## wpiea2021130-print-pdf

## Source details

**Canonical URL:** [wpiea2021130-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021130-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021130-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021130-print-pdf.pdf.json)

---

### Overview and motivation
- Widespread immunization is essential for fighting COVID-19 and supporting the global economic recovery.
- Vaccine rollouts constrained by supply and availability; even after supply issues, insufficient vaccine demand could prevent achieving herd immunity.
- WHO definition used: “the reluctance or refusal to vaccinate despite the availability of vaccines.”
- Vaccine hesitancy ranges from slight reluctance to outright refusal; concerns center on safety, potential side effects, and efficacy, often fueled by misinformation or lack of trust in government and health systems.

### Data and empirical scope
- Primary data source: COVID-19 Behavior Tracker (YouGov and the Institute of Global Health Innovation at Imperial College London).
- Survey coverage: 17 countries; waves mostly bi-weekly between November 2020 and April 2021.
- Sample size: over 114,000 individual observations.
- Typical survey wave size: around 1,000 respondents.
- Country coverage changes: Finland replaced by Israel in early 2021; United States replaced the Netherlands in March 2021.
- Survey items: willingness to take a COVID-19 vaccine (or vaccination status), perceptions of vaccine effectiveness and side effects, obstacles to vaccination, attitudes and information sharing within friends and family networks.
- Daily vaccinations data source: Our World in Data.

### Key stylized facts on vaccine hesitancy
- Aggregate shares:
  - 61 percent agree they will take the COVID-19 vaccine.
  - 22 percent are unsure.
  - 17 percent disagree.
- Cross-country averages:
  - less 40 percent willing to take the vaccine in France (averaged across sample periods).
  - about 77 percent willing in the United Kingdom (averaged across sample periods).
- Temporal trend: COVID-19 vaccine skepticism broadly declined over time; decline coincides with vaccine approval and rollouts. Countries that started vaccination early experienced faster declines.
- Regional clustering matters: regional deviations (e.g., German states; UK regions) can raise local herd-immunity thresholds when refusers cluster.

### Demographic, attitudinal, and peer determinants
- Age:
  - Older people report much higher willingness to be vaccinated than younger cohorts; hesitancy decreases with age.
- Gender:
  - Women are less likely to want to be vaccinated than men; the gender gap is largest among working-age women and mainly reflects more women being unsure rather than outright rejecting vaccination.
- Occupation:
  - Healthcare workers are most likely to want the vaccine, followed by public servants (essential workers) and the retired.
- Household composition:
  - Number of children is negatively associated with agreeing to be vaccinated: Table A.1 coefficients: -0.013**, -0.015***, -0.006**, -0.006* (columns (1)–(4)).
  - Logit odds ratios (Table A.4): 0.946***, 0.938***, 0.963**, 0.958** (columns (1)–(4)).
- Access:
  - Easy access to a vaccination site increases vaccination probability by 4 to 12 percentage points depending on specification.
  - About 55 percent of respondents report some obstacle to getting to a vaccination site.
  - Reported access concerns: Waiting times too long 11 percent; sites too far 6 percent; opening times inconvenient 4 percent; fear of being turned away 6 percent.
  - Table A.1 Easy Access coefficients: 0.115***, 0.113***, 0.044***, 0.045*** (columns (1)–(4)).
  - Table A.4 odds ratios for Easy Access: 1.695***, 1.703***, 1.349***, 1.349*** (columns (1)–(4)).
- Risk perceptions and compliance:
  - Individuals who strongly agree “I am worried about getting COVID-19” have vaccination probabilities more than 20 percentage points higher than those who strongly disagree.
  - Individuals who wear masks and comply with social distancing are more likely to want the vaccine.
  - Safe Behavior (Table A.1): 0.528***, 0.162***, 0.169*** (columns where shown).
  - Wears Masks (Table A.1): 0.098***, 0.045**, 0.047*** (where reported).
- Trust and beliefs:
  - Confidence in healthcare system increases vaccination probability by up to 5 percentage points.
  - Worry about vaccine side effects reduces vaccination intent by about 30 percentage points conditional on controls; Table A.1 category 5 (Str. Agree) -0.284***, -0.297*** (columns shown).
  - Belief that government will provide an effective vaccine increases vaccine demand by almost 50 percentage points relative to those with no trust at all; Table A.1 highest agreement: 0.496***, 0.476*** (columns shown).
  - Table A.4 odds ratios for highest agreement: 19.058***, 17.361*** (reported entries).
- Peer effects:
  - A one percentage point increase in the share of friends/family that will take the vaccine is associated with a 0.5 percentage point increase in own probability of vaccination (Table A.2 "% Peers Will Vac" coefficient 0.005***, standard error (0.000)).
  - Log(Vaccines/100 people) coefficients (Table A.2) across specifications: 0.015**, 0.022**, 0.008, 0.012, 0.004 (columns (1)–(5)); rollout effect attenuates once side-effect and effectiveness concerns are controlled for.
  - Endogeneity caveat: peer measures may be biased upwards because like-minded individuals cluster.

