How policies affect beliefs and preferences: The example of covid-19 vaccination willingness

Katrin Schmelz, Samuel Bowles 01 July 2021



Four decades ago, Robert Lucas rocked economics with a simple observation: taxes and other government interventions in the private economy affect not only the costs and benefits of actions citizens may take (as intended), but also their beliefs about the future actions of others (including the government), possibly in counterproductive ways. For example, announcing stiffer penalties for non-payment of taxes provides an incentive to pay up; but it also may convey the information that non-compliance is common, leading formerly honest citizens to cheat. 

Lucas’ point was that policymaking is not simply resetting the dials on a given model of how the economy works, but instead changing the structure of the model itself (Lucas 1976). In 1994, Henry Aaron of the Brookings Institution pointed to “the failure of economists to take the formation of preferences seriously”, and suggested that the Lucas critique be extended to cover preferences as well as beliefs (Aaron 1994).

At the time there was little evidence on which policymakers could draw if they wished to take account of the endogenous nature of preferences and beliefs. But experimental economics has filled many of the gaps (as surveyed by Bowles and Polania-Reyes 2012). The response to COVID-19 vaccinations provides rare new evidence from outside the lab (Schmelz 2021, Schmelz and Bowles 2021).

Framing and learning: How policies affect preferences and beliefs

There are two senses in which a policy may alter preferences and beliefs. We term these framing and learning. The first, a short-term effect, is that the incentives or constraints by which the policy is implemented frame the citizen’s decision in a particular way. The psychologist Mark Lepper and his colleagues wrote that “superfluous social constraints”, for example, unavoidably convey information about “the presumed motives” of the policymaker, as well as the reasons why the policy was adopted, and “the relationship” between the policymaker and the citizen (Lepper et al. 1982).

The second sense in which policies may alter preferences and beliefs is a longer-term change in preferences due to learning. By structuring the daily experiences of the citizen – in their interactions with others and the state – a policy may alter the longer-term evolution of a person’s preferences and beliefs (Bowles 2008, Bowles and Polania-Reyes 2012, Schmelz and Ziegelmeyer 2020). 

Our evidence: Enforced COVID-19 vaccination may crowd out voluntary compliance

We have found evidence of both framing and learning in citizens’ responses to vaccinations and other COVID-19 pandemic policies. 

Our representative online panel of 2,653 Germans, surveyed by the Cluster of Excellence "The Politics of Inequality" at the University of Konstanz both in April/May and in October/November of 2020, allows us to track the attitudes of the same individuals during the first and second lockdowns. In both waves, respondents were asked: “If there is an approved vaccine against the coronavirus, to what extent would you agree to be inoculated yourself if: … vaccination is strongly recommended by the government but remains voluntary? … vaccination is made mandatory and checked by the government?” Answers were given on a five-point Likert scale ranging from 0 (“not agree at all”) to 4 (“fully agree”). 

While support for voluntary vaccination remained high in the second wave, the fraction fully supporting enforced vaccinations was initially lower and then dropped from 44% to 28%, as shown in Figures 1A and 1B.

Vaccination willingness has declined among those who lost trust in public institutions

We found that trust in public institutions is a strong predictor of willingness to be vaccinated under voluntary or enforced policies. We measure trust in public institutions by the average of the subject’s expressed general trust in the federal government and specific trust in its truthful information about COVID-19, as well as their trust in the state government, in science, and in media. We term this measure ‘public trust’. 

The fall in support for enforced vaccination occurred disproportionately among those whose public trust had declined between the two waves of the survey. This lends some support to the idea that the positive correlation between trust and vaccine support reported in cross-section data (e.g., Bargain and Aminjonov 2020) may reflect a causal relationship. The distrust effect is substantial: a one-point drop in our public trust measure (ranging from 1 to 6.6) would account for 37% of the observed reduction in support for enforced vaccines between the two waves.

This is consistent with evidence from the UK that people in locations where intensive care units were under stress early in the pandemic were more resistant to vaccination later, perhaps because they had lost confidence in medical institutions (Blanchard-Rohner et al. 2021).

Figure 1 Reduced support for enforced vaccination

Notes: (A) Average agreement to get vaccinated if it is voluntary or enforced in the two waves of the survey (in Likert scale units). Error bars represent 95% CI. (B) Cumulative distributions of agreement in case of enforced versus voluntary vaccination for the two waves of the survey. For example, the dashed and solid red lines show that 44% and 28% of respondents fully agreed to get vaccinated in case of enforcement in the first and second waves of the survey, respectively. 

Why enforcement may crowd out intrinsic motivation to contribute to a public good

The fact that enforcement reduces support for vaccinations is surprising in light of what we know from experimental public goods games. Most anti-COVID-19 policies share the fundamental structure of public goods dilemmas where all-encompassing cooperation maximises the wellbeing of all citizens, but since cooperation is costly each individual has an incentive to free ride on others if the level of cooperation is sufficiently high (Gollier 2021). 

