Air pollution is a serious issue that affects billions of people worldwide. In 2019, David Boyd, the UN Special Rapporteur on Human Rights and the Environment, stated that the lives and health of more than 6 billion people around the world are seriously threatened by chronic exposure to highly polluted air, and as many as 7 million people die each year from indoor and outdoor air pollution. The conclusion that air pollution negatively impacts physical health has been supported by researchers in many disciplines and fields (McGlade and Landrigan, 2019; Qiu et al., 2019; Chen and Chen, 2020). Long-term exposure to air pollution increases the risk of premature death (Lelieveld et al., 2015), cardiovascular and cerebrovascular diseases (Lim et al., 2019; Mannucci et al., 2019), hypertension (Avila-Palencia et al., 2019; Curto et al., 2019; Li et al., 2020), respiratory disease (Losacco and Perillo, 2018; Tang et al., 2018), and chronic disease (Jiang et al., 2016; Hooper et al., 2018). Air pollution may also contribute to feelings of depression and anxiety (Gu et al., 2019; Fan et al., 2020; Lu et al., 2020). In addition, exposure to air pollution can harm people’s cognitive functioning, which includes attention, visuoconstruction, memory, math ability, reading comprehension, verbal intelligence, and non-verbal intelligence (Zhang et al., 2018; Adar et al., 2019). The impact of air pollution on cognitive functioning encompasses the entire life span, from prenatal development, childhood, and youth to young and old adulthood (Sun and Gu, 2008; Perera et al., 2009; Mohai et al., 2011; Lertxundi et al., 2015; Yorifuji et al., 2016). In addition, most people tend to engage in risk-averse behaviours to prevent the impact of air pollution. When air pollution is high, people tend to avoid outdoor activities, such as cycling, trips, and park usage (Bresnahan et al., 1997; Neidell, 2009; Moretti and Neidell, 2011; Saberian et al., 2017). Individuals in polluted regions also show increased interest in emigration and defensive expenditure. Qin and Zhu (2018) found that a 100-point increase in the air quality index (AQI) predicted a 2.3–4.8% increase in internet searches for “emigration” the next day. Zhang and Mu (2018) found that, in China, a 100-point increase in AQI increased the consumption of all masks by 54.5% and anti-PM_2.5 masks by 70.6%. What is even more concerning is that air pollution is associated with increased aggressive and unethical behaviours (Herrnstadt et al., 2016). For example, using 10-year panel data on counties in the contiguous United States, Burkhardt et al., 2019a found that an increase in PM_2.5 increased violent crime rates, especially assaults. Lu et al. (2018) found that air pollution predicted both violent crimes (murder, rape, robbery, and assault) and property crimes (burglary and motor vehicle theft). Bondy et al. (2020) reached similar conclusions using data from London.

The effect of air pollution is widespread and is not limited to the domains covered by the studies reviewed in this article (Lu, 2020). For example, some scholars have found that air pollution decreases the labour supply by increasing absenteeism (Hanna and Oliva, 2015). The impact of air pollution on human society is far-reaching, especially on social behaviour. Hence, the potential role of air pollution cannot be ignored when exploring a given social behaviour. Scholars from various disciplines and fields agree that air pollution induces risk-averse and unethical behaviours. However, researchers have neglected the relationship between air pollution and prosocial behaviour. Research exploring the relationship between air pollution and prosocial behaviour is relatively scarce. Prosocial behaviours are voluntary and benefit other people, such as charitable donations, justice, cooperation, trust, and altruistic punishment (Luo, 2018). Prosocial behaviour plays an important role in both individuals and society. For individuals, prosocial behaviour enhances their self-moral identity, social self-efficacy, and value of social commitment (Patrick et al., 2018). For society, prosocial behaviour is a powerful means of promoting productivity, increasing social wealth, and resisting social risk (Thielmann et al., 2020). Numerous systematic studies have examined prosocial behaviour. However, few scholars have directly studied the relationship between air pollution and prosocial behaviour. Exploring the association between air pollution and prosocial behaviour using a quasi-experimental approach, Zhao et al. (2020) found that the more polluted the air, the less likely Chinese university students were to engage in prosocial behaviour. Their finding implies that air pollution is significantly negatively associated with prosocial behaviour. However, their research has not provided a complete picture of the complex association between air pollution and prosocial behaviour. On the one hand, the quasi-experimental study allowed us to better observe the effects of air pollution on honest behaviour in the nearest actual situation. However, there are still differences from a completely objective and realistic environment. Obviously, there is a corresponding bias in the study findings. On the other hand, prosocial behaviour is so complex that it is impossible to determine its association with pollution based on only a few studies. Thus, examining the association between air pollution and prosocial behaviour is essential. In addition, Zhao et al. (2020) mainly use Chinese university students as the research participants. It remains to be tested whether the research findings could be extended to society as a whole. In this article, we explore the relationship between air pollution and prosocial behaviour to fill these research gaps.

Our study has important implications for research. First, the previous studies have explored the relationship between air pollution and antisocial or unethical behaviour. Comparatively, the relationship between air pollution and prosocial behaviour has not been thoroughly explored. Therefore, our analysis of the relationship between air pollution and prosocial behaviour could fill a research gap in this field. Second, most prior studies measure prosocial behaviour using questionnaires, experimental methods, or quasi-experimental methods. These methods constitute the dominant paradigm in the current research and are very effective. However, these studies all had methodological biases. It is extremely easy for researchers to incorporate their value judgements or social expectations into the study. This often results in self-reporting or social desirability biases when prosocial behaviour is reported or observed in a context that may rouse such biases in participants. Our study considers actual air pollution and prosocial behaviour occurring in a real environment, which effectively compensates for the biases in the previous studies. Third, we use blood donations occurring in natural scenarios to represent prosocial behaviour, which reflects the complex association between air pollution and prosocial behaviour more accurately than what is attempted in prior studies and thus deepens our understanding of that relationship.

The remainder of this article is organised as follows. The next section outlines the study’s data and provides the summary statistics. Section “Model Specification” presents our econometric model. Section “Results” presents the results of the study. Finally, Section “Discussion” discusses the results and provides conclusions.

Our baseline econometric specification examined the relationship between air pollution and prosocial behaviour, in which prosocial behaviour was measured as individual blood donation, as follows:

The dependent variable Bloodd⁢tj is the count of blood donation type j in district d on day t, and PM25_dt is the corresponding PM_2.5 pollutant index in district d on day t. X_dt is the covariate X in district d on day t. X contains a set of covariates that may affect the estimation results, such as daily maximum temperature, daily minimum temperature, weather, wind direction, and wind rating. Previous studies have found a strong correlation between the weather and air pollution (Burkhardt et al., 2019b). In addition, weather conditions are closely related to psychological activity and social behaviour (Zhao et al., 2020). Therefore, we also included a set of weather-related data as covariates in our model. λ_d denotes the district fixed effect. ξ_t denotes fixed effects for the month and day of the week. ∈^j_dt is the error term of blood donation type j in district d on day t. As individual blood donations are non-negative count variables with overdispersion, we employed the Poisson regression quasi-maximum likelihood (PQML) estimation, which produces consistent parameter estimates, assuming that the conditional mean function is correctly specified, even when the dependent variable is overdispersed (see Wooldridge (2010), Section 18.2 for details and Moeltner et al. (2013) for an example of implementation). To ensure correct inferences, standard errors (SEs) were clustered at the district level. In addition, the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) are not meaningfully different across models. When the PQML standard errors are not corrected for overdispersion, the standard errors are smaller as would be expected if overdispersion is present (Burkhardt et al., 2019b). Given this result, our preferred specification is the fully robust PQML model with standard errors clustered at the district level. This model specification was similar to that of Burkhardt et al. (2019b). As the number of blood donation stations varies from district to district, the number of blood donations used in Equation (1) is the relative number of donations.

