The effect of COVID certificates on vaccine uptake, health outcomes, and the economy

Governments implemented unprecedented policies during COVID-19, initially aiming either to eliminate the virus or slow its spread until vaccines became available. Non-pharmaceutical interventions (NPIs) such as closures and restrictions were informed by models and causal analyses. COVID certificates (proof of vaccination, recovery, or recent negative test) enabled targeted interventions but raised ethical and political concerns. As vaccine supply increased, hesitancy became the main barrier to coverage. Policy tools to increase uptake include communication, incentives, and mandates; COVID certificates emerged as a new tool possibly spurring vaccination. In Europe, EU travel certificates were agreed in June 2021, and France, Germany, and Italy soon extended certificates to domestic activities. The study focuses on these three comparable countries to measure how much widespread use of COVID certificates incentivised vaccine uptake, reduced adverse health outcomes, and strengthened the economy, thereby informing decisions on whether, when, and how to employ certificates.

The paper situates its contribution within prior work on NPIs’ effectiveness and modelling of policy impacts, and emerging analyses of COVID certificates’ effects on vaccination. It references ethical and behavioural considerations around certification, and notes survey-based and econometric estimates of mandates/certificates on uptake. Specifically, contemporaneous studies using synthetic control and time series found short-run increases in vaccination following certificate announcements in several countries, with larger effects among younger cohorts. The authors also build on innovation diffusion theory (Bass model) previously used for medical innovation and vaccine adoption, and on macroeconomic nowcasting literature using high-frequency indicators (OECD Weekly Tracker) to assess the impact of vaccination on economic activity.

Vaccine uptake counterfactuals: The principal approach uses innovation diffusion (Bass) theory to model cumulative first-dose uptake x(t) with parameters p (innovation), q (imitation), and K (capacity). Parameters are estimated via nonlinear least squares using data from the 100 days prior to each country’s certificate announcement and extrapolated to year-end 2021. Block bootstrapping (1000 iterations, 30 blocks) yields 95% confidence intervals. Counterfactual first-dose uptake without certificates is the fitted trajectory from the announcement date onward; realized data are used pre-announcement. Second-dose counterfactuals assume the realized ratio of second-to-first doses with a 3-week lag applies equally to the counterfactual. Booster doses are excluded as newly vaccinated post-announcement were not booster-eligible in 2021. Age-stratified uptake is modelled for 60+ in France and Italy using the same framework with a longer fitting window (start 2021 to announcement). Younger-group counterfactuals are inferred by difference. Robustness via synthetic control: For each treated country, a synthetic control is constructed from a donor pool of OECD/EU countries without day-to-day certificate announcements before 22 September 2021, using weights chosen to minimize pre-treatment error and including covariates predictive of vaccination (GDP per capita, pre-period cases/deaths, share 65+, mobility, vaccine acceptance). Placebo tests assess significance via RMSPE ratios. Synthetic trajectories are compared to the Bass counterfactuals through end-September 2021. Health outcomes modelling: Vaccine effectiveness against severe outcomes (Delta period) is set conservatively at 81% after one dose and 92% after two doses, aggregated across vaccine types and waning. Realized and counterfactual health outcomes (hospital admissions, ICU patients, deaths) by vaccination status are inferred using Bayes’ rule with vaccination shares V_t(v) and effectiveness by status, applying lags for vaccine protection and disease progression: l_vaccine=7 days; l_hosp=7; l_ICU=10; ICU stay total 8 days with midpoint at 4 days; l_death=14. Total lags: hospital admissions 14 days; ICU patients 21 days; deaths 21 days. Age-stratified modelling is applied where data permit (France: hospital admissions, deaths; Italy: deaths). Differences between realized and counterfactual cumulative outcomes from announcement to 31 Dec 2021 are attributed to certificates through their effect on uptake. Economic impact: A two-way fixed-effects panel regression over 46 OECD/G20 countries estimates the elasticity of weekly GDP (OECD Weekly Tracker, relative to pre-crisis trend) to vaccination: T_it = β V_it(lag 4 weeks) + γ I_it(lagged cases, deaths, R_t, mobility, stringency) + η X_it(trading partners’ vaccination, deaths, GDP, cases) + Z_it(temperature) + country FE + week FE + ε_it. β is estimated accounting for uncertainty in the Weekly Tracker via multiple (300) bootstrap series and Rubin’s rules. Robustness checks examine alternative lags (1–10 weeks), time-varying effects via Double Machine Learning (R-learner with overlapping period dummies), and alternative dependent variables (Google Mobility Index; official quarterly GDP). Country-specific counterfactual weekly GDP paths are computed by plugging uptake counterfactuals into the regression to quantify prevented GDP losses in H2 2021.

