PDRI/MLP/USAID-DRG Policy Brief 2: Press Freedom and Reporting on Corruption

This policy brief was prepared by Jitender Swami and Erik Wibbels from the University of Pennsylvania and Diego Romero from Utah State University. It highlights findings from the Machine Learning for Peace (MLP) project, funded by USAID’s Center for Democracy, Human Rights, and Governance (DRG).

Key Takeaways

1. Legal vs. Regulatory Changes:
   • Legal changes targeting press freedom have limited immediate impact on corruption reporting.
   • Regulatory changes, such as restructuring media oversight bodies, result in significant and immediate declines in corruption reporting.
2. Timing of Regulatory Changes:
   • Regulatory changes are often implemented after increases in media attention on government corruption, suggesting they are reactive tools used by regimes.
3. Policy Implications:
   • Support for independent journalism should prioritize monitoring and mitigating regulatory changes over formal legal reforms.
   • Computational tools and non-journalistic methods (e.g., trade and budget analysis) offer alternative ways to detect and combat corruption in repressive environments.

Introduction

Press freedom plays a critical role in exposing corruption and holding governments accountable. However, authoritarian regimes often restrict media freedom, limiting journalists’ capacity to report on corruption. Using MLP’s unique high-frequency civic space data, this study explores how legal and regulatory restrictions on press freedom influence media coverage of corruption in 62 developing countries from 2012–2024.

The Challenge

Corruption thrives in environments with weak press freedom, but researchers have lacked rigorous evidence on how different restrictions affect media reporting. This study distinguishes between two key types of restrictions:

1. Legal changes (e.g., laws criminalizing dissent or restricting media operations).
2. Regulatory changes (e.g., restructuring of media oversight bodies or altering their leadership).

MLP data enables a detailed analysis of the effects of these changes on corruption coverage.

Approach

1. Data Collection:
   • 110+ million articles classified into civic space events using a fine-tuned large language model.
   • Analysis of press freedom declines based on the RSF Press Freedom Index.
2. Methodology:
   • Identification of 46 major declines in press freedom, attributing 15 to legal changes and 4 to regulatory changes.
   • Examination of corruption reporting before and after these events using statistical methods.

Findings

Legal Changes:

• Impact: Minimal short-term effect on corruption reporting.
• Explanation: Long legislative timelines allow media to adapt coverage strategies in advance.

Regulatory Changes:

• Impact: Significant and immediate declines in corruption coverage.
• Examples:
  • Cameroon: Appointment of a new head to the National Communications Council under the guise of combating “fake news.”
  • Hungary: Media authority restructuring to tighten government control.
• Timing: Often enacted during periods of increased media focus on corruption, suggesting reactive suppression tactics.

Policy Implications

1. Regulatory Frameworks:
   • Regulatory changes can quickly suppress corruption reporting. Policymakers should increase monitoring of media oversight structures in backsliding democracies.
2. Support for Journalism:
   • Develop targeted strategies to assist journalists and media outlets in circumventing regulatory restrictions.
3. Alternative Anti-Corruption Tools:
   • Encourage non-journalistic methods, such as computational analyses of trade and budget anomalies, to detect corruption without endangering journalists.
4. Data-Driven Monitoring:
   • High-frequency civic space data should be integrated into policy frameworks to track real-time media freedom trends.

Acknowledgments

This brief was made possible through support from USAID’s Center for Democracy, Human Rights, and Governance. The authors acknowledge the contributions of the MLP team and its research collaborators.