Enforcement may crowd out voluntary support for COVID-19 policies, especially where trust in government is weak and in a liberal society

Abstract

Effective states govern by some combination of enforcement and voluntary compliance. To contain the COVID-19 pandemic, a critical decision is the extent to which policy makers rely on voluntary as opposed to enforced compliance, and nations vary along this dimension. While enforcement may secure higher compliance, there is experimental and other evidence that it may also crowd out voluntary motivation. How does enforcement affect citizens' support for anti-COVID-19 policies? A survey conducted with 4,799 respondents toward the end of the first lockdown in Germany suggests that a substantial share of the population will support measures more under voluntary than under enforced implementation. Negative responses to enforcement-termed control aversion-vary across the nature of the policy intervention (e.g., they are rare for masks and frequent for vaccination and a cell-phone tracing app). Control aversion is less common among those with greater trust in the government and the information it provides, and among those who were brought up under the coercive regime of East Germany. Taking account of the likely effectiveness of enforcement and the extent to which near-universal compliance is crucial, the differing degrees of opposition to enforcement across policies suggest that for some anti-COVID-19 policies an enforced mandate would be unwise, while for others it would be essential. Similar reasoning may also be relevant for policies to address future pandemics and other societal challenges like climate change.

Keywords: cooperation; crowding out intrinsic motivation; institutions; social norms; state capacities.

Fig. 1.

Cumulative distributions of agreement under the enforced vs. voluntary implementation of the five measures (panels A–E). For example, A shows that 18% and 36%, respectively, of respondents fully agree to use a contact-tracing app if it is enforced and voluntary, respectively. The sum of those expressing either agreement level 3 or 4 amounts to 29% and 51%, respectively, in case of enforcement and voluntariness, respectively. Strongest opposition (level 0) was expressed by 41% and 22%, respectively, under enforced and voluntary implementation, respectively (the final step in the upper left graph). The sample sizes are as follows. Use tracing app: n = 4,787 (if voluntary) and n = 4,777 (if enforced); Get vaccinated: n = 4,787 (if voluntary) and n = 4,786 (if enforced); Limit contacts: n = 4,790 (if voluntary) and n = 4,792 (if enforced); Limit traveling: n = 4,784 (if voluntary) and n = 4,794 (if enforced); Wear a mask: n = 4,776 (if voluntary) and n = 4,781 (if enforced).

Fig. 2.

Shares of types of responses to enforced vs. voluntary policies. Responses to enforcement are negative (neutral and positive, respectively) if agreement is lower (equal to and higher, respectively) in case of enforced compared to voluntary implementations of a measure. Negative responses are termed control aversion. The sample sizes are as follows. Use tracing app: n = 4,770; Get vaccinated: n = 4,776; Limit contacts: n = 4,783; Limit traveling: n = 4,781; Wear a mask: n = 4,769. The underlying choice distributions are provided in SI Appendix, Fig. S3.

Fig. 3.

Predictors of control aversion (as measured by the difference between voluntary agreement and agreement under enforcement) for the five measures. Shown are the coefficients and 95% CI on control aversion, estimated in ordinary least squares linear regressions with standardized variables. SI Appendix, Tables S4 and S5 provide the full regressions. (A) For example, (a) shows that a SD difference in the extent to which people distrust the German government is associated with increased control aversion of somewhere between 10% and 15% of a SD for contacts, travel, and masks. (B) The East Germany effect reduces control aversion. The values are based on regressions identical to those of Fig. 3A except that the continuous variable for age is replaced by a dummy indicating born before 1970. The top group of bars shows that for all domains other than wearing a mask, older respondents from the East are about 13% to 16% of a SD less control-averse than those from the West.