Experiments with public goods games around the world have shown that in the absence of an option to punish free riders, substantial levels of initial cooperation typically decline as contributors become discouraged or angered by those not contributing (Herrmann, Thöni, and Gächter 2008). But a belief that most others will cooperate – or be punished if they do not – encourages conditional cooperators to do the same (Fischbacher, Gächter, and Fehr 2001, Shinada and Yamagishi 2007). This would lead us to expect that average agreement to follow anti–COVID-19 measures should be higher if a measure is enforced than if it remains voluntary.

But the lesser support for mandated than voluntary vaccinations that we found suggests a different line of reasoning. An enforcement-based approach might be compromised because it crowds out voluntary commitment. Economists term this ‘control aversion’, a particular case of intrinsic motivation being diminished by explicit constraints or incentives (Falk and Kosfeld 2006, Ziegelmeyer, Schmelz, and Ploner 2012). 

Our data suggest that this occurs in the case of vaccine willingness. We have identified three mechanisms which might account for this (Bowles and Polania-Reyes 2012).

The first, termed ‘psychological reactance’, is the result of individual strivings for freedom or ‘self-determination’ (Lepper et al. 1982, Rudorf et al. 2018). It is consistent with our finding that the opposition to enforced vaccinations is substantially greater among respondents who reported that it would restrict their ‘freedom’. 

Second, ‘moral disengagement’ occurs when the provision of explicit incentives or constraints frames the decision problem as one in which ethical convictions are not salient (Falk and Szech 2013). Voluntary vaccination policies may trigger moral deliberation and the desire to be a good citizen. By contrast, enforcement might relieve the citizen of any need to deliberate, and thus crowd out those moral convictions (Bowles 2016). 

Resistance to vaccination provides an example of moral disengagement. In the second wave of the survey, those reporting greater altruism – willingness to help others – were also more likely to support voluntary but not enforced vaccinations. The negative impact of enforcement is greater among citizens reporting more other-regarding preferences. 

The third mechanism through which enforcement may crowd out intrinsic motivation is by diminishing trust. Enforcement may signal that the policymaker knows that the vaccine is not something citizens would willingly subject themselves to. In addition to ‘bad news’ about the vaccine, the distrust communicated by enforcement signals low expectations about citizens’ behaviour (Sliwka 2007). In the eyes of citizens, this may result in a mutually distrusting relationship (Bartling, Fehr, and Schmidt 2012, Fehr and Rockenbach 2003), promoting vaccine hesitancy, as our panel data show.

Learning control aversion: The role of institutions

Because the control-averse reactions shown in Figures 1A and 1B were responses to a hypothetical question about mandated vaccination, we consider them to be temporary framing effects, subject to reversal by reframing the decision, instead of a genuine change in preferences.

But our survey also provides evidence of longer-term preference change. 

Respondents who had lived under liberal institutions in West Germany since the end of WWII were more control averse than those who had experienced two generations of authoritarian rule beginning of 1933. Figure 2 shows that the East–West difference is qualitatively larger for older Germans who were at least 20 years old when the Berlin Wall came down (mean age at survey time: 62 years) than for younger Germans who experienced less than 20 years under different regimes (mean age: 35 years). This finding is in line with evidence from an online experiment in a setting unrelated to COVID-19 policies (Schmelz and Ziegelmeyer 2020).

Figure 2 Effect of experience in East (compared to West) Germany on control aversion with respect to anti-COVID-19 policies

Notes: Control aversion is measured by the difference between voluntary agreement and agreement in case of enforcement in the five domains. Shown are the coefficients and 95% CI on control aversion. For example, the upper part of the figure shows that older East Germans are somewhere between 13% and 16% of a standard deviation less control averse than older West Germans in all domains except for masks.

Older East Germans are less control-averse than older West Germans in all domains except for masks. This is consistent with the effect of mere exposure (Zajonc 2001): those brought up in East Germany prior to 1990 were subject to ubiquitous surveillance, compulsory vaccination, and restrictions on movements. 

By contrast, wearing masks is rather exotic and not part of the experience of East (or West) Germans prior to the pandemic. This result is remarkable as it shows that people who experienced the coercive East German regime three and more decades ago are still less averse to enforced anti-COVID-19 measures.

Policy: Putting the survey results to use

Legally required vaccination against measles and other diseases is an essential part of public health policies around the world. COVID-19 vaccination mandates, for example, targeting those caring for the elderly may be an important element of a policy response to the pandemic. 

But mandating COVID-19 vaccination may also have a substantial negative impact on voluntary compliance. Given limited state capacities and persistent citizen opposition, ending the pandemic in a timely manner by universal enforcement could be impossible. Moreover, enforcement could bear costs including heightened social conflict and citizen alienation from their government or professional elites. The result could be a downward spiral of public distrust fuelling vaccine resistance, triggering more extensive enforcement and in turn further eroding public trust. Costly errors may be avoided if policymakers understand that citizens’ preferences are not fixed but will be affected by the crowding-out effect of enforcement - in varying degrees depending on the policies, as we have found (Schmelz 2021).