Vaccine uptake: Announcements of COVID certificates were associated with increases in first-dose uptake by end-2021 of 13.0 p.p. (95% CI 9.7–14.9) in France, 6.2 p.p. (2.6–6.9) in Germany, and 9.7 p.p. (5.4–12.3) in Italy. Counterfactual first-dose shares without certificates: 65.2% (France), 67.3% (Germany), 70.4% (Italy) versus realized 78.2%, 73.5%, 80.1%. Effects grew with subsequent extensions (e.g., workplace rules; booster integration) and were larger among younger cohorts; among 60+, additional uptake: France 8.9 p.p. (8.0–9.4), Italy 4.4 p.p. (2.9–5.2). Synthetic control analyses up to late September 2021 support the counterfactual trajectories. Health outcomes: From announcement to 31 Dec 2021, additional hospital admissions averted were estimated at 32,065 (26,566–35,306) in France, 5,229 (−1,774–6,822) in Germany, and 8,735 (2,999–12,261) in Italy. Additional deaths averted: France 3,979 (3,453–4,298); Germany 1,133 (−312 to 1,358); Italy 1,331 (502–1,794). Relative to realized totals, counterfactual increases would have been 31.3% (hospital) and 31.7% (deaths) higher in France, 5.0% and 5.6% higher in Germany, and 15.5% and 14.0% higher in Italy. By the last week of 2021, without the uptake increase, hospital admissions (deaths) would have been approximately 46% (49%) higher in France, 14% (11%) higher in Germany, and 29% (26%) higher in Italy. ICU pressure and lockdown benchmarks: In France, ICU patients per million at end-2021 were 52.4 observed; counterfactual estimated 76.1 (72.4–78.3), a 45% increase, near levels at which prior lockdowns were instated, suggesting certificates may have helped avert renewed lockdowns. In Germany, added uptake was insufficient to avoid high ICU pressure; in Italy, ICU pressure remained low regardless. Economic impact: Panel estimates indicate a 1 p.p. increase in vaccinated share raises weekly GDP (after one month) by 0.052 p.p. (95% CI 0.033–0.070). Applying country-specific uptake uplifts, weekly GDP by end-2021 would have been lower by 0.6% (0.5–0.9) in France, 0.3% (0.1–0.4) in Germany, and 0.5% (0.3–0.7) in Italy absent certificates, implying prevented GDP losses across H2 2021 of €6.0 (5.9–6.1) billion in France, €1.4 (1.3–1.5) billion in Germany, and €2.1 (2.0–2.2) billion in Italy. A complete vaccination rollout would be associated with a 5.2 p.p. GDP increase relative to pre-crisis trend (about 85% of 2020 loss).

The study finds that COVID certificates were associated with substantial increases in vaccination, improved health outcomes, and higher economic activity in France, Italy, and to a lesser extent Germany (significant mainly from late November 2021). Conservative assumptions on vaccine effectiveness and epidemiological lags were used. While the causal pathway is inferred through the estimated uptake response, direct causality cannot be established conclusively, and the analysis does not model how certificates may have changed epidemic dynamics or other policy choices (e.g., behaviour, natural immunity). Cross-country differences likely reflect announcement timing, policy scope, and governance: centralized, nationwide, and clearly communicated implementations in France and Italy versus gradual, heterogeneous adoption in Germany. Certificates may serve as an inclusive alternative to mandates, emphasizing the benefits of vaccination/testing. Policymakers should consider supply of vaccines/tests, political trust, accessibility for marginalized groups, and international coordination for mutual recognition to avoid inequities. Further research should assess broader epidemic effects and long-term social and economic costs/benefits as more data accrue.

COVID certificates were associated with meaningful increases in vaccine uptake that translated into fewer hospital admissions and deaths, reduced ICU pressure (notably in France, potentially avoiding lockdown-level strain), and improved economic activity in the second half of 2021. The paper contributes a counterfactual approach using innovation diffusion theory corroborated by synthetic control, and links vaccination to macroeconomic activity via high-frequency GDP proxies. Future work should strengthen causal identification, quantify effects on transmission and policy responses, examine heterogeneity across institutional contexts and timing, and evaluate longer-term social and economic impacts, including equity considerations.

- Causality: The design associates certificates with increased uptake but cannot directly infer causality; unobserved shocks may remain despite robustness checks. - Model assumptions: Innovation diffusion (Bass) model assumes constant parameters and excludes major exogenous shocks; validity may wane with Omicron emergence in 2022. No explicit modelling of supply/eligibility constraints beyond fitting window choices. - Data limitations: Age-stratified vaccination data unavailable for Germany pre-mid-September; some health outcomes not age-stratified outside France/Italy. Booster effects excluded for 2021 period. - Epidemiological scope: Health impact estimates focus on direct protection; reductions in transmission, behavioural changes, and evolving natural immunity are not modelled. - Synthetic control constraints: Donor pool shrinks as more countries adopt certificates, limiting analysis to late September 2021; potential cross-border policy spillovers challenge the no-interference assumption. - Economic measurement: Weekly GDP proxied by OECD Weekly Tracker (machine-learning-based nowcast) with measurement uncertainty; estimates rely on average elasticities across countries and assume a static, time-invariant effect (though robustness checks suggest stability). - Germany timing: Later, more heterogeneous implementation may confound effect size and significance timing.