Autonomy and Performance Pay - A Field Experiment with a Large Retail Chain

How to cite

Vogelsang, Timo; Manthei, Kathrin; Sliwka, Dirk. 2017. "Autonomy and Performance Pay - A Field Experiment with a Large Retail Chain." AEA RCT Registry. AEARCTR-0002126. https://doi.org/10.1257/rct.2126-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Timo Vogelsang* , Kathrin Manthei , Dirk Sliwka

* Primary investigator

Registry ID
AEARCTR-0002126
DOI
https://doi.org/10.1257/rct.2126-1.0
First registered
Last updated
Status
In development
Trial dates
2017-03-24 – 2017-12-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/2126

Abstract

Within a large retail chain we introduce a bonus payment and provide more autonomy in decision making for store managers, implementing a 2x2 experimental design. The bonus is based on the store’s profit contribution margin (= sales - costs of goods sold - inventory shrinkage - personnel costs).
Store managers in one region in South-East Germany are randomly assigned to one of four treatments. 25% of store managers receive a bonus based on the contribution margin. Another 25% of store managers receive more autonomy in their work by allowing them to decide on the items placed in front of the cash desk. 25% of store managers receive both, the bonus and the additional autonomy. The remaining 25% serve as control group.
Based on a theoretical model we study the following hypotheses: The bonus payment will increase the contribution margin and will do so to a stronger extent when it is accompanied by an increase in decision autonomy. We also hypothesize that the granting of autonomy alone (without the bonus) can have countervailing effects. On the one hand, store managers can make better use of local knowledge in decision-making and may be motivated by the increase in autonomy. On the other hand, decentral autonomy can have a cost if less able store managers cannot make effective use of their local knowledge.

Experimental design

The intervention takes place in one region (south-east) of Germany. We randomly assign districts within each region to either TreatAUT (1), TreatBON (2), TreatAUTBON (3), or the Control group (4) using stratifications depending on the predicted contribution margin in the first treatment month. Randomization at the district level, although the unit of observation is the store level, gives us the possibility to separate treatment and control groups and only inform them concerning their specific treatment interventions.

Importantly, all store managers (including the control group) get access to an online training on the contribution margin prior to the intervention. Additionally, they receive information about the relative profit margin of each product (this is novel information to them).
All store managers further receive a monthly overview about the development of the contribution margin (and the components) apart from the possible monthly bonus notifications.

TreatBON: Store managers receive a three-month bonus for an increase of their contribution margin above 80% of its planned value (following from the accounting plan for each store). In each month, store managers receive a bonus value of 0.05 * (contribution margin in € - 80% of the plan value). Cumulative bonus values are paid out after three month (capped at zero). Store managers receive monthly feedback on the respective bonus values.

TreatAUT: Store managers receive more autonomy for the experiment’s duration of three month. More precisely, they are now allowed to place items in front of the checkstand according to what they consider best.

TreatAUTBON: Is the combination of TreatBON and TreatAUT.
Control: The control group only receives the online training and the access to information about relative profit margins. They are not made aware of the other treatment groups..
We briefed the district managers how to react to different sorts of questions concerning the experimental design.

Control: The control group only receives the online training and the access to information about relative profit margins. They are not made aware of the other treatment groups.

We briefed the district managers how to react to different sorts of questions concerning the experimental design.

The company’s worker council agreed to this intervention. This serves as an IRB substitute.

Design details

Randomization method
Stratification method
Randomization unit
District
Clusters
32 districts
Observations
228 stores
Arms
Control: 58 stores TreatBON: 59 stores TreatAUT: 56 stores TreatAUTBON; 55 stores Note: There can be slight changes in the number of treated stores as pay experience has shown that the firm may reconfigure district compositions for administrative reasons at short notice. However, this should concern only very few stores and we document reasons and procedures.
Primary outcomes
The contribution margin on the store level (as well as all its components), further retail key figures, questionnaire data (pre and post)

Cited by

No citation data available for this registration yet.

Keywords