1. Introduction
How do people in Japan practice prevention measures to protect themselves from the risks of coronavirus disease 2019 (COVID-19)? Do they rationally make decisions about prevention measures? To combat COVID-19, whether people practice appropriate preventive behavior is the key to not only protecting themselves, but also to preventing the spread of the disease. Since the externality of preventive behavior is large, governments in many countries locked down cities and shut down stores and offices. Many studies have been conducted to seek effective prevention measures against COVID-19 [
1,
2,
3,
4,
5]. In the field of economics, research on quarantine [
6,
7,
8] and prevention measures [
9,
10] has been carried out.
In Japan, there is no legislation to force people to self-isolate in their homes or offices. Therefore, the Japanese government requests the cooperation of citizens and applies moral pressure on them, relying on altruism and conformity to the social norm. To fulfill the request of voluntary self-control for preventive purposes, individuals are required to make a rational plan. This study investigates whether residents in Japan make rational preventive behavioral decisions.
Rational people will consider the benefits and costs associated with the COVID-19 in their decisions about adopting prevention behaviors. Benefits depend on how severe the disease is when one is infected, the probability of an infection, and the effectiveness of prevention measures, while costs refer to monetary and opportunity costs as well as the discomfort caused by the measures. These considerations were formalized as the health belief model (HBM), which has been one of the most extensively used theories to explain preventive health behavior in terms of certain belief patterns [
11,
12,
13]. This model has been adapted to examine various health behaviors, including vaccination [
14,
15,
16,
17,
18,
19]. The HBM comprises several main categories that were identified as predictive factors of decisions regarding the influenza vaccine: perceived susceptibility, perceived severity, perceived benefits, perceived barriers, self-efficacy, and cues to action [
12]. For example, several studies, such as [
14,
15], found the HBM categories among the predictive factors of the decision to receive a flu shot, such as higher levels of perceived seriousness of the illness and higher levels of perceived susceptibility (higher probability of contracting the disease). In addition, a large meta-analysis study indicates that risk likelihood, susceptibility, and severity significantly predicted vaccination behavior [
20] (18 Brewer et al., 2007).
Numerous studies have been undertaken about vaccination against COVID-19 and the health belief model in the COVID-19 pandemic. For example, many papers reported the willingness to pay (WTP) for a hypothetical vaccine against COVID-19. Using an online survey answered by 566 individuals conducted between 18 April and 5 May 2020, in Chile, ref. [
21] found that the individual’s willingness to pay (WTP) for a hypothetical COVID-19 vaccine was US
$184. Similarly, using the data of 531 individuals conducted in July–August 2020, ref. [
22] found that WTP of a vaccine was US
$232 per vaccine, which is quite high in Chile. Using an online survey conducted from 2 April to 7 April 2020, in Ecuador, ref. [
23] found that the average WTP values ranged from US
$148 to 197. Ref. [
24] conducted an online discrete choice experiment survey between June and July 2020, in China. They asked about a hypothetical vaccine and found that there was strong public preference for high effectiveness of the vaccine, followed by long protective duration, very few adverse events and being manufactured overseas. Price was the least important attribute. Using a cross-sectional survey conducted from 3 to 12 April 2020, in Malaysia, ref. [
25] found that WTP was US
$30.66 ± 18.12. In addition, they found respondents believe that the vaccination decreases the chance of infection. These studies asked about a hypothetical vaccine like the current study, and they found WTP was quite high. In addition, the conclusion of ref. [
25] that HBM was supported is also consistent with the current study.
In contrast, those which studied these topics in Japan are relatively few. Some papers simply reported how many people showed the willingness to take a hypothetical vaccine using a cross-sectional survey [
26,
27,
28] assessed the intention to be vaccinated for influenza and rubella, providing information about severe risks and susceptibility. However, the effect of the information was not necessarily clear-cut. In addition, they concluded that older individuals demonstrated vaccine hesitancy for both vaccinations, which seems opposite to the fact of the COVID-19 cases. Ref. [
29], using a cross sectional survey, reported that perceived risks of COVID-19 and those of vaccine are associated with the willingness to take vaccine, being consistent with HBM. Using a cross sectional survey results, ref. [
30] examined the psychological factors that predict staying at home during the COVID-19 pandemic. They found that perceived severity and self-efficacy significantly predicted greater levels of staying at home.
