Supervisors’ reactions to penalty contracts

The intervention in this study operates within a principal-agent framework, where participants are randomly assigned either a bonus or penalty contract (between-subject design). Agents work on a real effort task. Agents’ possible effort levels are quantified in points reached in the real effort task and grouped into categories. Supervisors benefit financially from each point achieved by their assigned agent. Supervisors evaluate agents across effort categories. In the bonus treatment, supervisors determine whether the agent receives a bonus or not for each effort category. In the penalty treatment, agents receive an upfront payment and supervisors determine whether the agent must pay a penalty or not for each effort category. Importantly, the financial incentives are equivalent across both treatments. The key distinction lies in the framing: receiving a low (high) payment in the beginning and a potential bonus (penalty) later.

To investigate whether supervisors' evaluation behavior is influenced by reciprocity considerations, the study also employs a within-subject design involving two treatments. In one treatment, agents have the option to reward or penalize the supervisor in response to their subjective performance evaluation. In the other treatment, this option is absent. This design allows for a nuanced examination of how the presence or absence of reciprocity options may impact supervisors' rating behaviors in the context of the described bonus and penalty contracts.

2024-11-08

2025-11-30

Our primary outcome is supervisor’s evaluation behavior. We collect data of supervisors’ evaluation behavior using the strategy method. For each effort category, representing all different effort levels possible in the real effort task, the supervisors make a binary decision by determining whether the agent receives the bonus (must pay the penalty) or not. We will base our analysis on the threshold category, i.e. the lowest category for which the supervisor assigns a bonus (do not assign a penalty) to the employee.

Behavioral change in supervisors’ evaluation behavior for treatments with agents’ reward/ punishment option and without this option; beliefs on leniency bias; employees’ effort levels; employees’ reciprocal behavior

To shed light onto the interplay between framed incentive contracts and subjective performance evaluation, we will conduct a laboratory experiment comprising five parts and a survey. Participants will act in a principal agent setting. Thus, there are two different roles: agent (employee) and principal (supervisor). Participants will remain in the same role throughout the experiments and will be paired with a player from the other role. Notably, in Parts 2 through 5, each participant in the role of the agent will be randomly matched with one other participant in the role of the principal, following a perfect-stranger design.

In Part 1, participants in the role of the agent will work on a real effort task to familiarize them with the task. In Parts 2 and 3, participants are randomly assigned either a bonus or penalty contract. In the bonus treatment, agents perform the same real effort task as in Part 1 and receive a bonus or not based on their supervisors’ evaluation. In the penalty treatment, agents receive an upfront payment before starting to work on the real effort task, and the supervisors' decisions determine whether they have to pay a penalty or not. The incentives are financially equivalent in both treatments. The only distinction between the treatments lies in how they are framed (receiving a low (high) payment in the beginning and a potential bonus (penalty) based on the supervisors’ decision later). To elicit supervisors’ evaluation behavior, we will implement the strategy method. I.e. for each effort category, which reflects the various possible effort levels in the real effort task, supervisors make a binary decision: whether the agent receives a bonus (must pay a penalty) or not.

To assess whether supervisors’ evaluation behavior is driven by reciprocity considerations, participants undergo a within-subject design in Parts 2 and 3: in one part, agents will have the option to reward or punish the supervisor as a reaction to their subjective performance evaluation, and in the other part, this option will be absent. We will randomize whether the reward/punishment option will be implemented in Part 2 or in Part 3. One of these parts will be randomly chosen for payment.

Further, we measure the time the supervisors take for their decisions in Part 2 and Part 3.

In addition, we will elicit supervisors’ and employees’ beliefs of the expected rating behavior. Belief elicitation will be incentivized.

In Part 4, we will assess agents’ loss aversion and supervisors’ loss aversion for other peoples’ payments. In Part 5, participants play a dictator game with the supervisors in the role of the dictator.

Supervisors will complete a survey on self-report scales (risk aversion, self-image concerns, empathy, personality traits) and answer demographic questions. Employees will also complete a survey on risk aversion and demographic questions.

Not available

computer

Experimental session

No

In total about 544 participants

In total about 544 participants

For each of the experimental treatments (bonus / malus treatment) about 136 groups, each consisting of an employee and a supervisor (between-subject design)

For calculating the required sample size, we set the error probability to 0.05, the power to 0.80, and the minimum-detectable effect size to 0.35. Based on a two-sided Wilcoxon-Mann-Whitney test, these parameters yield a required sample size of 136 groups (272 participants) per treatment.