Voluntary Disclosure of Cartel Information under Leniency Programs: Theory and Experiment

Description

This dataset contains the experimental data and replication materials for the study “Voluntary Disclosure of Cartel Information under Leniency Programs: Theory and Experiment.” The study examines how the design of leniency programs affects both the decision to report cartel activity and the amount of useful information voluntarily disclosed. The experiment compares three leniency programs. The OLD and NEW treatments model leniency systems adopted by the Japan Fair Trade Commission (JFTC): OLD represents the former Japanese system, in which fine reductions depend mainly on the order of application, while NEW represents the reformed system, in which the amount of information disclosed can also affect fine reductions for non-first applicants. REFINED is a hypothetical extension of the NEW system, REFINED does not currently exist in practice; under REFINED, information disclosure also affects the fine reduction available to the first applicant. The study therefore compares the past and current Japanese systems and then evaluates a possible modification intended to improve the current system. The experiment was conducted online using oTree between December 2021 and June 2023 with 186 undergraduate students: 66 in OLD, 63 in NEW, and 57 in REFINED. Participants were assigned to groups of three and repeatedly chose reporting strategies and an information-content level from 0 to 5. The main finding is that information disclosure is lowest under OLD, higher under NEW, and highest under REFINED. This suggests that making the amount of information provided more directly relevant to fine reductions can strengthen incentives to submit useful evidence. The analyses reported in the paper were conducted using Microsoft Excel and R. The replication package provides R code that reconstructs the analysis datasets and reproduces the reported tables, statistical tests, and underlying figure data; it also includes supporting Excel files used for the final presentation of selected figures.

Steps to reproduce

Download and extract the replication package, keeping Leniency_repliation_report_20260915.rmd, Leniency_replication_ALL_in_one20260915_with_sessionInfo.R, and Raw_data.zip in the same directory. Open Leniency_repliation_report_20260915.rmd in RStudio and Knit the document. The R Markdown file automatically runs the master R replication script. The script extracts the raw data, reconstructs the analysis datasets, performs validation checks, reproduces Tables 4–8 and Figures 2–7, runs the statistical analyses, and creates sessionInfo.txt. All generated analysis files are saved in the replication_output/ directory. The knitted report presents the results in the same order as in the manuscript. Required R packages are checked and installed automatically by the master script when necessary. For the optional wild-cluster bootstrap analysis, fwildclusterboot and dqrng are required.

Institutions

• Kansai University

  Osaka, Suita

Categories

Funders

• Japan Society for the Promotion of Science

  Ministry of Education, Culture, Sports, Science and Technology

  Tokyo, Tokyo
• The Joint Usage and Research Center, Research Institute for Socionetwork Strategies, Kansai University

  Grant ID: 2020001

Related Links

Licence