The following is not related to HBM nor vaccination, including two papers by economists. Using official labor data, ref. [
31] documented heterogeneous changes in employment and earnings in response to the COVID-19 shocks. They reported that contingent workers are hit harder than regular workers, younger workers than older workers, females than males, and workers engaged in social and non-flexible jobs than those in ordinary and flexible jobs. Using prefecture-level daily data that contain the numbers of infectious and recovered people, ref. [
32] calibrated the SIR-Macro model to find that a voluntary lockdown and a request-based lockdown play an important role in the low proportion of infectious individuals and the large decrease in consumption in Japan. Ref. [
33] reported suicide cases in 2020 in Japan have increased from late July to November especially for women. Investigating the effects of weather, population and host factors on the outcome of COVID-19 in Japan, ref. [
34] found that the strongest correlation was detected between fatalities and population density followed by total population and by humidity. Using a chatbot-based healthcare system named COOPERA (COVID-19: Operation for Personalized Empowerment to Render Smart Prevention and AN Care Seeking), ref. [
35] analyzed 353,010 participants from Tokyo recruited from 27 March to 6 April 2020. They found that 95.6% of participants had no subjective symptoms. Ref. [
36] found that a substantial proportion of older family carers had relatively low vaccine literacy.
Given the scarcity of vaccination and HBM study in Japan, in this study we comply with the HBM and formalize the model as follows: first, we introduce a hypothetical vaccine, which is expensive but effective, and investigate respondents’ willingness to purchase the vaccine. Since the effectiveness and the cost are provided here, rational models including the HBM, predict that the severity of the disease and the probability of infection are the main predicting factors. Even though in reality various preferences and attributes could play a role in addition to these basic factors, we focus on these two factors in this study by estimating the fixed effect model (FE). In addition, we estimate an FE model by incorporating several time-variant variables, which we conducted in three waves, to check the robustness of the results.
We conducted a panel survey from 13 March to 13 April 2020, which determined the willingness to pay for a hypothetical vaccine, how individuals evaluate the severity of COVID-19, how they predict the probability of infection, and their risk attitudes. The survey collected panel data during three waves, and enabled us to analyze how Japanese prevention behavior changed with the rapid spread of COVID-19.
Whereas most previous studies are based on cross-sectional survey, a merit of the current study is that it is based on a panel survey of larger number of respondents (around 4000), which enabled us to estimate a FE model to investigate the effect of the within-individual changes in the subjective probability of infection and severity of symptom when infected on the willingness to take a vaccine.
The remainder of this paper is organized as follows. In
Section 2, we explain the survey, the COVID-19 situation during the observed period, our hypotheses, and the model tested. In addition to the baseline model, which examines whether HBM applies to all observations, we investigate the HBM model in detail with regard to three aspects. Specifically, we investigate how the sensitivity of the willingness to purchase the vaccine regarding the three explanatory variables changed during the three waves, and how the magnitude and sensitivity of the variables contributed to an increase in willingness. In addition, we examine whether the model applies to sub-samples classified by age, sex, income, or education. Furthermore, we examine whether the baseline model applies to the rule between individuals, although the baseline model focuses on the rule regarding within-individual changes. In
Section 3, we explain how the perception of COVID-19, the willingness to pay for the vaccine, and risk attitude changed as the COVID-19 situation became more severe during the observation period. In
Section 4, we present our empirical results. In
Section 4.1, we show the results of the baseline estimation on how the willingness to purchase the vaccine depends on the perception of COVID-19 (the HBM) and risk attitude. In
Section 4.2,
Section 4.3 and
Section 4.4, we present the results of the investigation of the extended models. Finally,
Section 5 concludes the paper.
5. Conclusions
This study investigated how their willingness to purchase the (hypothetical) vaccine depends on their perceptions of COVID-19 and their risk attitudes. We conducted a large-scale (approximately 4000 respondents for each wave) panel survey three times between 13 March and 13 April and found the following:
When respondents confronted the possibility of COVID-19 infection, their willingness to purchase a vaccine, subjective probability of becoming infected, and possible symptoms if infected increased over time, while they became more risk tolerant.
By estimating the fixed effect model (FE), we found that a person is more willing to buy the vaccine when the person perceives COVID-19 as a more severe disease, when considering a higher probability of infection, and becomes more risk averse. The result provides evidence that people in Japan considered their preventive behavior rationally, as the HBM predicts.
The HBM also applies to between people variation, that is, a person who perceives COVID-19 as a more severe disease, predicts a higher probability of infection, and is more risk averse are more willing to purchase the vaccine.
We analyzed how the three factors contributed to the increase in willingness to purchase the vaccine. Blinder-Oaxaca decomposition reveals that the change in sensitivity contributed more than that of the magnitude of variables, the contribution of possible symptoms if infected is larger than subjective probability of becoming infected. The decrease in risk aversion (i.e., becoming more risk tolerant) during the 4-week period has two contradicting effects: while the decrease in Magnitude is negatively affected, the increase in sensitivity increased willingness to purchase vaccine. Since the effect of risk aversion is complex, an interpretation might be helpful. First, higher risk aversion promotes willingness to buy the vaccine shown in
Table 5. On average, people became more risk tolerant over time (
Table 1 and
Figure S5), which decreased willingness to purchase vaccine (“Magnitude” in
Table 7). Meanwhile, the sensitivity of willingness to purchase vaccine on risk aversion became larger (cross terms in
Table 6), which promoted willingness to purchase vaccine (“Sensitivity” in
Table 7).