### Empirical model and robustness
- Baseline outcome: indicator = 1 if respondent answered “4- Agree” or “5 – Strongly Agree” to “I will take the COVID-19 vaccine if it becomes available to me”; 0 if “1 – strongly disagree” or “2 – disagree”. Unsure respondents excluded in main specification.
- Model: linear probability model with extensive fixed effects and controls (location-by-week fixed effects, δ_jt); standard errors clustered at the country level.
- Controls: demographic D, access A, vaccine perceptions V, compliance S, trust G; all regressions condition on δ_jt.
- Robustness checks: alternative specification comparing unsure with agreeers; logit latent-variable model with location plus time fixed effects to reduce incidental-parameter bias; Table A.4 presents odds ratios consistent with baseline.
- Model note: linear probability model predicted probabilities < 0 or > 1 for 4 percent of observations.

### Peer effects, misinformation, and information-sharing
- Exposure to warnings against vaccines from peers or online is associated with lower willingness to vaccinate; hesitant respondents more likely to share negative information.
- Figure 8 description: plots average share who have seen or shared warnings against COVID-19 vaccines, segmented by agreement/disagreement/uncertainty about taking a vaccine.
- Faster rollouts increase the likelihood people know someone vaccinated; if adverse side effects remain small, concerns about uptake could taper off.

### Epidemiological model extension and vaccine hesitancy implementation
- SIR extension features: endogenous infection rates, asymptomatic transmission, random testing, and vaccines.
- Vaccine hesitancy introduced by:
  - Imposing a cap on the share of the population that gets vaccinated.
  - Reducing the number of vaccinations administered each day when demand fails to meet supply (including vaccine shopping and unused doses).
- Baseline vaccination path: projected path for the United Kingdom from Radzikowski and Dizioli (2021) based on Airfinity data.

### Vaccine hesitancy thresholds (Table 2)
- Share of the population that wants to be vaccinated:
  - Baseline: Vulnerable 89.35% ; Young 74.07%
  - Median: Vulnerable 75.00% ; Young 62.07%
  - Max: Vulnerable 57.48% ; Young 34.71%

### Scenario definitions and implementation
- Scenario 1:
  - Both hesitancy cap and slowdown in the pace of vaccination operate.
  - Daily vaccinations for a group set to zero once cumulative share vaccinated in that group reaches the threshold in Table 2.
  - Pace of rollout affected in proportion to the level of hesitancy.
- Scenario 2:
  - Hesitancy cap maintained, but governments maintain the same pace of vaccination regardless of hesitancy.
  - No reallocation of vaccines across groups when a group reaches its cap.

### Model findings — Scenario 1 (cap and slower rollout)
- Model matches UK daily death data fairly well.
- Aggregate impact (Feb-2020 to mid-Jul-2021):
  - Increasing hesitancy from baseline to maximum level would cost roughly 18,000 lives.
  - 18,000 lives is 11 percent more than the total death toll expected in the baseline.
- Higher hesitancy leads to a higher peak in daily cases; slower rollout causes earlier divergence and greater infection risk.

### Model findings — Scenario 2 (cap but maintained pace)
- Timing when caps reached (pace maintained):
  - 57.48 percent cap for the vulnerable is reached at the beginning of April.
  - 75 percent cap for the vulnerable is reached by the end of April.
  - 34.71 percent cap for the young is reached in June.
- Aggregate impact (Feb-2020 to mid-Jul-2021):
  - Excess deaths relative to baseline are much lower than in Scenario 1.
  - Comparing baseline to maximum hesitancy yields slightly less than 1,000 excess deaths.
  - Excess deaths in Scenario 2 are almost 20 times lower than in Scenario 1.

### Model assumptions and caveats
- Non-linearity: higher hesitancy leads to caps reached earlier and fewer vaccinated by that time; slowdown compounds effects in Scenario 1.
- Conservative behavioral assumptions: rate of infections remains at its lowest point (β0) once reached; number of daily contacts slowly increases but never reaches pre-pandemic levels within time frame.
- Relaxing vigilance or increasing contacts would increase deaths under both scenarios.
- Health-system pressures (e.g., overcrowded ICUs) not modeled but noted as likely to raise mortality with higher peaks.

