Air pollution is a leading cause of death and morbidity worldwide, particularly in low- and middle-income countries where monitoring is sparse and public policy responses are often weak. But while the scale of the problem has been well documented, that scale itself raises an important question: why isn’t there more demand for action on pollution and greater investment in defensive technology? One possibility is that the lack of robust monitoring networks in many of these places have led to substantial information failures that inhibit individual actions and reduce demand for cleaner air.
A recent session at the 2025 EAERE conference brought together three papers that all focus on this second possibility, that information provision is below the optimum. They each examine the impact of providing better information about pollution from different angles in three different cities in LMICs around the world.
Collectively the authors run RCTs in Delhi, Tbilisi, and Mexico City to help shed light on how providing better information is (a) valued by the recipients and (b) the extent to which it changes behaviors among the recipients. Together, they suggest that information matters and that there is scope for better information to improve outcomes with respect to pollution exposure, but that there are limits to its effectiveness.
Measuring demand for clean air in Delhi
The first study, conducted in low-income neighborhoods in Delhi, India, uses an experiment to estimate willingness-to-pay (WTP) for clean air. Researchers randomly varied the price they offered to participants for a high-quality pollution mask. They interpret the resulting demand curve they estimate as measuring demand for clean air. On average, they find that people were willing to pay about $1.14 for a 10 unit annual reduction in PM2.5. This is a relatively low WTP by global standards, and remains low even when compared to estimates from other LMICs.
However, the authors find that when they provide participants with information about air pollution it dramatically increases their WTP. Participants were randomized into receiving a simple handout and short video describing the health harms of pollution. Those that received this information treatment had WTP that was more than 5x higher than among those who did not receive information. Respondents who received the information were also more sensitive to variation in pollution levels when deciding whether to purchase a mask. Importantly, this effect was concentrated among those with lower education, suggesting that information gaps may be most important for the most vulnerable. However, even after the increase in WTP induced by the information treatment, WTP remains low relative to global standards, suggesting limits to how much action can be generated by providing more information.
Highly local alerts and health gains in Tbilisi
The second study (and one that I was involved in conducting), this time in Tbilisi, Georgia, aimed to provide a different kind of information about air pollution. Rather than provide information about the harms, the authors provided information from a network of highly local monitors they established throughout the city. The goal was to determine to what extent timely and local information can help individuals change their behavior to avoid pollution in the first place.
In the RCT, households were assigned to receive different levels of air quality information: a pamphlet about the harms of pollution exposure, daily SMS alerts about outdoor pollution, and in some cases, real-time indoor pollution monitors. The treatments stacked on one another (e.g. those who received daily SMS alerts also received a pamphlet, etc.).
The pamphlet alone had no measurable effect. But daily SMS messages about outdoor pollution led to more self-reported awareness and avoidance behavior. These changes did not manifest as improvements in health within the timeframe of the intervention however. Households that were also informed about indoor pollution levels (which were often higher than outdoor levels), made larger changes; most importantly reductions in indoor smoking. In this group these changes were large enough to produce detectable health improvements during the intervention.
Despite the improvements in measured outcomes the information generated, households weren’t necessarily willing to pay for this information. Households broadly had low WTP to pay either for monitors or for air purifiers after experiencing the RCT intervention and there was no meaningful change in WTP across the treatment arms.
WTP for information in Mexico City
The third study, in Mexico City, directly examines WTP for information (vs. cleaner air), which neither of the previous papers examine directly. Households were offered a subscription to air quality alerts via SMS. The authors then measured how much people were willing to pay for the alerts and how they responded after receiving them.
Most people reported positive willingness to pay, and the alerts increased avoidance behavior. But experiencing actual high-pollution days didn’t predict willingness to pay, rather self-reported perceptions of high-pollution days (that were not necessary accurate) predict higher WTP. People seem to respond to a general sense that pollution is often bad but not the specific days on which it is bad.
Receiving the alerts seems to reinforce these generalized responses rather than increasing targeting of behavioral responses. After receiving alerts participants were more likely to stay indoors, but not necessarily on the dirtiest days. Alerts seem to shift general beliefs about average pollution levels rather than helping people target specific days.
What these studies tell us
Collectively, these three papers provide new evidence on how people in cities with poor air quality respond to information. There are a few points of broad agreement:
- Information does lead to behavior changes. Exactly how much behavior change and what types of information lead to changes vary across the contexts however.
- WTP for both information and improvements in air quality seem relatively low. Certainly in Delhi and Tbilisi the measured WTP was low. The story in Mexico City is a bit more mixed but it is not obvious that WTP is especially high.
- There are limits to the benefits of providing information. WTP remains low after information provision even if it may be responsive to information. Behavior changes are detectable in response to information provision but are, in many cases, relatively small.
There remain a number of important open questions that these studies individually and collectively raise:
- Two of these studies find that SMS alerts are effective to varying degrees. But what about the SMS alert matters? Is it the informational content of the message or the reminder feature (i.e. you get a daily reminder that pollution is bad)? What is the right frequency to send these alerts to maximize impact without overwhelming recipients?
- Why isn’t information provision, especially in settings where existing information is relatively sparse (i.e. Tbilisi), more effective in changing behavior? Is the scope for behavior change too constrained (i.e. by existing commuting patterns)?
- Why do people seem to be willing to pay for information on outdoor pollution (Mexico) even if it only leads to limited behavioral change (Tbilisi)? How generalizable are these two results to other cities?
Air pollution imposes large health costs and information provision is relatively cheap and scalable, which has generated a lot of interest globally in improving the infrastructure to monitor and provide information. These three studies suggest this could have benefits but that there are limits to how much can be achieved through information provision alone.
Understanding how the mode of delivery and informational content of the messages is important for making this new infrastructure effective. Real-time, repeated information seems to be valuable and might be effective. It also seems that focusing on pollutants or behaviors people can control—like indoor air pollution or smoking—may yield bigger health gains than focusing only on ambient pollution levels.
But there remains a great deal of space for research in this area that can improve our understanding of how to target these information interventions most effectively.
Join the Conversation