The HBM is supported for subgroups classified by sex, income, or education. However, it is not supported by young and middle-aged groups; the model applies only to the old-aged group who consider COVID-19 more seriously. The result is contrary to [
28], which assessed the intention to be vaccinated for influenza and rubella. In the context of the COVID-19 pandemic, however, it is widely known that older people face more risks when infected, therefore, it is reasonable that they are more willing to take vaccines. For example, as of 6 January 2021, the cumulative number of deaths caused by COVID-19 among those aged over 60 was 95% of the total Japanese [
41]. The Ministry of Labour, Health, and Welfare announced on 25 February 2021, “for the elderly and those with underlying illnesses, we will lead to earlier and more appropriate consultations, keeping in mind that they are likely to become severely ill” [
42]. Therefore, the results obtained in this study are reasonable.
Providing information about severe risks and susceptibility. However, the effect of the information was not necessarily clear-cut. In addition, they concluded that older individuals demonstrated vaccine hesitancy for both vaccinations, which seems opposite to the fact of the COVID-19 cases.
Furthermore, the importance of this study lies with the fact that it analyzes the data of individuals faced with the mega risk COVID-19. This enabled us to analyze the prevention behavior under the situation of a mega risk. Another accomplishment is the use of the panel data during the limited period of four weeks, which enabled us to assume that most attributes are fixed, while the situation concerning possible infections from COVID-19 changed radically. A difficulty of the HBM analyzed in this study is the endogeneity problem among key variables. First, willingness to purchase vaccine, subjective probability of becoming infected, possible symptoms if infected, and risk aversion may depend on hidden fixed confounders and time-variant factors during the observed period. The use of the FE and limited observation period minimized this problem. Another cause of the endogeneity is the reverse causality from prevention behavior to subjective probability of becoming infected and possible symptoms if infected. If a person takes the vaccine, subjective probability of becoming infected and possible symptoms if infected will be lowered. Questioning respondents about a hypothetical vaccine probably mitigates the reverse causality, if not perfects it. Given that previous literature on vaccines and HBM in the COVID-19 pandemic is scarce in Japan, we hope that the present study contributes to the accumulation of knowledge in the field.
Naturally, there are many limitations. Whether people truly make rational decisions and adopt rational behavior under the threat of COVID-19 and whether the prolonged request of self-control is effective, may be some of the most important problems to solve. This is more important in Japan, because the government is not allowed to restrict peoples’ and companies’ behavior, so that it only requests them to follow the guidelines, such as avoiding of going out or closure of operations. In this study, we found that citizens in Japan make rational decisions regarding willingness to purchase a hypothetical vaccine. However, the COVID-19 pandemic has been continuing for more than a year since the observation period. According to [
32], which uses the data until 27 December 2020, the voluntary stay at home in response to the risk of infection and the government’s request to stay at home play an important role in the low proportion of infectious individuals and the large decrease in consumption in Japan, suggesting that Japanese people have been rational in their response to COVID-19. In addition, as of 19 November 2021, 75.8% of the total population completed the second dose [
43], which is relatively high compared with other developed countries [
44]. However, it is important to investigate in future work whether Japanese people have continued to be rational concerning vaccination.
Whether people rationally adopt true prevention behavior, such as avoiding going out and receiving government-approved vaccines also remains to be assessed in future research. Regarding the progress of vaccination in countries, easy and fast availability of the vaccine, low vaccination costs, having no serious side effects, having strong effectiveness, and availability of the vaccine at any facility are important factors. In addition, belief in religion also affects people’s willingness to be vaccinated [
45]. Furthermore, there are three factors to consider when compelling people to adopt appropriate prevention behaviors. First, to fill the gap between plans (willingness) and behavior, people require a strong will to achieve the plan. Psychologists and behavioral economists argue that people have present bias and various misperceptions [
46,
47,
48]. Therefore, if people fail to exercise preventive behavior such as “self-control regarding no outings,” we need to devise measures to promote appropriate behavior.
The second problem refers to the socially desirable level of preventive behavior that exceeds the desirable level of selfish individuals because of the externality of the infection. Discrepancies between the social optimum and the outcome of freedom may lead to limited freedom. Ref. [
49] found that altruism plays an important role in promoting flu vaccines in Japan, suggesting that the externality problem might be mitigated if people are more considerate of others.
Third, although the free-rider problem will obstruct the internalization of the externality of the infection, Japanese individuals are characterized with more collectivistic rather than individualistic traits [
50,
51,
52]. Economic experiments using public games found that Japanese people punish those who do not cooperate more than Americans [
53,
54]. These types of forces, such as spiteful behavior towards outliers, which was referred to as “Hachibu” in the past meaning “neglect from the community,” could discipline people to cooperate. Even though collectivism, companion consciousness within a small group, and spirit of community has been considered primitive in Japanese society, severe trials due to COVID-19 might stimulate reconsideration of communitarianism. If successful, request-based policies without enforcement by law could be effective in preventing the spread of the infection.