### Policy-relevant conclusions and recommendations
- Principal implication: vaccine hesitancy can significantly affect pandemic dynamics if it slows vaccination pace.
- Operational measures to speed rollout until every willing adult is vaccinated:
  - Move quickly through priority groups if demand among vulnerable fails to meet supply.
  - Offer flexible days and times for appointments.
  - Provide multiple vaccination sites to reduce commuting difficulties.
  - Allow individuals to choose vaccine brand within site possibilities to avoid brand hesitancy.
  - Permit pre-registration or signaling of intent to assess demand in advance.
  - Allow different vaccination sites to tailor policy mixes to local hesitancy and impediments.
- Behavioral and communication measures:
  - Accurately and repeatedly inform the public about evidence on vaccine safety and effectiveness to address side-effect concerns.
  - Credibly and regularly inform the public about vaccine efficacy and keep updated information about rollout timing and locations.
  - Use cues and nudges, increased accessibility, and diverse messengers (politicians, celebrities, health experts) to generate positive peer effects and reach the unwilling.
  - Maintain vigilance on mobility and individual precautions (mask-wearing, social distancing) because relaxing these could increase deaths even if vaccination pace is maintained.
- Broader benefits:
  - Faster rollout helps reach herd immunity, reduces chances of new variants, and facilitates safer economic reopening and recovery.

### Key empirical statistics and model metrics
- Aggregate sample: over 114,000 observations; Table A.1 Observations: 83,805; Table A.2 Observations: 53,396; Table A.3 Observations: 71,701; Table A.4 Observations: 83,799 (selected).
- Linear model Within R2 examples: Table A.1: 0.0300, 0.0452, 0.3141, 0.3201 (columns as reported).
- Table A.2 Within R2 examples: 0.0337, 0.0386, 0.3277, 0.3372, 0.3800.
- Table A.4 Pseudo R2: 0.0656, 0.0693, 0.3035, 0.3098 (columns as reported).
- Peer effects: "% Peers Will Vac" coefficient 0.005*** (standard error (0.000)).
- Log(Vaccines/100 people) coefficients reported across Table A.2: 0.015**, 0.022**, 0.008, 0.012, 0.004.

### Suggested avenues for future research
- Targeting geographic and social-network clusters of hesitancy and monitoring local flare-ups after country-level herd immunity.
- Evaluating optimality and implementation of mandatory vaccination strategies given correlations between hesitancy and distrust of institutions.
- Further investigation of higher hesitancy among working-age women and households with children.
- Embedding modelling results in broader economic models to assess short- and medium-term economic impacts of vaccine hesitancy.

*Source: wpiea2021130-print-pdf*

### 1. Introduction ........................................................................................................

### 1. Introduction

### Overview and motivation
- Widespread immunization is essential for fighting COVID-19 and supporting the global economic recovery.
- Vaccine rollouts have been constrained by supply and availability; even when supply issues are resolved, insufficient vaccine demand could prevent achieving herd immunity.
- Vaccine hesitancy ranges from slight reluctance to outright refusal and predates COVID-19; concerns typically center on safety, potential side effects, and efficacy, often fueled by misinformation or lack of trust in government and health systems.
- The WHO defines vaccine hesitancy as “the reluctance or refusal to vaccinate despite the availability of vaccines.”

### Data and empirical scope
- Primary data source: COVID-19 Behavior Tracker (YouGov and the Institute of Global Health Innovation at Imperial College London).
- Survey coverage: data on vaccine attitudes for 17 countries; survey waves mostly bi-weekly between November 2020 and April 2021.
- Sample size: over 114,000 individual observations.
- Typical survey wave size: around 1,000 respondents.
- Country coverage changes noted: Finland was replaced by Israel in early 2021, and the United States replaced the Netherlands in March 2021.
- Survey items include willingness to take a COVID-19 vaccine (or vaccination status), perceptions of vaccine effectiveness and side effects, obstacles to vaccination, and attitudes and information sharing within friends and family networks.

### Key stylized facts on vaccine hesitancy
- Averaging across the full sample of over 114,000 observations, only about 61 percent of respondents agree or strongly agree that they would take the COVID-19 vaccine if available to them.
- There is significant variation in vaccine hesitancy:
  - Across countries and within countries.
  - Over time; vaccine hesitancy appears to be declining over time, especially in countries further along their vaccination programs.
- Even if some countries achieve herd immunity, neighboring countries or regions may not, potentially prolonging the pandemic.

### Demographic and attitudinal determinants (high-level)
- Age:
  - Older people report much higher willingness to be vaccinated than younger cohorts.
- Gender:
  - Women are less likely to want to be vaccinated than men.
  - The gender gap is largest among working-age women.
- Beliefs and trust:
  - Respondents who strongly believe that the government will provide them with an effective vaccine are almost 50 percentage points more likely to take the vaccine than those who do not.
  - Respondents who express strong concerns about side effects are 30 percentage points less likely to take the vaccine than those who do not.
- Peer effects:
  - A one percentage point increase in the number of friends and family that want the vaccine is associated with a 0.5 percentage point increase in own probability of vaccination, indicating clustering of hesitant respondents within social networks.

### Epidemiological modeling and counterfactuals
- The paper extends a canonical susceptible-infectious-recovered (SIR) model to account for COVID-19 features: endogenous infection rates, asymptomatic transmission, random testing, and vaccines; vaccine hesitancy is introduced by:
  - Imposing a cap on the share of the population that gets vaccinated.
  - Reducing the number of vaccinations administered each day when demand fails to meet supply (including vaccine shopping and unused doses).
- Counterfactual findings:
  - Increasing vaccine hesitancy in the United Kingdom to the levels observed in France could have worsened the death toll in the United Kingdom by approximately 18,000 deaths between February and July 2021.
  - If governments can maintain a rapid pace of rollouts so that hesitancy only caps the eventual share vaccinated (but rollout speed is preserved), the model predicts excess deaths would be almost 20 times lower than when rollout speed also falls.
- Policy implication from the model:
  - Increasing vaccination speed substantially reduces cumulative deaths, time to herd immunity, and risk of new variants.
  - It is also important to reduce hesitancy, especially when the share willing to be vaccinated is below levels needed for herd immunity.

### Relationship to the literature
- Builds on literature examining behavioral responses to COVID-19 and vaccination preferences shaped by perceptions of benefits and risks, peer information, and trust in institutions.
- Extends work by Radzikowski and Dizioli (2021) by incorporating vaccination policy and hesitancy into an extended SIR framework.
- Confirms pre-pandemic findings that mistrust, concerns about side effects, and misinformation are key drivers of lower vaccine take-up; emphasizes the role of social media and distrust of traditional media sources.

### Policy-relevant conclusions
- Public health policies and communications that inform the public about vaccine safety and effectiveness are key to containing vaccine hesitancy.
- Policies that increase trust and assuage concerns about potential side effects can have a dramatic impact on vaccine acceptance.
- Speeding up vaccine rollouts is critical—faster rollout mitigates the compounding harms of supply-demand mismatches and reduces excess deaths even when hesitancy caps final coverage.

*Source: wpiea2021130-print-pdf - 1. Introduction*

### appendix provides additional details on some of the questions included in the survey.

### wpiea2021130-print-pdf - appendix provides additional details on some of the questions included in the survey.

### Data and controls
- Survey measures vaccine intent with the statement: “I will take the COVID-19 vaccine if it becomes available to me.”
- Dataset includes controls on location (state or region), gender, age, health, employment status and occupation, household size, and number of children in the household.
- Includes measures of individual attitudes towards COVID-19, confidence in government handling of COVID-19, and confidence in national health systems.
- Daily vaccinations data are taken from the COVID-19 databases compiled by Our World in Data.
- Sample countries: Australia, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan, Netherlands, Norway, Singapore, Republic of Korea, Spain, Sweden, United Kingdom, and the United States.

### Vaccine hesitancy across countries and over time
- Aggregate shares:
  - 61 percent agree they will take the COVID-19 vaccine.
  - 22 percent are unsure.
  - 17 percent disagree.
- Cross-country variation (averaged across sample periods):
  - less 40 percent willing to take the vaccine in France.
  - about 77 percent willing in the United Kingdom.
- Regional clustering matters: regional deviations within countries (example: German states; UK regions) show sizeable differences from national means and can raise local herd-immunity thresholds when refusers cluster.
- General vaccine attitudes track COVID-19 vaccine intent: pre-pandemic measures of agreement that vaccines are safe, important, and effective (from Figueiredo et al. (2020), 2015–2019) are strongly and positively correlated with COVID-19 vaccine intent.
- Temporal trends:
  - COVID-19 vaccine skepticism has broadly declined over time.
  - The start of the decline coincides with vaccine approval and rollouts.
  - Countries that started vaccination early (in Europe and North America) experienced a faster decline in hesitancy.

### Empirical model (baseline)
- Outcome variable: indicator equal to 1 if individual i in location j and date t agrees to take the COVID-19 vaccine (answered “4- Agree” or “5 – Strongly Agree”); 0 if answered “1 – strongly disagree” or “2 – disagree”. Unsure respondents are excluded in main specification.
- Model: linear probability model with extensive fixed effects and controls (equation (1) in source).
- Controls:
  - D: demographic variables (age cohorts: 18—24, 25—34, 35—44, 45—54, 55—64, 65—75, and 75+; gender; household size; number of children; occupation), pre-existing health conditions, COVID-19-related symptoms.
  - A: self-reported ease of access to vaccination site (dummy indicating perceived difficulty due to distance, times, inability to go alone, long waits, fear of being turned away).
  - V: questions about COVID-19 perceptions and vaccine safety/efficacy concerns, worry about COVID-19, confidence government will provide effective vaccine.
  - S: compliance with social distancing and mask wearing summarized as indices between 0 and 1.
  - G: trust in government handling of COVID-19 and confidence in healthcare system response.
- All regressions condition on location-by-week fixed effects, δ_jt. Standard errors are clustered at the country level.
- Robustness checks: alternative specification comparing unsure respondents with agreeers; logit specification; findings consistent across specifications.

### Results — Demographic characteristics and vaccine access
- Age and gender:
  - Older people are much more willing to get vaccinated; vaccine hesitancy decreases with age.
  - Women are less likely to want to be vaccinated compared to men across all age cohorts; the gender gap mainly reflects a higher proportion of women being unsure rather than outright rejecting vaccination.
- Occupation effects (controlling for other characteristics):
  - Healthcare workers are the most likely to want the vaccine, followed by public servants (essential workers) and the retired.
- Access effects:
  - Having easy access to a vaccination site increases the chances a person gets the vaccine by 4 to 12 percentage points, depending on model specification.
  - About 55 percent of respondents report some obstacle to getting to a vaccination site.
  - Reported common access concerns:
    - Waiting times at vaccination sites are too long: 11 percent of respondents.
    - Sites are too far: 6 percent.
    - Sites open during inconvenient times: 4 percent.
    - Fear of being turned away without receiving vaccine: 6 percent.

### Results — Beliefs concerning COVID-19, vaccines, and trust in government
- Risk perceptions and compliance:
  - Individuals who strongly agree with “I am worried about getting COVID-19” have a probability of vaccination more than 20 percentage points higher than those who strongly disagree.
  - Individuals who wear masks and comply with social distancing are also more likely to want to take the vaccine.
- Trust in healthcare system:
  - Confidence in the capacity of the healthcare system to respond to COVID-19 increases the probability of vaccination by up to 5 percentage points.
- Vaccine side-effect concerns and government effectiveness perceptions:
  - Widespread concerns about side effects reduce vaccination intent by about 30 percentage points, conditional on controls.
  - Strong trust that the government will provide an effective vaccine increases vaccine demand by almost 50 percentage points relative to those with no trust at all.
- Policy-relevance:
  - Beliefs about side effects and trust in government vaccine provision have the largest estimated effects and are potentially amenable to public health policies and communication campaigns to inform the public about existing evidence on vaccines.

*Source: wpiea2021130-print-pdf - appendix provides additional details on some of the questions included in the survey.*

### 4.3 Peer Effects and Information

### 4.3 Peer Effects and Information

### Peer effects on vaccination decisions
- Evidence suggests significant peer effects on vaccination through mechanisms such as information sharing and imitation (Rao 2007; Bodine-Baron et al., 2013; Sato and Takasaki 2019).
- Two opposing peer-information channels highlighted:
  - Exposure to warnings against the vaccine from peers can reduce willingness to vaccinate.
  - Individuals with reservations may seek and share negative information with peers, reinforcing hesitancy.
- The results plotted in Figure 8 suggest that both channels play a role in vaccine hesitancy.

### Vaccination rollout and network effects
- To measure rollout effects, the log of vaccinations per 100 people in individual i’s country at date t is included in the regression (equation (1)), and fixed effects are changed from location-by-week to location and week (δ_location + δ_week) to capture rollout impact on intent.
- Empirical findings from Table A.2 (appendix) summarized in text:
  - The rollout has a positive and significant effect on the probability of vaccination in columns (1) and (2) which only include demographic variables.
  - Once concerns about side effects and vaccine effectiveness are controlled for (columns 3 and 4), the effect of the rollout becomes considerably smaller and is no longer statistically different from zero.
  - The effects of other variables are largely in line with Table A.1.

### Measuring peer effects and endogeneity concerns
- Naïve peer measure: percentage of each respondent’s close friends and family that will take the vaccine.
  - Endogeneity concern: individual i wanting to be vaccinated makes it more likely that their friends and family are like-minded, biasing estimates.
- Empirical association reported:
  - In Column (5) of Table A.2, a one percentage point increase in the share of friends/family that will take the vaccine is associated with a 0.5 percentage point increase in the probability of vaccination.
  - This coefficient could be biased upwards due to endogeneity.
- Comparing specifications:
  - Including peer effects attenuates the size of other coefficients between columns (4) and (5) but maintains their directions.
  - Despite interpretational challenges from endogeneity, results are qualitatively robust to network and peer effects.

### Misinformation, side-effect concerns, and observed shares
- Figure 8 (described) plots average share of respondents who have seen or shared warnings against COVID-19 vaccines (online or from peers), segmented by respondents’ agreement, disagreement, or uncertainty about taking a COVID-19 vaccine if available.
- Faster rollouts increase the likelihood that people know someone vaccinated, providing more data on side effects; if adverse side effects remain consistently small, concerns about uptake could taper off.

### Key empirical and methodological notes
- Linear probability model is the main specification for ease of interpretation; drawback: predicted probabilities can be < 0 or > 1 for 4 percent of observations.
- Alternative latent-variable logit model:
  - Net benefit y_i* follows a logistic error term; individual i takes the vaccine if y_i* ≥ 0.
  - Location plus time fixed effects are used (instead of location-by-time) to reduce incidental-parameter bias.
  - Table A.4 presents odds ratios; results are consistent with the baseline.

