Stronger pre-pandemic institutions linked to lower COVID-19 deaths and faster vaccination

Countries that entered the COVID-19 pandemic with stronger institutions recorded substantially fewer deaths and vaccinated their populations faster, according to a cross-country study published in Globalization and Health. Diego A. Diaz, Cristian Larroulet and Andres Marticorena of the Universidad del Desarrollo combined 2019 World Bank Worldwide Governance Indicators with pandemic outcomes measured through the end of 2021, and found that in the fullest statistical models a one-unit increase in the composite governance index was associated with 384 fewer reported COVID-19 deaths per million, 1,860 fewer excess deaths per million, and 15.5 percentage points higher full vaccination coverage by the end of 2021. The associations held up across a battery of robustness checks, though the authors stress the findings are observational and should not be read as definitive causal estimates.

The study addresses a puzzle that has run through much of the pandemic literature: reported COVID-19 death counts are shaped not only by how deadly the virus was in a country but also by how completely deaths were counted. The researchers compared official reported mortality with excess mortality, defined as the difference between observed all-cause deaths and expected deaths based on historical trends, which is widely regarded as a more reliable indicator of pandemic impact. In their data, the unconditional correlation between the governance composite and reported deaths per million was actually positive, at 0.29, because richer countries with stronger institutions both report deaths more accurately and have older populations more vulnerable to the virus. The correlation with excess mortality, by contrast, was negative at minus 0.43. Once the models controlled for income and demographic structure, the reported-deaths coefficient flipped sign, a textbook instance of Simpson’s paradox that underscores why confounders matter in cross-country comparisons.

The governance measure itself was built from the six Worldwide Governance Indicators for 2019: Voice and Accountability, Political Stability, Government Effectiveness, Regulatory Quality, Rule of Law, and Control of Corruption. The authors averaged the six dimensions, each measured on roughly the same minus 2.5 to 2.5 scale, into a single composite. Diagnostics supported that aggregation: the six dimensions showed high internal consistency with a Cronbach’s alpha of 0.970, and the first principal component explained 87.0 percent of their variance and correlated almost perfectly with the simple average. Using 2019 values ensured the institutional measure was predetermined with respect to pandemic outcomes, and the composite in the sample ranged from minus 2.03 to 1.88 with a standard deviation of roughly 0.93.

The empirical design was a specification ladder of cross-sectional OLS regressions, beginning with bivariate associations and progressively adding demographic controls such as log GDP per capita, the share of population aged 65 and above, population density and urbanization; health-infrastructure controls including health expenditure, hospital beds and physicians per 1,000 people; the mean 2020 stringency of containment policy from the Oxford COVID-19 Government Response Tracker; diabetes prevalence as a comorbidity proxy; and finally continent fixed effects. Standard errors were heteroskedasticity-robust. In the full reported-deaths specification, a one-standard-deviation increase in institutional quality corresponded to about 360 fewer deaths per million, meaningful against a sample mean of roughly 931 deaths per million. The excess-mortality equivalent was about 1,730 fewer excess deaths per million per standard deviation. The share of people aged 65 and above was consistently positive and significant in the mortality models, as expected given the steep age gradient in COVID-19 fatality risk.

Sample attrition was substantial and matters for interpretation. The reported-deaths ladder shrank from 210 countries in the simplest model to 111 in the full specification, while the excess-mortality ladder fell from 121 to 76. Countries retained in the fuller models were on average richer, older and institutionally stronger than those excluded; the mean governance composite among countries in the full excess-mortality sample was 0.48, compared with minus 0.18 among excluded countries. The authors therefore caution that the excess-mortality estimates speak most directly to relatively well-governed, better-resourced countries with stronger mortality-data systems, rather than to the full set of countries where pandemic-preparedness needs may be greatest. Re-estimating the models on common full-model samples preserved the main patterns, however, suggesting that changing country composition does not fully explain the results.

Vaccination emerged as one of the strongest and most consistent associations in the study. In the full specification, a one-unit increase in the governance composite was associated with 15.5 additional percentage points of the population fully vaccinated by end-2021, or about 14 points per standard deviation, operating above and beyond income, health infrastructure and policy stringency. Log GDP per capita also remained positive and significant, indicating that resources mattered but that institutional quality was not simply a proxy for income. To probe whether vaccination could be a channel linking governance to lower mortality, the researchers ran an exploratory, temporally ordered mediation analysis in which the mediator was the share fully vaccinated by September 30, 2021 and the outcome was deaths per million accumulated between October 1 and December 31, 2021, so the mediator was measured before the mortality window. The estimated indirect effect was minus 46.7 deaths per million, with a bias-corrected bootstrap confidence interval of minus 95.4 to minus 17.9, representing about 35 percent of the total governance association with late-2021 mortality. That leaves most of the association unexplained by vaccination alone, with other institutional channels such as public trust, policy compliance, health-system coordination and mortality reporting likely contributing.

