 2, participants play a modified version of the “rule-following game” by Kimbrough and Vostriknutov (2018). Each participant is asked to allocate 30 balls between a blue and a yellow basket. For each ball that is placed into the blue basket, the individual chance of winning an additional payment of 2 € increases by 1%. For each ball allocated to the yellow basket, the chance of receiving the payment increases by 2%. However, the instructions further introduce a trivial but costly rule to all subjects, which asks them to put all 30 balls into the blue (unprofitable) basket. The lottery draw (based on the individual ball allocation) is conducted by the computer program at the end of the experiment. This task allows us to measure participants general inclination towards rule-following, which has been shown to be a good predictor for pro-social (norm consistent) behavior in individual decision-making (dictator games). We want study the predictive power of rule-following, in regard to pro-social behavior, in a more complex hierarchical workplace setting.
At the beginning of stage 3, subjects are allocated into work groups of three: one subject is randomly chosen to act as the group manager and while the two remaining individuals take the roles of regular employees. Each work group is entrusted with a donation of 10 €, which is going to be transferred to a randomly chosen children’s hospice in Germany. We ensure that the donation procedure is verifiable to all participants. Participants have the possibility to generate personal income for themselves by decoding letters in a simple real effort task (Erkral et al., 2011) for 10 minutes. We ensure a thorough understanding of the task by introducing a 1-minute trial round. It will be stressed that own effort translates directly into own income. The explicit task compensation will be determined at the group level by the implemented production type. Production type “A” promises a high piece rate of 8 (7) Cents for the manager (employees), while production type “B” offers a low piece rate of 6 (5) Cents. However, production under “A” adversely affects the allocated group donation. For each letter that is encoded by one of the group members, the group donation decreases by a fixed amount of 2 Cents. Therefore, each individual who works under the high wage production scheme A – the one including the negative externality - is confronted with the internal conflict of balancing selfish motives (generating personal income) against ethical concerns (reducing the donation for children and families in need).
Starting from this basic group production setup, we vary the workplace setting across two dimensions (= 2x2 treatment design). First, we will manipulate the procedure on how the production type is implemented – either by a deliberate decision made by the group manager or through an impersonal draw conducted by the computer program. This manipulation allows us to see whether workers have a higher tendency to justify selfish behavior on the job, if they can shift the personal responsibility for the implementation of the production externality to an internal authority. The second dimension varies the information available to the employees. Under the full info condition, employees receive information on both production types. This contrasts the limited info scenario, under which workers only receive information about the production type that was ultimately implemented. We hypothesize that employees, who are fully aware of the payoff differential (about 33%) between both production types, are more likely to perceive the production externality as being less despicable, as its introduction serves a clear self-benefitting purpose.
Summary of the four treatment conditions:
In the Choice + Info Treatment, the manager can freely choose between production type A and B. The employees know that the manager implemented the production type deliberately. The employees receive information on both available production types.
In the Computer + Info Treatment, the manager cannot choose between production type A and B. The computer program implements the production type automatically. The employees know that the computer implemented the production type automatically. The employees receive information on both available production types.
In the Choice + No Info Treatment, the manager can freely choose between production type A and B. The employees know that the manager implemented the production type deliberately. The employees only receive information about the type that was implemented.
In the Computer + No Info Treatment, the manager cannot choose between production type A and B. The computer program implements the production type automatically. The employees know that the computer implemented the production type automatically. The employees only receive information about the type that was implemented.
To test our above-mentioned conjectures, our main analysis will compare employees’ effort level (number of encoded letters) under production type A across our four treatment settings (between subjects design).
After finalizing the production task, all participants receive information on all relevant monetary outcomes generated throughout the experiment. Subsequently, all individuals complete a short questionnaire that gathers information on socio-demographic characteristics, prosociality, risk aversion, reciprocal inclination, and the Big 5 personality traits. These personal characteristics allow us to derive a better understanding on how different types of workers may react to the introduction of negative production externalities and whether certain types are more prone to justify selfish behavior than others. Furthermore, participants are asked to evaluate the social appropriateness of implementing either production type A or B in a hypothetical workplace scenario similar to the “Choice + Info Treatment”, using the incentivized method by Krupka and Weber (2013). Finally, subjects have the opportunity to donate a fraction of their income to a charity organization of their choice (“German Red Cross”, “Médecins Sans Frontières (Ärzte ohne Grenzen)”, “UNICEF” or “Brot für die Welt”). Income is paid out in cash.