### Policy implications (linked to peer effects and information)
- Because concern about potential side effects has one of the largest impacts on vaccine propensity:
  - Health authorities should accurately and repeatedly inform the public about the evidence on vaccine safety and address salient concerns.
  - Clear communication can generate positive peer effects: promoting take-up among selected individuals can indirectly encourage peers.
- Because belief that government will provide an effective vaccine is strongly related to propensity to vaccinate:
  - Authorities should credibly and regularly inform the public about vaccine efficacy as new information becomes available.
  - Keep updated information about the vaccination rollout (when and where individuals can expect to get the shot).
- Additional strategies with supporting evidence:
  - Cues and nudges (Milkman et al., 2011) and increased accessibility (Brewer et al. 2017), particularly among those planning to be vaccinated.
  - Communicating information delivered by diverse individuals and expertise (Alsan and Eichmeyer, 2021) can help reach those unwilling to be vaccinated.
- Examples of adopted approaches include public vaccination of politicians and celebrities, appearances by health experts, and easy-to-access up-to-date web pages with vaccine information.

*Italicized source attribution below.*

### 5.2 Adding Vaccine Hesitancy

### 5.2 Adding Vaccine Hesitancy

### Vaccine hesitancy mechanisms
- Vaccine hesitancy affects vaccination efforts through two channels:
  - It introduces a cap on the share of people who will get the vaccine, which can be below the required threshold for herd immunity.
  - It can reduce the average number of people vaccinated per day, for example when low uptake leaves doses unused or discarded.
- Implementation in the model:
  - Both outcomes are introduced by changing the number of vaccines delivered per day relative to the baseline vaccination path depicted in Figure 10.
  - The baseline vaccination path is the projected vaccination path for the United Kingdom produced by Radzikowski and Dizioli (2021) based on data obtained from Airfinity.

### Vaccine hesitancy thresholds (from Table 2)
- Share of the population that wants to be vaccinated:
  - Baseline: Vulnerable 89.35% ; Young 74.07%
  - Median: Vulnerable 75.00% ; Young 62.07%
  - Max: Vulnerable 57.48% ; Young 34.71%

### Scenario definitions
- Scenario 1:
  - Both the hesitancy cap and a slowdown in the pace of vaccination operate.
  - The vaccine hesitancy cap is implemented by changing the number of daily vaccinations for a group to zero once the cumulative share vaccinated in that group reaches the threshold in Table 2.
  - The pace of the vaccine rollout is affected in proportion to the level of hesitancy.
  - Example (notation preserved from source): if baseline daily vaccines for the vulnerable are 푣1,3 and for the young 푣1,6, changing hesitancy to its maximum level yields counterfactual daily vaccines of 푣1,3 × 0.5748? (source shows the numeric transformed expressions as 푣1,3ୈହ଻.ସ଼଼ଽ.ଷହ and 푣1,6ୈଷସ.଻ଵ଻ସ.଴଻).
- Scenario 2:
  - The hesitancy cap is maintained, but governments are able to maintain the same pace of vaccination regardless of hesitancy.
  - No reallocation of vaccines across groups is allowed when a group reaches its cap.

### Findings — Scenario 1 (both cap and slower rollout)
- Model fit:
  - The model matches UK daily death data fairly well, picking up both waves of the disease.
- Aggregate impact (Feb-2020 to mid-Jul-2021):
  - Increasing hesitancy from the baseline to the maximum level would cost roughly 18,000 lives.
  - 18,000 lives is 11 percent more than the total death toll expected in the baseline.
- Additional notes:
  - A higher level of hesitancy leads to a higher peak in daily cases (in scenario 1).
  - Slower rollout causes curves to diverge earlier, increasing chances of infection at any point in time.

### Findings — Scenario 2 (cap but maintained pace)
- Timing of caps reached (when pace maintained):
  - 57.48 percent cap for the vulnerable is reached at the beginning of April.
  - 75 percent cap for the vulnerable is reached by the end of April.
  - 34.71 percent cap for the young is reached in June.
- Aggregate impact (Feb-2020 to mid-Jul-2021):
  - Vaccine hesitancy remains costly in terms of more cases and deaths, but the number of excess deaths relative to the baseline is much lower than in scenario 1.
  - Comparing baseline to maximum hesitancy levels yields slightly less than 1,000 excess deaths.
  - The excess deaths in scenario 2 are almost 20 times lower than in scenario 1.

### Model assumptions and their roles
- Non-linearity:
  - The impact of hesitancy on deaths is non-linear because higher hesitancy leads to caps being reached earlier and fewer people vaccinated by that time.
  - In scenario 1, slower rollout compounds this effect.
- Mobility and adherence assumptions:
  - The model adopts a conservative stance: the rate of infections remains at its lowest point (β0) once it is reached, implying continued adherence to masks and safety protocols even as vaccination progresses.
  - The number of daily contacts slowly increases but never reaches pre-pandemic levels within the time frame considered.
  - Relaxing these assumptions (less vigilance, higher contacts) could lead to more deaths due to vaccine hesitancy in both scenarios.
- Health-system pressures:
  - Not modeled but noted: higher peaks in daily cases can lead to higher mortality via overcrowded ICUs.