Our findings may bear lessons beyond COVID-19. Consider policies to address climate change. Sustaining environmentalist values will be essential for the passage and implementation of the necessary mandatory policies. Moreover, green social norms will be necessary to promote the lifestyle changes that reduce our carbon footprint, given limited state enforcement capacities in this realm. So, the feasibility and effectiveness of environmentalist initiatives may depend on the ways policies themselves alter beliefs and preferences. 

Important considerations to evaluate the wisdom of voluntary versus enforced policies go beyond whether enforcement evokes substantial control aversion and include the level of compliance required for a policy to be successful, the share of citizens who would comply with a policy voluntarily, and the extent to which effective enforcement is feasible. 


Aaron, H (1994), "Public Policy, Values, and Consciousness", Journal of Economic Perspectives 8 (2):3-21.

Bargain, O and U Aminjonov (2020), "Trust and compliance with public health policies in the time of COVID-19",, 23 October.

Bartling, B, E Fehr, and K M Schmidt (2012), "Screening, Competition, and Job Design: Economic Origins of Good Jobs", American Economic Review 102(2): 834-864. doi: 10.1257/aer.102.2.834.

Blanchard-Rohner, G, B Caprettini, D Rohner, and H-J Voth (2021), "From tragedy to hesitancy: How public health failures boosted COVID-19 vaccine scepticism",, 1 June. 

Bowles, S (2008), "Policies Designed for Self-Interested Citizens May Undermine "The Moral Sentiments", Evidence from Economic Experiments", Science 320(5883): 1605-1609. doi: 10.1126/science.1152110.

Bowles, S (2016), The Moral Economy - Why Good Incentives are no Substitutes for Good Citizens, Yale University Press.

Bowles, S and S Polania-Reyes (2012), "Economic Incentives and Social Preferences: Substitutes or Complements?", Journal of Economic Literature 50(2): 368-425. doi: 10.1257/jel.50.2.368.

Falk, A and M Kosfeld (2006), "The Hidden Costs of Control", American Economic Review 96(5): 1611-30.

Falk, Armin, and Nora Szech. 2013. "Morals and Markets", Science 340(6133): 707-711. doi: 10.1126/science.1231566.

Fehr, E, and B Rockenbach (2003), "Detrimental effects of sanctions on human altruism", Nature 422(6928): 137-140.

Fischbacher, U, S Gächter, and E Fehr (2001), "Are people conditionally cooperative? Evidence from a public goods experiment", Economics Letters 71(3): 397-404.

Gollier, C (2021), "Cost-benefit analysis of the Covid vaccination campaign in France", VoxEU, 9 April.

Herrmann, B, C Thöni, and S Gächter (2008), "Antisocial Punishment Across Societies", Science 319(5868): 1362-1367. doi: 10.1126/science.1153808.

Lepper, M R, G Sagotsky, J Defoe, and D Greene (1982), "Consequences of Superfluous Social Constraints: Effects on Young Children's Social Inferences and Subsequent Intrinsic Interest", Journal of Personality and Social Psychology 42(1): 51-65.

Lucas, R E, Jr. (1976), "Econometric Policy Evaluation: A Critique", Carnegie-Rochester Conference Series on Public Policy 1: 19–46.

Rudorf, S, K Schmelz, T Baumgartner, R Wiest, U Fischbacher, and D Knoch (2018), "Neural Mechanisms Underlying Individual Differences in Control-Averse Behavior", The Journal of Neuroscience 38(22): 5196-5208. doi: 10.1523/jneurosci.0047-18.2018.

Schmelz, K (2021), "Enforcement may crowd out voluntary support for COVID-19 policies, especially where trust in government is weak and in a liberal society", Proceedings of the National Academy of Science 118. doi:

Schmelz, K and S Bowles (2021), "Overcoming COVID-19 vaccination resistance when alternative policies affect the dynamics of conformism, social norms, and crowding out", Proceedings of the National Academy of Sciences 118(25): e2104912118. doi: 10.1073/pnas.2104912118.

Schmelz, K and A Ziegelmeyer (2020), "State coercion and control aversion: Evidence from an internet study in East and West Germany", TWI Research Paper Series 117.

Shinada, M and T Yamagishi (2007), "Punishing free riders: direct and indirect promotion of cooperation", Evolution and Human Behavior 28(5): 330-339.

Sliwka, D (2007), "Trust as a Signal of a Social Norm and the Hidden Costs of Incentive Schemes." American Economic Review 97(3): 999-1012.

Zajonc, R B (2001), "Mere exposure: A gateway to the subliminal", Current Directions in Psychological Science 10(6): 224-228. doi: 10.1111/1467-8721.00154.

Ziegelmeyer, A, K Schmelz, and M Ploner (2012), "Hidden costs of control: four repetitions and an extension", Experimental Economics 15(2): 323-340. doi: 10.1007/s10683-011-9302-8.



Topics:  Covid-19

Postdoctoral Researcher, University of Konstanz and Thurgau Economic Institute

Research Professor and Director of the Behavioral Sciences Program, Santa Fe Institute


CEPR Policy Research