The economic results were more phase-specific. Institutional quality showed little robust association with GDP growth in 2020, the year of the initial global contraction, across fuller specifications, and adding the Oxford tracker’s Economic Support Index did not materially change the coefficient. The authors interpret 2020 GDP growth as reflecting a mixture of epidemiological timing, containment severity, sectoral exposure and policy responses rather than a simple governance gradient. For 2021, however, the picture differed: in the full specification the governance coefficient was positive at 2.494 and statistically significant, indicating that countries with stronger institutions tended to recover faster after conditioning on the same controls. When economic support was added, the coefficient remained positive but became marginally significant at 2.011, while economic support itself was positively associated with 2021 growth. The researchers read this as a conditional association rather than evidence of a single mechanism, since fiscal support, sectoral composition, external demand and vaccination could all contribute to recovery.

Robustness checks tested the findings from several angles. The mortality and vaccination results were essentially unchanged when the simple-average composite was replaced with the first principal component of the six governance dimensions, and when the V-Dem 2019 electoral democracy index was added as a control, though the mortality magnitudes shifted in opposite directions, with the reported-deaths coefficient moving from minus 384 to minus 601 and the excess-mortality coefficient attenuating from minus 1,860 to minus 1,365. The authors attribute this to the strong empirical overlap between electoral democracy and governance quality, noting the two indices correlated at 0.749 and that the democracy index correlated at 0.939 with Voice and Accountability. Timing checks showed the association was present for deaths accumulated by September 2021, deaths accrued in the fourth quarter of 2021, and vaccination measured at both cutoffs. Influence diagnostics flagged Singapore as the most influential observation in the excess-mortality model, with a Cook’s D of 3.717 and leverage of 0.954, followed by Georgia and the United States; the leave-one-out range for that coefficient ran from minus 1,999 to minus 1,650, so no single omitted country reversed the sign, but influential observations affected the estimated size and uncertainty. Quantile regression showed the protective association concentrated in the upper tail of the mortality distribution, significant at the 75th percentile but imprecisely estimated at the 25th and 50th. Legal-origin instrumental-variable estimates, presented only as supplementary sensitivity analysis, produced larger effects than OLS, but the Sargan overidentification test rejected at p equal to 0.018, reinforcing the authors’ view that the exclusion restriction is contestable and that the multivariable OLS patterns deserve greater weight.

Disaggregating the six governance dimensions one at a time, Government Effectiveness showed the largest negative association with COVID-19 mortality, while Rule of Law and Control of Corruption were also negative and statistically significant; Regulatory Quality, Voice and Accountability and Political Stability were imprecisely estimated. Because the six dimensions are highly correlated, the authors interpret these coefficients as descriptive comparisons rather than sharply identified rankings. They also used Chile as a descriptive case study illustrating how the mechanisms co-moved within Latin America. Chile scored substantially higher than large regional peers on all six governance dimensions in the 2019 data, began mass vaccination in February 2021, reached 50 percent full vaccination by July 2021 and over 84 percent by end-2021, the highest rate among the peers examined. Its economy contracted 6.1 percent in 2020 and rebounded 11.3 percent in 2021, reaching the highest 2021 GDP-per-capita-PPP index among the peers shown, at 103.1 relative to 2019. The authors note that this recovery also reflected unusually large macroeconomic support, including fiscal transfers and pension withdrawals, and that aggressive stimulus later contributed to overheating and inflation, a reminder that stronger institutions may facilitate crisis response while the specific policy choices made within that framework remain consequential.

The authors conclude that pre-pandemic institutional quality was consistently associated with pandemic health outcomes and vaccine deployment, while economic associations were more phase-specific. They frame the results as consistent with the view that strengthening government effectiveness, regulatory capacity, rule of law and anti-corruption mechanisms should be understood as pandemic preparedness measures, since capabilities such as cold chains, health information systems, procurement capacity and public trust are difficult to build during a crisis. The limitations they acknowledge are considerable: cross-country regressions cannot establish causality, the composite compresses heterogeneous governance dimensions, the cross-sectional design cannot capture pandemic dynamics fully, attrition limits external validity, excess mortality is available for fewer countries and is sensitive to high-influence observations, and the pandemic was a unique event that warrants caution in extrapolating to future health crises. Future work, they suggest, should combine richer within-country designs with better mortality coverage and explicitly test the roles of fiscal buffers and international economic openness to determine which institutional capabilities matter most when acute health shocks arrive.

Subject of Research: Cross-country associations between pre-pandemic institutional quality and COVID-19 mortality, vaccination rollout and economic outcomes

Article Title: Institutional quality and pandemic resilience: cross-country evidence on health and economic outcomes during COVID-19

Article References: Diaz, D. A., Larroulet, C., & Marticorena, A. (2026). Institutional quality and pandemic resilience: cross-country evidence on health and economic outcomes during COVID-19. Globalization and Health, 22(1), Article 71. https://doi.org/10.1186/s12992-026-01230-y

Image Credits: AI Generated

DOI: 10.1186/s12992-026-01230-y

Keywords: institutional quality, COVID-19, excess mortality, vaccination rollout, Worldwide Governance Indicators, economic recovery, cross-country regression, pandemic preparedness, mediation analysis, Chile, Globalization and Health, observational study