### Policy implications
- Principal implication:
  - Vaccine hesitancy can have a significant impact on pandemic dynamics if it slows down the pace of vaccination.
- Recommended operational measures to speed rollout until every willing adult is vaccinated:
  - Move quickly through priority groups if demand among the vulnerable fails to meet supply.
  - Offer flexible days and times for appointments.
  - Provide multiple vaccination sites to reduce commuting difficulties.
  - Allow individuals to choose their vaccine brand within site possibilities to avoid brand hesitancy.
  - Permit pre-registration or signaling of intent to assess demand in advance.
  - Allow different vaccination sites to tailor policy mixes to local hesitancy and impediments.
- Behavioral and communication measures:
  - Maintain vigilance on mobility and individual precautions (mask-wearing, social distancing) because relaxing these could increase deaths even if vaccination pace is maintained.
  - Manage information about COVID-19 vaccines via public health policies and targeted communication about vaccine safety and effectiveness to contain hesitancy.
- Broader benefits of reducing hesitancy:
  - Faster rollout facilitates reaching herd immunity and safely reopening the economy, aiding economic recovery and growth.
  - A larger vaccinated share reduces the chances of new variants evolving.

### Conclusion (key summary points from section 6 related to hesitancy)
- Across the sample of 17 countries (Nov-2020 to Apr-2021), nearly 40 percent of respondents are either unsure or unwilling to take the vaccine.
- COVID-19 vaccine demand has been rising since the first effective vaccine announcement in November 2020, indicating hesitancy can be mitigated.
- Demographic patterns:
  - Higher demand among older cohorts and among men; a gender gap that is consistent across ages.
  - Vaccine demand correlates with concerns about COVID-19 severity, self-reported compliance with protective behaviors, and overall trust in government.
- Peer effects and information sharing:
  - Exposure to warnings against the vaccine from peers is associated with lower willingness to be vaccinated.
  - Hesitant respondents are more likely to share negative information about COVID-19 vaccines with peers.
- Modelling takeaways:
  - When hesitancy reduces both the share vaccinated and the speed of rollout, the impact on COVID-19 dynamics can be dramatic.
  - Policies that increase vaccination speed and reduce hesitancy can yield very large benefits.
- Suggested avenues for future research:
  - Targeting geographic and social-network clusters of hesitancy and monitoring local flare-ups after country-level herd immunity.
  - Evaluating the optimality and implementation of mandatory vaccination strategies given correlations between hesitancy and distrust of institutions.
  - Further investigation of higher hesitancy among working-age women and households with children.
  - Embedding the modelling results in broader economic models to assess short- and medium-term economic impacts of vaccine hesitancy.

*Source: wpiea2021130-print-pdf - 5.2 Adding Vaccine Hesitancy*

### 304. doi: 10.1016/j.vaccine.2013.07.013. Epub 2013 Jul 13. PMID: 23859839.

### wpiea2021130-print-pdf - 304. doi: 10.1016/j.vaccine.2013.07.013. Epub 2013 Jul 13. PMID: 23859839.

### Key empirical findings (Tables A.1–A.4)
- Number of children is negatively associated with agreeing to be vaccinated:
  - Table A.1 coefficients: -0.013**, -0.015***, -0.006**, -0.006* (columns (1)–(4)).
  - Table A.4 (logit odds ratios): 0.946***, 0.938***, 0.963**, 0.958** (columns (1)–(4)).
- Easy access to vaccine is strongly positively associated with agreeing to be vaccinated:
  - Table A.1 coefficients: 0.115***, 0.113***, 0.044***, 0.045*** (columns (1)–(4)).
  - Table A.4 (odds ratios): 1.695***, 1.703***, 1.349***, 1.349*** (columns (1)–(4)).
- Safe behaviors and mask wearing correlate with higher vaccination agreement:
  - Table A.1 Safe Behavior: 0.528***, 0.162***, 0.169*** (columns (1)–(3) where shown).
  - Table A.1 Wears Masks: 0.098***, 0.045**, 0.047*** (where reported).
  - Table A.4 (odds ratios) Safe Behavior: 3.090***, 1.551***, 3.090*** (reported entries).
  - Table A.4 (odds ratios) Wears Masks: 1.017, 1.040, 1.353*** (reported entries).
- Worry about COVID-19 is positively associated with vaccination agreement (graded effects):
  - Table A.1 (coefficients across categories): examples include 0.044***, 0.090***, 0.159***, 0.225*** for increasing levels (where reported).
  - Table A.4 (odds ratios) corresponding category examples: 1.444***, 1.964***, 3.093***, 5.001*** (where reported).
- Worry about vaccine side effects is negatively associated with vaccination agreement:
  - Table A.1 coefficients: category 5 (Str. Agree) -0.284***, -0.297*** (columns shown).
  - Table A.4 (odds ratios) category 5 (Str. Agree): 0.162***, 0.146*** (reported entries).
- Belief that government will provide an effective vaccine strongly increases vaccination agreement:
  - Table A.1 coefficients for highest agreement: 0.496***, 0.476*** (columns shown).
  - Table A.4 (odds ratios) highest agreement: 19.058***, 17.361*** (reported entries).
- Peer effects and vaccination coverage:
  - Table A.2 shows Log(Vaccines/100 people) coefficients across specifications: 0.015**, 0.022**, 0.008, 0.012, 0.004 (columns (1)–(5)).
  - Table A.2 reports "% Peers Will Vac" coefficient 0.005*** (standard error (0.000)) indicating a positive and statistically significant peer influence.
- Model fit and sample sizes (selected):
  - Table A.1 Observations: 83805, 83805, 83805, 60521; Within R2: 0.0300, 0.0452, 0.3141, 0.3201.
  - Table A.2 Observations: 53396, 53396, 53396, 38335, 29826; Within R2: 0.0337, 0.0386, 0.3277, 0.3372, 0.3800.
  - Table A.3 Observations: 71701, 71701, 71701, 51615; Within R2: 0.0191, 0.0272, 0.1772, 0.1814.
  - Table A.4 Observations: 83799, 83799, 83799, 60519; Pseudo R2: 0.0656, 0.0693, 0.3035, 0.3098.
- Standard controls and estimation details:
  - All specifications control for age group by gender, pre-existing health conditions, whether individuals had COVID-19 symptoms, and household size. Occupation and location-by-week (or week and location) fixed effects included as noted per table. Standard errors are clustered at the country level. *, ** and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively.

### Survey items and behavioral measures (Table A.5)
- Safe behaviors (asked about last 7 days): Worn a face mask outside your home; Washed hands with soap and water; Covered nose and mouth when sneezing or coughing; Cleaned frequently touched surfaces; Avoided contact with symptomatic/exposed people; Avoided small/medium/large social gatherings.
- Barriers to vaccination asked (examples): COVID-19 vaccination is not yet available for me; COVID-19 vaccination costs too much; I can’t go on my own; The vaccination site is too far away; Opening times inconvenient; People are turned away; Waiting time too long; Nothing. It would not be hard.
- Additional items: Tested for coronavirus in last 7 days; Perception of government handling of Coronavirus; Belief government health authorities will provide effective vaccine; Confidence in healthcare system; Worry about getting COVID-19; Worry about potential vaccine side effects; Perceptions of peers’ vaccination intentions; Exposure to negative information online and from friends/family; Sharing own concerns about vaccine negative effects.
- Note in source: full list of survey questions can be found here: https://github.com/YouGov-Data/covid-19-tracker

### Figures and pandemic dynamics (Figures B.1–B.2 and Section C)
- Data availability: Figure B.1 indicates survey wave dates by country and week.
- Mobility and contacts: Figure B.2 displays Mobility Relative to Pre‐pandemic Baseline and Daily Contacts alongside predicted mobility.
- Vaccine hesitancy scenarios and COVID-19 cases (Section C):
  - Scenario 1: "Vaccine hesitancy slows down the pace of vaccination and introduces a cap at the total number of people vaccinated at any point in time."
  - Scenario 2: "Authorities adapt to hesitancy and keep the vaccination pace, but the cap remains."
- Case-level outcomes (figures):
  - Daily Number of COVID‐19 Cases plotted up to "1400" (Number of Cases per Day (Thousands)) on vertical axis.
  - Total number of COVID‐19 Cases (Feb‐20 to Jul‐21) shown with numeric labels including 148, 152, 156, 160, 164 (plotted along baseline/hesitancy series).
  - Separate plots indicate Predicted, Baseline Hesitancy; Predicted, Median Hesitancy; Predicted, Maximum Hesitancy for both daily cases and cumulative cases.

### Interpretation highlights (from reported estimates)
- Accessibility and expectations about vaccine effectiveness (government provision) show some of the largest positive associations with willingness to vaccinate (e.g., Easy Access 0.115*** in linear models; Gov. Provide Effective Vaccine up to 0.496***; and odds ratios up to 19.058*** for highest agreement).
- Concerns about side effects are strongly negatively associated with willingness to vaccinate (e.g., -0.284***; odds ratios 0.162***).
- Peer beliefs and local vaccination coverage are positively associated with individual willingness to vaccinate (Log(Vaccines/100 people) and % Peers Will Vac coefficients reported).
- Behavioral correlates (safe behavior index and mask wearing) are positively associated with vaccine willingness, consistent across linear and logit specifications.

*Source: wpiea2021130-print-pdf - 304. doi: 10.1016/j.vaccine.2013.07.013. Epub 2013 Jul 13. PMID: 23859839.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021130-print-pdf.pdf_