Rotation of staff in microfinance
How to cite
Stryjan, Miri; Deserranno, Erika. 2019. "Rotation of staff in microfinance." AEA RCT Registry. AEARCTR-0004891. https://doi.org/10.1257/rct.4891-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This project studies the effect of implementing staff rotation in a large microfinance institution. Specifically, we compare between offices randomly assigned to regularly rotate their loan officers between borrowers (portfolios) to offices randomly assigned to freeze rotation.
The incentives faced by employers and employees are often misaligned - the classic principal-agent problem. Our project will study this phenomenon in the context of managerial practices in the financial sector, and in particular, in relation to policies pertaining to staff rotation. In developing counties, a large share of loans is extended by microfinance institutions, where personal and frequent contact between borrowers and lenders is a necessary norm. This makes microfinance a particularly relevant context for the study of staff rotation – a management practice that directly impacts the scope for personal relationships to form between loan officers and borrowers.
The study will examine the effect of loan officer rotation on portfolio quality, as well as staff and borrower behavior, types and welfare. It is carried out in partnership with BRAC Microfinance in Uganda. For a study period of 12 months, 53 randomly selected branches (local offices) rotate groups between the loan officers every six months, following a schedule prepared by the research team together with BRAC, while 53 control branches will freeze rotation. Detailed administrative and survey data is collected on both staff and borrowers, before, during and after the study period.
Whereas the related empirical literature has utilized endogenous organizational policies for loan officer rotation to document features of loan officer behavior and draw conclusions about their incentives (see e.g. Herzberg, Liberti and Paravisini, 2010; Bhowal , Subramanian and Tantri 2014; Drexler and Schoar, 2014 and Fisman, Paravisini and Vig, 2017), we propose to implement a randomized experiment to study the causal effect of implementing a rotation policy. Our experiment introduces exogenous variation in within branch rotation. While the related previous papers study variation in rotation at the individual level, we randomize rotation at the branch level, which allows us to study organizational responses to and group level effects of rotation, such as its impact on team work at the branch level.
Intervention
We collaborate with BRAC Uganda Microfinance to evaluate the effects of a rotation policy in which clusters of borrower groups are rotated among credit officer on a regular basis. We randomize branch offices to either
(a) rotation in credit officer-to-group allocation
(b) freeze in credit officer- to-group allocation
and ask how staff rotation affects outcomes such as credit officer effort, timing of loan approvals, team-work, corrupt activities in the field and borrower pool characteristics.
Experimental design
In January 2019, BRAC Uganda Microfinance implemented a schedule for rotation of credit officers to clusters of borrower-groups in 53 of its branches while in 53 branches, rotation was put on hold. In rotation branches, rotation was to be implemented in January and July 2019, while in control branches, rotation was not permitted for all of 2019. The researchers collaborated with BRAC in this process and randomly chose branches to be assigned to rotation and freeze in rotation, respectively. Within the rotation branches clusters of borrower-groups were re-allocated between credit officers according to a randomized order, prepared by the researchers. The new rotation policies for 2019 were announced to the 106 branches in December 2018.
The randomization is done at the branch level and stratified on area (branches are grouped in areas with 4-5 branches in each), and on the following branch level variables: repayment performance in the past year, number of staff and rural/urban status of the branch.
Design details
Cited by
No citation data available for this registration yet.