Expert adoption of composite indices: A randomized experiment on migrant resettlement decisions in Bangladesh
MetadataShow full item record
- Publications 
Original versionBergen: Chr. Michelsen Institute (CMI Working Paper WP 2022:03) 32 p.
In settings where complex social decisions are made, information is often aggregated into indices to facilitate decision making. The value added of such composite indices depend, inter alia, on the extent to which decision makers trust and make use of them. This paper presents a randomized experiment on the use of an index designed to inform migrant resettlement decisions, using 410 graduate students in Bangladesh as respondents. Respondents were randomly assigned to control and treatment conditions. In the control group, respondents faced a discrete choice experimental set-up where they were asked to allocate 1000 migrants between two locations described by five attributes (availability of cropland, distance to hospital, distance to school, poverty incidence, and frequency of floods, droughts and cyclones). In the treatment group, respondents also had access to the migrant resettlement index for the two locations, and we also had a second control group where an irrelevant attribute was included instead of the resettlement index. The results show that the resettlement index is used by the study participants, and mechanism analyses suggest that this is due to perceptions of improved benefits to costs from using the index to make decisions. Results from the control group also suggest that past adverse environmental events are particularly important for resettlement decisions. Use of the index in decision making does not depend much on respondent background characteristics, but the perceived importance of the five attributes in the control group does vary with background, sometimes in surprising ways. Notably, respondents who grew up in locations where land was scarce or floods, droughts and cyclones were frequent, placed less emphasis on these attributes in their migrant resettlement decisions.