Impact of Personalised Information on the Efficiency of Vehicle Choices: Evidence from Nepal

How to cite

Srinivasan, Suchita; Filippini, Massimo; Kumar, Nilkanth. 2024. "Impact of Personalised Information on the Efficiency of Vehicle Choices: Evidence from Nepal." AEA RCT Registry. AEARCTR-0013486. https://doi.org/10.1257/rct.13486-1.0

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

Suchita Srinivasan* , Massimo Filippini , Nilkanth Kumar

* Primary investigator

Registry ID
AEARCTR-0013486
DOI
https://doi.org/10.1257/rct.13486-1.0
First registered
Last updated
Status
Completed
Trial dates
2023-02-01 – 2023-11-11
Countries
Nepal (Kathmandu Valley)
JEL codes
D1, D8, Q4, Q5
Registry record
http://www.socialscienceregistry.org/trials/13486

Abstract

Information on operating costs can enable potential buyers to purchase more fuel-efficient vehicles. We test this hypothesis in a developing country setting using personalised information provision in a randomised controlled trial: we recruit respondents looking to purchase a motorcycle in Kathmandu, Nepal, and provide them information on annual operating cost-savings or five-year operating cost-savings of different models using energy labels on a unique web-based platform. We find that both information treatments were effective in improving the fuel economy of the motorcycles actually purchased by respondents, and the motorcycle that they stated they would like to purchase, with a stronger effect for the five-year operating cost savings. Furthermore, being shown the five-year operating cost savings information also made respondents buy relatively more efficient motorcycles in reality, compared to the model they indicated they would buy in the stated choice setting. Our treatments were particularly effective with respondents of an impatient nature, or those who had low levels of energy-related financial literacy. This study provides novel evidence on the effectiveness of information provision in low-education settings, where the opportunity cost of collecting information may be high.

Intervention

In collaboration with a data collection company, we sampled for individuals looking to purchase a motorcycle with a high probability (at least 30%) in the next 6-7 months. Respondents were sampled at motorcycle showrooms, office and college parking lots, and at public transport connection points.

Potential motorcycle buyers are randomly allocated into one of three groups, the control group, Treatment group 1 or Treatment group 2. Respondents in all three groups access a web-based platform, where they can browse different motorcycle models, and select which motorcycle model they prefer. They can compare up to three motorcycles at a time. Respondents in the control group are shown energy labels (fuel economy labels) in which they are provided information on the fuel economy of the selected motorcycles (in kilometres per litre). Treatment group 1 respondents are shown the annual operating cost savings/expenses from using each motorcycle model, relative to the average vehicle of the same engine size, whereas Treatment group 2 are shown the five-year operating cost savings/expenses.

All respondents are provided information on the make, model, engine size, price as well as other non-efficiency related technical specifications of the selected motorcycles. The fuel economy labels are similar in design to those used in India for cars, and we tested the design and color scheme of the label during a pre-test.

Experimental design

Phase 0 of the study was the recruitment phase, in which respondents were approached and invited to participate in our study. We also collected some basic socio-economic information, on the basis of which they were randomly allocated. Respondents are randomly allocated into one of three groups, control group (which sees the fuel economy of the motorcycles), Treatment group 1 (which sees the annual operating cost savings/expenses of each motorcycle model), and Treatment group 2 (which sees the five-year operating cost savings/expenses of each model).

In Phase 1, respondents were sent a text message with a link to access the web-based platform. On the platform, they could view information on different motocycle models. This intervention involved showing information in the form of fuel economy labels, as described above. Respondents can use the website several times if they wish. Respondents were then sent another text message, with a link to the baseline survey, in which we collected more detailed socioeconomic information.

Phase 2 was the reminder or notification phase, in which respondents were reminded to keep using the website in case they were still looking to purchase a motorcycle.

Lastly, in Phase 3, we conducted a short follow-up survey in which we asked the respondents about which motorcycle models they purchased/intend to purchase, or why they haven't already purchased one.

Design details

Randomization method
The data collection company shared basic socioeconomic information with our research team (including gender, age, income, occupation, probability of purchasing a motorcycle in the next 3-6 months, and educational attainment) and we did the randomisation using this information, and employing a re-randomisation procedure.
Randomization unit
The unit of randomisation is the individual in our study.
Clusters
3300 individuals
Observations
3300 individuals
Arms
1100 control respondents, 1100 in treatment group 1, 1100 in treatment group 2
Primary outcomes
We are interested in both the revealed preference as well as in the stated preference outcomes in this study. In the revealed preference outcome, we evaluate the impact of the treatments mainly on the logarithm of the fuel economy of the motorcycle model actually bought by respondents (measured in kilometres per litre). In the stated preference analysis, which we conduct on the platform, we are interested in two main variables, the logarithm of the fuel economy of the selected motorcycle (in kilometres per litre), as well as on the likelihood of selecting the most efficient model in the choice set.
Secondary outcomes
In the revealed preference analysis, we also evaluate the impact of our treatments on the total annual operating costs associated with the motorcycle models purchased, as well as on the total lifetime cost (which is proxied by the purchase price plus the discounted sum of total operating costs).

Cited by

No citation data available for this registration yet.

Keywords