Financial incentives for vaccination do not have negative unintended consequences

The study investigates whether offering financial incentives to encourage COVID-19 vaccination leads to negative unintended consequences, such as reduced future vaccination uptake, erosion of prosocial motivations and civic responsibility, diminished trust in healthcare providers, or increased perceptions of risk and coercion. Despite evidence that incentives can spur initial behaviour change, a broad literature cautions that such incentives may crowd out intrinsic motivation, signal riskiness, or undermine ethical norms. Measuring these unintended effects is challenging due to non-randomized policy rollouts and the need for rich behavioural and attitudinal data. The authors address these gaps using randomized exposure to incentives, administrative vaccination records, and comprehensive surveys to causally assess impacts in Sweden, and a complementary information experiment in the USA. The work is important for informing policy decisions on using financial incentives for health and prosocial behaviours amid ongoing vaccination efforts.

Prior research shows incentives can increase healthy behaviours, including blood donation, cancer screening, smoking cessation, and vaccination. However, theoretical and empirical work has warned about potential downsides: crowding out of intrinsic and prosocial motivation, signaling of risk or unpleasantness that may lower trust and perceived safety, changes in ethical attitudes toward incentives, and feelings of coercion or reduced self-determination. Influential contributions include philosophical critiques (Sandel; Satz), motivation crowding theory (Frey; Bowles; Bénabou & Tirole), and evidence that sanctions or controls can harm altruism or intrinsic motivation (Fehr & Rockenbach; Falk & Kosfeld; Gneezy & Rustichini). Policy bodies (WHO; ECDC) have advised caution regarding incentives in contexts like blood donation. During COVID-19, U.S. states implemented varied incentive programmes (cash/gift cards and lotteries), spurring debate with mixed evidence on effectiveness. Overall, concerns persist that incentives may produce unintended harms that could offset initial benefits, motivating the present causal assessment.

Sweden: The authors leverage a prior randomized controlled trial (May–July 2021) in which 1,131 participants were offered SEK 200 (~US$24) to receive a first COVID-19 vaccine dose within 30 days (incentives condition) and 3,888 participants were not offered payment (control). First-dose uptake increased by 4 percentage points at 30 days in the original RCT. For unintended consequences, they link all 5,019 RCT participants to national administrative vaccination records (second-dose outcomes) in December 2021 and conduct two pre-registered surveys: January 2022 (attitudes, intentions, other health behaviours) and June 2022 (third-dose uptake). Survey response: first survey n=3,238 (726 incentives; 2,512 control), second survey n=2,706 (606 incentives; 2,100 control), with balanced participation and no differential attrition on key characteristics. Outcomes include second- and third-dose uptake and timing; intentions for third dose (including hypothetical payments of SEK 100/500); flu shot uptake and intentions; blood donation; indices of morals/civic responsibility, perceived safety and efficacy (including child vaccine safety), trust in vaccination provision (researchers, public health agency, pharmaceutical companies); attitudes toward ethical acceptability of payment; and feelings of self-determination, coercion, and regret. Some survey measures were incentivized (implemented for a subset of participants) to capture consequential choices. Analytic approach: per pre-registration, outcomes are standardized. Ordinary least squares regressions with heteroscedasticity-robust standard errors estimate effects of the incentives condition relative to control, using pre-registered controls (gender, age, region, age×region interactions, at-risk status, civil status, children in household, employment, education, parents' birthplace, income). Two-sided t-tests report P-values. Power: 80% to detect effects as small as -0.12 SD at 5% significance. Equivalence tests assess whether effects more negative than -0.2 SD (Cohen’s d) can be rejected. Robustness checks include item-level analyses, alternative control sets, and secondary outcomes. A complementary Swedish vignette experiment (n=1,001) tests whether reactions differ if incentives are attributed to researchers vs the Public Health Agency of Sweden; analyses mirror the main approach with basic controls (gender, age, education, income). USA: A pre-registered information experiment (June–July 2022) with a general population sample from 12 states that implemented vaccine incentive programmes (n=3,062). Many participants (62.3%) were unaware of state programmes. Participants were randomized to receive detailed information about their state's programme (incentives condition) or no information (control). Outcomes were measured about five days later in an apparently unrelated follow-up survey to reduce demand effects. Outcomes include willingness to receive a further dose within 6 months, in case of a new outbreak, and if offered $20; intentions for next-season flu shot and blood donation; morals/civic responsibility; perceived safety/efficacy; and trust in state government. Analyses use OLS with heteroscedasticity-robust SEs and controls (gender, age, education, employment, income, 2021 state of residence). Equivalence testing again uses -0.2 SD as the smallest effect size of interest. Heterogeneity analyses probe vaccine hesitancy, political attitudes, sociodemographics, and state characteristics.

Sweden: No evidence of negative unintended consequences from offering SEK 200 for first-dose uptake. Administrative outcomes: second-dose uptake not reduced (B=0.055, SE=0.033, P=0.097) and no delay for second dose (B=0.046, SE=0.033, P=0.164); among first-dose recipients, second-dose uptake unaffected (B=0.049, SE=0.035, P=0.158). Survey outcomes: third-dose intention unaffected (B=-0.026, SE=0.044, P=0.560); intention if offered SEK 100 (B=0.001, SE=0.043, P=0.983) or SEK 500 (B=-0.008, SE=0.043, P=0.850) unchanged. Actual third-dose uptake unaffected (B=-0.007, SE=0.046, P=0.879); no change in time to third dose (B=0.030, SE=0.046, P=0.524); among second-dose recipients, third-dose uptake unaffected (B=-0.016, SE=0.049, P=0.745). Other health behaviours: flu shot uptake (B=0.001, SE=0.042, P=0.982), flu shot intention (B=0.012, SE=0.044, P=0.792), and blood donation (B=-0.021, SE=0.043, P=0.619) unaffected. Morals/civic responsibility index unchanged (B=0.001, SE=0.043, P=0.979); donations to NGOs promoting vaccination unaffected (B=-0.018, SE=0.045, P=0.679; and B=0.069, SE=0.044, P=0.117). Perceived safety/efficacy index (B=-0.030, SE=0.044, P=0.499), belief that vaccines are generally safe for children (B=-0.028, SE=0.044, P=0.528), and trust in vaccination provision index (B=0.004, SE=0.044, P=0.928) unchanged. Other concerns: ethical acceptability of payments (B=0.025, SE=0.044, P=0.560), regret (B=-0.040, SE=0.043, P=0.352), and feelings of coercion (B=-0.011, SE=0.043, P=0.797) unchanged. Across 21 coefficients, none of the 95% CIs crossed the -0.2 SD small-effect threshold; equivalence tests strongly reject effects more negative than -0.2 SD (largest P=0.0001). Complementary Swedish study: reactions did not differ when incentives were attributed to researchers vs the Public Health Agency; equivalence tests again reject even small negative effects. USA: Informing residents about state incentive programmes produced no negative consequences. Willingness to receive a further dose within 6 months (B=0.039, SE=0.036, P=0.276), during a new outbreak (B=0.062, SE=0.035, P=0.077), or if offered $20 (B=0.041, SE=0.036, P=0.256) did not decline; flu shot intention (B=-0.022, SE=0.035, P=0.528) and blood donation intention (B=0.027, SE=0.036, P=0.454) unchanged. Morals/civic responsibility (B=0.066, SE=0.035, P=0.060), perceived safety/efficacy (B=0.027, SE=0.035, P=0.448), and trust in state government (B=-0.027, SE=0.036, P=0.450) showed no negative impacts. No 95% CIs crossed the -0.2 SD threshold; equivalence tests reject effects more negative than -0.2 SD across outcomes. Heterogeneity analyses in both countries found no consistent negative effects across vaccine hesitancy levels, sociodemographics, or states.

The findings directly address concerns that financial incentives for vaccination may erode prosocial motivation, reduce trust and perceived safety, or depress subsequent health behaviours. Using randomized exposure to incentives in Sweden and randomized information about state programmes in the USA, combined with administrative records and rich surveys, the study finds no evidence of such harms. Equivalence tests further indicate that even small negative effects (Cohen’s d ≤ -0.2) can be ruled out across the wide range of behavioural and attitudinal outcomes. These results suggest that modest financial incentives can be employed without risking downstream reductions in vaccination adherence, other health behaviours, or civic and ethical attitudes. The lack of detectable negative impacts across subgroups (including vaccine-hesitant and sociodemographic strata) strengthens the generalizability within similar contexts. Although positive unintended consequences were also possible in theory, consistent improvements were not observed, indicating that incentives primarily affect initial uptake without broader attitudinal spillovers. The work informs policy by alleviating key ethical and practical concerns about using incentives to promote health behaviours.

Offering modest financial incentives for COVID-19 vaccination did not produce negative unintended consequences in Sweden or the USA. Across administrative and survey outcomes—including future vaccination uptake, timing, intentions, other health behaviours, morals and civic responsibility, perceived safety and efficacy, trust, regret, and coercion—effects were statistically indistinguishable from zero and equivalence tests rejected even small negative effects. The study contributes causal evidence to a long-standing debate and provides guidance to policy-makers considering incentives to boost healthy and prosocial behaviours. Future research should test different incentive magnitudes and contexts (e.g., organ donation, cancer screening), examine settings beyond high-income Western countries, and further explore long-term dynamics and ethical considerations.

• Generalizability: Samples come from high-income Western countries; findings may not extend to low-income settings or those with limited social safety nets. - Study awareness: Participants knew they were in a study, which could influence responses, though necessary for linking surveys to administrative data. - Context specificity: Focus on COVID-19 vaccination incentives; results may differ in other domains (e.g., organ donation, cancer screening) or with different incentive sizes/types. - Vaccine hesitancy: The Swedish sample did not include a very hesitant population, limiting insights for extremely hesitant groups. - Ethical considerations: While the study informs ethical debates about paying for vaccination, normative judgments are not settled by empirical results alone.