Estimating and explaining cross-country variation in the effectiveness of non-pharmaceutical interventions during COVID-19

Nicolas Banholzer, Stefan Feuerriegel, Werner VachScientific 2022 · DOI

Abstract

To control the COVID-19 pandemic, countries around the world have implemented non-pharmaceutical interventions (NPIs), such as school closures or stay-at-home orders. Previous work has estimated the effectiveness of NPIs, yet without examining variation in NPI effectiveness across countries. Based on data from the first epidemic wave of 40 countries, we estimate country-specific differences in the effectiveness of NPIs via a semi-mechanistic Bayesian hierarchical model. Our estimates reveal substantial variation between countries, indicating that NPIs have been more effective in some countries (e.g., Switzerland, New Zealand, and Iceland) as compared to others (e.g., Singapore, South Africa, and France). We then explain differences in the effectiveness of NPIs through 12 country characteristics (e.g., population age, urbanization, employment, etc.). A positive association with country-specific effectiveness of NPIs was found for government effectiveness, gross domestic product (GDP) per capita, population ages 65+, and health expenditures. Conversely, a negative association with effectiveness of NPIs was found for the share of informal employment, average household size and population density. Overall, the wealth and demographic structure of a country can explain variation in the effectiveness of NPIs.

Introduction

During the initial wave of the COVID-19 pandemic, governments worldwide deployed a suite of non-pharmaceutical interventions (NPIs)—including school closures, gathering bans, and stay-at-home orders—to curb viral transmission. While early research successfully estimated the average global effectiveness of these measures, it largely overlooked the significant regional differences in outcomes.

Treating NPI effectiveness as a universal constant ignores the local context. For instance, a stay-at-home order in a nation with robust digital infrastructure and government subsidies differs fundamentally from one in a country characterized by high population density and a large informal workforce. This study bridges this gap by quantifying country-specific variation and identifying the socioeconomic and demographic drivers that determine NPI success.

Methodology

The researchers employed a two-step analytical approach using a semi-mechanistic Bayesian hierarchical model.

Estimation of Variation

The model links unobserved daily new infections and contagious subjects to observed reported cases. By decomposing the NPI effect into a cross-country average and a country-specific deviation ($\theta_j$), the authors quantified how much more or less effective NPIs were in a specific nation compared to the global mean.

Explanation of Variation

To understand the drivers behind these deviations, 12 country characteristics were analyzed using two methods:

  1. Univariate Models: Testing the individual association of characteristics (e.g., GDP, population density) with NPI effectiveness.
  2. Latent Factor Model: Mapping the 12 characteristics onto two overarching factors: "Wealth" (comprising GDP, health spending, and government effectiveness) and "Demographics" (comprising density, household size, and age structure).

The study utilized epidemiological data from the Johns Hopkins Coronavirus Resource Center (February to May 2020) and NPI implementation data covering 8 specific interventions across 40 countries.

Results: National Variation in Effectiveness

The analysis revealed substantial heterogeneity in how effectively countries implemented NPIs. Nations such as Switzerland, New Zealand, and Iceland demonstrated the highest effectiveness, significantly exceeding the global average. Conversely, countries like Singapore, South Africa, and France showed lower-than-average effectiveness during the study period.

Figure 5: Estimated change in NPI effectiveness by country, showing significant variation from the cross-country average.

Drivers of NPI Success

The study identified several key socioeconomic and demographic predictors of NPI effectiveness:

  • The "Wealth" Advantage: Higher GDP per capita and government effectiveness were strongly associated with NPI success. Wealthier nations likely possess the fiscal capacity to provide social safety nets, such as furlough schemes, making compliance economically feasible for citizens. Government effectiveness saw a +2.3% relative increase in avoided infections per standard deviation (SD) increase.
  • The Demographic Hurdle: High population density and large average household sizes were the strongest predictors of reduced effectiveness. NPIs primarily target public interactions; they are less effective at mitigating "within-household" transmission, which dominates in crowded living conditions. Population density was associated with a -2.4% relative decrease in effectiveness per SD.
  • The Age Paradox: Countries with a higher proportion of elderly citizens (65+) saw higher NPI effectiveness. This is likely due to higher compliance among high-risk populations.
  • Informal Labor: A high share of informal employment was negatively associated with effectiveness (-2.8% per SD). Workers in the informal sector often lack labor protections and must continue working to survive, regardless of mandates.
Figure 6: Association of 12 country characteristics with NPI effectiveness, highlighting the positive impact of wealth and the negative impact of density and informal labor.

Discussion and Limitations

The latent factor model indicated that "Wealth" and "Demographics" explain approximately 65% of the observed cross-country variation. However, several limitations remain. The study identifies correlations rather than direct causation. Furthermore, the 40-country sample is predominantly European, which may limit the generalizability of the findings to regions with different governance or cultural strategies. Future work should investigate whether specific country characteristics favor certain types of interventions over others and account for voluntary behavioral changes that often precede official mandates.

Summary updated

Abstract

To control the COVID-19 pandemic, countries around the world have implemented non-pharmaceutical interventions (NPIs), such as school closures or stay-at-home orders. Previous work has estimated the effectiveness of NPIs, yet without examining variation in NPI effectiveness across countries. Based on data from the first epidemic wave of [Formula: see text] countries, we estimate country-specific differences in the effectiveness of NPIs via a semi-mechanistic Bayesian hierarchical model. Our estimates reveal substantial variation between countries, indicating that NPIs have been more effective in some countries (e. g. Switzerland, New Zealand, and Iceland) as compared to others (e. g. Singapore, South Africa, and France). We then explain differences in the effectiveness of NPIs through 12 country characteristics (e. g. population age, urbanization, employment, etc.). A positive association with country-specific effectiveness of NPIs was found for government effectiveness, gross domestic product (GDP) per capita, population ages 65+, and health expenditures. Conversely, a negative association with effectiveness of NPIs was found for the share of informal employment, average household size and population density. Overall, the wealth and demographic structure of a country can explain variation in the effectiveness of NPIs.
BibTeX
@inproceedings{banholzer2022estimating,
  author = {Nicolas Banholzer and Stefan Feuerriegel and Werner Vach},
  title = {{Estimating and explaining cross-country variation in the effectiveness of non-pharmaceutical interventions during COVID-19}},
  booktitle = {Scientific},
  year = {2022},
  doi = {10.1038/s41598-022-11362-x},
}
Figure