New Pay: An experimental analysis of self-set pay and social comparisons

Last registered on February 07, 2023

Pre-Trial

Trial Information

General Information

Title
New Pay: An experimental analysis of self-set pay and social comparisons
RCT ID
AEARCTR-0010897
Initial registration date
February 06, 2023

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
February 07, 2023, 11:36 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Region

Primary Investigator

Affiliation
University of Bern

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2023-01-31
End date
2023-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
New work demands for new pay, and while still many companies are hesitant to change their traditional incentive practices, more and more pioneering companies are experimenting with new forms of pay. One such new pay model incorporates the idea that employees can determine their own salaries. First studies examining the causal implications of self-set pay show higher effort provision, but also document relatively high wage levels.

Even though self-set pay can be, per definition, decided for individually, information about peer performance might still be highly relevant. The main goal of this online study is to examine the influence of performance comparison on self-set wages and to vary whether people are informed about the average output that was generated by others working on the same task. Specifically, the chosen wage levels are analyzed and also the proportion of the participants selecting the highest wage in both conditions (with and without peer performance comparison). Furthermore, I analyze the effect of delegating decision-rights to employees on effort provision by running two treatments with fixed pay, allowing comparison of performance under self-selected versus fixed wages in the same task.
External Link(s)

Registration Citation

Citation
Schumacher, Stefanie. 2023. "New Pay: An experimental analysis of self-set pay and social comparisons." AEA RCT Registry. February 07. https://doi.org/10.1257/rct.10897-1.0
Experimental Details

Interventions

Intervention(s)
Intervention (Hidden)
Intervention Start Date
2023-02-08
Intervention End Date
2023-02-28

Primary Outcomes

Primary Outcomes (end points)
- Wage level chosen in T1 and in T2 (between 0 and 2 GBP, in 10 cents increments).
- Proportion of people choosing the maximum wage in T1 and T2 (max. wage is 2 GBP).
- Effort provision in all four treatments (number of correctly solved words; comparing the data from T1 + T2 to T3 and to T4 (self-set salaries vs. fixed wages)).
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
For T1 and T2 only:
- Wage levels and wage choices of low performers.
- Wage levels and wage choices of high performers.
This is analyzed by looking at the two groups (high / low) and by looking at the interaction effect between effort and treatment (dependent variable = wage level)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The study is conducted online with four treatments. In all treatments, participants receive a base pay for participating (1 GBP) and an additional payment for working on a real-effort task (encryption task). In the two main treatments, participants can self-select their wage within a pre-defined wage band. In treatment 1, participants are informed upfront that they will first work on the task, second, they will see their performance, and third they will determine their own wage. The wage they set will be implemented and paid out after the study. The only difference in treatment 2 is that participants will additionally be informed about the average output generated by other workers working on the same task. This information is gathered in a pre-study, where participants face the same conditions as in the self-set wage treatments. Pre-collecting this information allows random assignment to all four treatment conditions in the main data collection phase.

Treatments 3 and 4 allow a comparison of effort provision under two different fixed wages to the treatments with self-set wages. One treatment offers a "great" salary (1 GBP), and one offers a salary that represents the upper bound of the self-set wage band of T1 and T2 (2 GBP).
Experimental Design Details
Task: As a real effort task, the Word Encryption task with Double Randomization (WEDR task) (Benndorf et al. 2019) is implemented.The task is to encode words into three-digit numbers. Participants can proceed with the next word only if they encoded all letters correctly. Effort is measured as the number of solved words. I chose this task for several reasons: It requires no special knowledge or cognitive abilities, learning possibilities are trivial, there is no room for guessing, and it is gender neutral (Benndorf et al. 2019).

Payment: Participants receive a base pay of 1 GBP for participating in the study. In addition, they can self-set their wage for their performance in the working stage in T1 and T2.

The procedure is done in two steps:
Performance: After performing in the encryption task, participants receive feedback about the number of correctly solved tasks.
Wage: Within a range of £0 - £2 participants can then set their own wage for their performance.

The study takes approximately 10 minutes and prolific recommends a total hourly salary of 9 GBP or 1.5 GBP for an assignment of 10 minutes. The lowest hourly wage that can be set is 6 GBP. A wage of 12 GBP, or 2 GBP for 10 minutes is considered a "great" salary according to the official webpage (as of February 2023). Choosing the maximum salary of 2 GBP and thus being paid 3 GBP for a 10 minute assignment would hence result in a salary far above standard or even great wages that are paid on prolific.

In T3 and T4, the wage participants receive for performing in the working stage is fixed to 1 GBP or 2 GBP, respectively, to mimic a "great" salary and to have a comparison of the people who self-set their wage to 2 GBP and of those who received 2 GBP as a fixed wage.
Randomization Method
Randomization is done by a computer. Every participant who clicks on the study link is randomly assigned to one of the four treament conditions. As more observations will be gathered in T1 and T2, the allocation ratio is 1/3 for the fixed pay and 2/3 for the self-selected wages treatments.
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
n=480 individuals

We will exclude subjects who do not complete the study within 44 minutes of starting and who are not approved for any other reason.
We will analyze the data both, for all participants and for those participants, who solve at least one task correctly in the working stage.
We will exclude participants who do not manage to solve the practice questions in 3 trials or answer the question on the payment scheme wrong twice or more times.
Sample size: planned number of observations
n=480 individuals
Sample size (or number of clusters) by treatment arms
Approximately n=160 in T1 and T2 (each) and n=80 in T3 and T4 (each).
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Analyzing self-set wages, I will compare wages levels in T1 to wages in T2. Based on a two-sided Wilcoxon-Mann-Whitney test, an error probability of 0.05, and a power of 0.80, we require about 160 participants per treatment to detect an effect of Cohen’s d of 0.32. For the proportion of people choosing the maximum wage, an effect of 14.6 percentage points will be detectable with 160 participants. For effort, I will compare effort provision in T1 + T2 to effort levels in T3 (1 GBP fixed wage) and T4 (2 GBP fixed wage). According to the literature, this effect should be rather large (approximately d=0.5). Based on a two-sided Wilcoxon-Mann-Whitney test, an error probability of 0.05, and a power of 0.80, we require about 80 participants per treatment to detect an effect of Cohen’s d of 0.36 (compared to T1 + T2). Between T3 and T4, the minimum detectable effect size is d=0.46 with 80 participants per treatment.
IRB

Institutional Review Boards (IRBs)

IRB Name
The Ethics Committee of the Faculty of Business Administration, Economics and Social Sciences of the University of Bern
IRB Approval Date
2023-02-06
IRB Approval Number
052023

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials