Abstract
We have developed and made accessible for a multidisciplinary audience a unique global dataset of the behavior of political actors during the COVID-19 pandemic as measured by their policy-making efforts to protect their publics. The dataset presents consistently coded cross-national data at subnational and national levels on the daily level of stringency of public health policies by level of government overall and within specific policy categories, and reports branches of government that adopted these policies. The data on these public mandates of protective behaviors is collected from media announcements and government publications. The dataset allows comparisons of governments' policy efforts and timing across the world and can serve as a source of information on policy determinants of pandemic outcomes—both societal and possibly medical.
Introduction
The global response to the COVID-19 pandemic was characterized by a rapid proliferation of public health mandates. While existing datasets successfully tracked the "what"—the specific nature of lockdowns, mask mandates, and travel restrictions—a critical gap remained in documenting the "who" and "how." Understanding pandemic outcomes requires more than just a list of policies; it requires an understanding of the political processes and institutional origins behind them.
The Protective Policy Index (PPI) addresses this gap by providing a granular, consistently coded record of policy-making behavior across 81 countries. By documenting which branches of government (executive, legislative, judicial, or bureaucracy) enacted specific rules at both national and subnational levels, the PPI enables researchers to analyze how political structures and institutional lead-actors influenced the trajectory of the pandemic response.
Methodology and Policy Categorization
The PPI was constructed using a multi-stage workflow to collect and verify data from media announcements and government publications. Policies are categorized into 15 distinct dimensions, ranging from international border closures to mandatory PPE usage.
To account for the evolving scientific understanding of the virus, the authors implemented two distinct weighting methods. Method 1 reflects the early-2020 perspective, placing higher emphasis on border closures. Method 2 reflects the later consensus, prioritizing social distancing and mask mandates.
All indices are normalized on a scale from 0 to 1 to facilitate cross-national comparison. For federal systems, country-level aggregates are calculated by weighting subnational data by population share, ensuring the data reflects the "average experience" of a country's citizens rather than just central government decrees.
Dataset Structure and Accessibility
The PPI Global Dataset is publicly hosted on GitHub and openICPSR. It provides high-resolution subnational data at the ISO 3166-2 level, which is essential for capturing the fragmented nature of governance in federal systems like the United States or Germany. The dataset includes:
- Country-level aggregates for both weighting methods.
- Regional data providing subnational insights.
- Policy "episodes" documenting every individual change and the specific branch of government responsible.
Validation and Key Results
The robustness of the PPI was validated through comparisons with Item Response Theory (IRT)-based indices and the Oxford COVID-19 Government Response Tracker (OxCGRT).
As shown in the validation analysis, the PPI correlates extremely strongly with IRT-based indices (0.983 for Method 1), suggesting that the manual weighting system is statistically sound. While it also correlates positively with the Oxford Stringency Index (0.85–0.89), the PPI captures more variation in specific snapshots, offering a more nuanced view of policy shifts. Preliminary epidemiological analysis suggests that a 10-percentage-point increase in the Method 1 PPI was associated with approximately an 8% reduction in the virus spread rate in the US during 2020.
Institutional Insights and Limitations
The PPI measures political effort—the willingness of incumbents to enact restrictive measures—rather than the effectiveness of implementation or public compliance. By tracking the "institutional origins" of a response, the dataset can distinguish between executive-driven responses and those negotiated through legislatures.
However, users should note the "implementation gap." A high PPI score indicates stringent mandates, but does not account for public enforcement or compliance. Future research should pair PPI data with mobility data to explore this gap. While the current dataset covers 81 countries and focuses on the 2020 mitigation phase, it provides a scalable template for documenting political behavior in future global crises.
Summary updated