Using Lotteries To Encourage Saving: Experimental Evidence from Kenya

Justin Abraham, Merve Akbas, Dan Ariely, and Chaning Jang

This is a public repository for "The Role of Regret in Prize-Linked Savings: Experimental Evidence from Kenya". We registered a pre-analysis plan with the AEA RCT Registry that outlines the research question, the experiment, the data, the identification strategy, and our hypotheses. The most recent version of the paper can be found here.

Replication

Releases typically contain the data, source code, survey instruments, and a summary of the research design. Simply run akiba_master.do to reproduce tables and figures presented in the manuscript and appendix. These will be output in tables/ and figures/, respectively. Experimental data was collected using Z-Tree. Results published in the manuscript were analyzed using Stata 13.1.

Directories

The scripts rely on this particular file structure.

• code: Contains source code for cleaning raw data and running analyses.
  • ado: Contains canned scripts used in data manipulation, analysis, and publishing.
  • do: Primary .do files, esttab tables, and .ado files required for execution.
  • php: Scripts for acquiring raw data.
  • ztree: .ztt files used in the lab study.
• data: Contains data in various formats.
  • clean: Cleaned and merged datasets in wide and long .dta format.
  • lab: Raw Z-Tree subject tables from the lab study.
  • ledger: Transaction data from SMS savings.
  • lottery: Data on lottery results.
  • pilot: Pilot study.
  • questionnaires: Endline questionnaire, Akiba post-survey, and enrollment data.
• documents: Contains surveys, lab protocols, applications, IRB, and notes.
• figures: Contains output graphs and figures.
• presentation: Contains slides for presentation.
• publication: Contains materials for the manuscript, analysis plan, and online appendix.
• tables: Contains output tables in .tex format.

Data

All individual-level data has been de-identified to protect respondent privacy. Access to personally identifiable data is restricted to authorized users. The primary datasets used for primary results are

• akiba_wide.dta: This is a cross-section of individual participants that includes baseline lab data, saving outcomes aggregated over the entire 60-day saving period, and data from the endline questionnaire.
• akiba_long.dta: This is an individual-by-period dataset containing daily outcomes as well as the time-invariant individual characteristics in akiba_wide.dta.

Contact

The corresponding author is Justin Abraham.