Air pollution is likely endogenous in Equation (1) because there may be unobservable factors between PM_2.5 and individual blood donation that vary between districts and times. For example, PM_2.5 and individual blood donation may be correlated with population density, population structure, industrial activity, traffic density, household consumption, etc. Failure to control for these observations effectively may lead to a biased estimation of γ_PM. By including district-fixed effects, the model specification in this study can compare the number of different blood donations between days with different PM_2.5, as the potential for biased estimation is eliminated. We also included a month fixed effect to control for seasonal variation in PM_2.5 and the number of blood donations. In addition, there may be periodic co-movement between PM_2.5 and the number of blood donations; these are not fully identified by the month fixed effects, which are constant across districts. For example, we would like to know the difference between busy weekdays and quiet Sundays in terms of both PM_2.5 and the number of blood donations. To identify such time-varying variables, dummy variables for the day of the week were included in Equation (1) to solve the potential short-run seasonality problem described above.

We also controlled for weather variables, such as daily maximum temperature and daily minimum temperature, and used dummies for weather phenomena such as wind direction and rating. A body of research has shown that there is a close relationship between weather conditions, air pollution, and unethical behaviour (Zannetti, 2013; Bondy et al., 2020). Although our focus is on prosocial behaviour rather than unethical behaviour, blood donation is also a type of social behaviour. Thus, we have a reason to expect a connection between weather and individual blood donation, making it necessary to control for weather variables.

In this section, we first discuss the link between air pollution and prosocial behaviour, measured as PM_2.5, and individual blood donation. Next, we examine the strength and consistency of the main results using several specifications. Then, we explore the heterogeneity of the relationship between air pollution and prosocial behaviour by disaggregating blood donation in terms of gender and age and exploring the moderating role of temperature. Finally, we conduct robustness checks to examine the consistency and reliability of the results.

We explored the relationship between air pollution and prosocial behaviour. We merged data on PM_2.5, individual blood donations, and weather in the district under study. We used PM_2.5 to reflect air pollution and used individual blood donation to reflect prosocial behaviour. We employed PQML to estimate the link between air pollution and prosocial behaviour. We also used a series of high-dimensional fixed effects and alternative model specifications to address endogeneity issues and verify robustness.

Our primary finding is that not all prosocial behaviours are significantly associated with air pollution. We found that only component blood donation was significantly associated with air pollution. People aged 30–60 years are more likely to engage in prosocial behaviours, such as component blood donation, and are more likely to be associated with air pollution. In general, our results suggest that there is a significantly negative relationship between air pollution and specific prosocial behaviours (i.e., component blood donation). Furthermore, a significant difference was found in the relationship between air pollution and both blood donation and whole blood donation. Thus, prosocial behaviour is very complex and must be explored in terms of its several categories.

The results indicate that air pollution and prosocial behaviours are not unrelated; they have a complex and close relationship. Only specific prosocial behaviours can be associated with air pollution. Individual blood donation is a good indicator of prosocial behaviour. To examine prosocial behaviour clearly, it is necessary to categorise individual blood donations into whole blood donations and component blood donations. Although both whole and component blood donations can be considered as the types of individual blood donation, such donations imply different body images and cognitive preferences because of the differences in the donation form, donation duration, and interval time. Whole blood donations mainly occur in vans and houses and thus could also be called “street blood donations.” By contrast, component blood donations have a relatively short interval period and occur at fixed blood donation stations. Component blood donations occur in only two districts. As whole and component blood donations have different interval periods, we must count the number of these two blood donations separately. It was thus reasonable to divide individual blood donations into whole and component blood donations in our study. The data we obtained showed that the first blood donation was the main type of whole blood donation and that regular blood donation was the main type of component blood donation. This study found that regular blood donation in component blood donation was significantly associated with PM_2.5, which could indicate that air pollution may be associated with blood donors who have regular blood donation habits. Our results are largely consistent with those of Zhao et al. (2020), who found that people’s integrity behaviour decreased when air pollution increased. However, we are sceptical of the reliability of their conclusions. First, prosocial behaviour is so complex that it is impossible to identify its association with air pollution based on a few studies alone. In other words, it is necessary to re-examine the association between air pollution and prosocial behaviour. Second, this study mainly used Chinese university students as the research participants, which means the results may not necessarily translate to society as a whole. Third, the quasi-experimental study allowed us to better observe the effects of air pollution on behavioural integrity in the nearest real situation. However, there are still differences from a completely objective and realistic environment. Our conclusion also corroborates the relationship between air pollution and unethical behaviour. A previous study showed that air pollution is negatively associated with violent crimes (Berman et al., 2019). Violent crimes constitute traceable behaviour. Police authorities can find relevant traces of a crime. Our results indicate that air pollution is mainly associated with fixed blood donation in component blood donation. Most regular blood donations are traceable. Blood centres can predict when regular blood donors are likely to donate blood based on their records of past blood donations. From this perspective, it is not surprising to find that air pollution has a negative association with blood donation.

We lacked the data required to identify the mechanism by which air pollution is associated with blood donation. We attempted to explain the reasons for the association between air pollution and blood donations. Component blood donations are mainly given by regular blood donors. Most regular blood donors donate at relatively regular intervals. To be able to donate within the pre-planned time window, they may maintain healthy habits. A connection was observed between the most regular blood donors at the blood centre. They exchanged knowledge and information related to blood donation through WeChat. These donors could be more sensitive to surrounding circumstances and risks. They could cancel or postpone their blood donation plan when PM_2.5 increases. Regular blood donors may feel strong faith in and responsibility for their blood donation and may also feel that health is the basis of it. Thus, the number of component blood donations may decrease when PM_2.5 increases. Our results show that most regular blood donors were aged >30 years. We suggest that this is the reason for the difference between donors under 30 and those over 30. Blood donors over 30 may have better physical health management habits than those under 30. In China, most young people aged 18–29 are still students or at the beginning of their careers and are less capable of managing their physical health. Some scholars have found that the youth struggle at this stage of their lives because of many changing pressures. They often lack control over their own lives and require balance. They struggle with health-related issues such as unhealthy eating, lack of sleep, obesity, and lack of exercise. Quick-fix technological solutions are sought to add discipline to their lives, but this is often insufficient in the face of external temptation and low self-management efficacy (Tu and Wang, 2019). Our data show that the majority of first-time blood donors are under 30. The blood donation behaviour of first-time donors may show less of a pattern than that of regular donors. Thus, we did not find a significant association between PM_2.5 and first-time blood donation. In addition, a previous study found that men expressed less concern for practically all the risks studied. Although men and women seem to worry about the same risks, women worry slightly more (Gustafsod, 1998). From this point of view, it is not surprising that the negative relationship between PM_2.5 and component blood donation is stronger among women in the same age group (30–60). Finally, we found a moderating role played by daily minimum temperature in the relationship between PM_2.5 and regular blood donation. Some scholars have found that PM_2.5 concentration is the highest in winter and the lowest in summer in China (Wang and Fang, 2016; Wang et al., 2017). The lower the temperature, the higher the PM_2.5 concentration (Xu et al., 2007). There are many reasons for the high PM_2.5 concentration at lower temperatures, such as incomplete emissions from vehicle exhausts, poor air circulation, and weak pollutant diffusion. Thus, the result indicating that lower daily minimum temperature is linked to a stronger relationship between PM_2.5 and regular blood donation in component blood donation is reasonable.

This study has several limitations. First, we did not perform a spatiotemporal comparison across years. We used only 1 year of data from H city to study air pollution and prosocial behaviour among various administrative districts. Although H city is a megacity in China and 1 year of data drawn from it can answer many questions, cross-year studies and comparisons with other cities are still necessary. In addition, the mechanism underlying the association between air pollution and specific prosocial behaviours needs to be further explored. Future studies should attempt to use more empirical evidence to examine these mechanisms.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

SZ: conceptualisation, methodology, software, formal analysis, data curation, writing—original draft preparation, and visualisation. SZ, LW, and ZG: validation, investigation, resources, and writing—review and editing. LW: supervision, project administration, and funding acquisition. ZG: conceptualization, methodology, data curation, writing—original draft preparation, supervision, project administration and funding acquisition of the study. All authors have read and agreed to the published version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.