We exploit a quasi-natural experiment in law-violation behaviors in response to the Clean Winter Heating Policy (CWHP) implemented in northern China from 2017 to 2019. Our results show that treated villagers were more likely to violate laws against burning agricultural waste and exhibit less prosocial behaviors in both incentivized dictator games and public goods games. The results hold for both the conventional Difference-In-Differences analysis and the newly proposed weighted CATT estimates. We find two factors that likely play a part. First, the CWHP was perceived as a negative income shock. Thus, in response, villagers would want to cut down expenditures on the disposal of corn stalks and straw and behave more selfishly in incentivized games. Second, the CWHP led to discontent and provoked farmers to burn straw. Additional evidence suggests that such law violations and less prosocial behaviors could have been avoided by granting more generous subsidies up front.

Understanding users' context is essential for successful recommendations, especially for Online-to-Offline (O2O) recommendation, such as Yelp, Groupon, and Koubei. Different from traditional recommendation where individual preference is mostly static, O2O recommendation should be dynamic to capture variation of users' purposes across time and location. However, precisely inferring users' real-time contexts information, especially those implicit ones, is extremely difficult, and it is a central challenge for O2O recommendation. In this paper, we propose a new approach, called Mixture Attentional Constrained Denoise AutoEncoder (MACDAE), to infer implicit contexts and consequently, to improve the quality of real-time O2O recommendation. In MACDAE, we first leverage the interaction among users, items, and explicit contexts to infer users' implicit contexts, then combine the learned implicit-context representation into an end-to-end model to make the recommendation. MACDAE works quite well in the real system. We conducted both offline and online evaluations of the proposed approach. Experiments on several real-world datasets (Yelp, Dianping, and Koubei) show our approach could achieve significant improvements over state-of-the-arts. Furthermore, online A/B test suggests a 2.9% increase for click-through rate and 5.6% improvement for conversion rate in real-world traffic. Our model has been deployed in the product of "Guess You Like" recommendation in Koubei.

Online individual fundraising has been a prominent component of online charitable giving due to the low cost of online solicitation. While studies of offline charity have shown that solicitation under peer pressure leads to higher donation, it is unclear whether this fundraising effect still holds online given several significant differences between offline and online settings. In this paper, we first examine whether online individual fundraising donors give more and what factors moderate such an effect. We found that in general this is true: An individual fundraising donor gives on average ¥6 more than an organic donor. However, this effect is overturned for anonymous donors, where the organic donors actually give more than those in individual fundraising. In addition, the effect varies along the life cycle of a project or individual fundraising, with the early and later donors showing greater effect than those coming in the middle of the life cycle. Next, we examine the interaction between individual fundraising and organic donation through information sharing. Counter-intuitively, we found that blocking organic donors’ access to individual fundraising will increase donations for most projects, with the aggregated gain across projects in our sample to be 4.6%. We provide explanations for our findings.

Does receiving a gift encourage the recipient to send more gifts? Causal identification of behavioral contagion is very challenging, especially based on observational data from social networks. Given that the online monetary gift (also known as a packet) in WeChat groups is randomly split between group members, we are able to conduct a natural experiment to identify the causal effect of receiving online red packets on the contagion of gift exchange. Analyzing gifting behavior using a large-scale dataset of 3.4 million WeChat users, we find that recipients on average repay 18.29% of the amount they receive. Moreover, our analysis shows that recipients' contagion may be driven by different types of reciprocity. We further find that the draw recipients repay 1.5 times more than other recipients, and identify a group norm that luckiest draw recipients should send payment back to the group.

Does receiving a gift encourage the recipient to send more gifts? Causal identification of behavioral contagion is very challenging, especially based on observational data from social networks. Given that the online monetary gift (also known as a "red packet") in WeChat groups is randomly split between group members, we are able to conduct a natural experiment to identify the causal effect of receiving online red packets on the contagion of gift exchange. Analyzing gifting behavior using a large-scale dataset of 3.4 million WeChat users, we find that recipients on average repay 18.29% of the amount they receive. Moreover, our analysis shows that recipients' contagion may be driven by different types of reciprocity. We further find that the "luckiest draw" recipients repay 1.5 times more than other recipients, and identify a group norm that luckiest draw recipients should send payment back to the group.

In this paper, we describe a laboratory experiment designed to investigate the behavioral motives for generalized indirect reciprocity. This refers to the behavior in which, upon receiving a transfer from a donor, the recipient is willing to repay a third party beneficiary in a one-shot interaction. We found that the recipients’ repayment to the beneficiaries increased significantly when beneficiaries and donors had prior social connections, especially when the donor was generous initially. However, a structured message from the donors asking for favorable treatment of the beneficiaries did not affect the recipients’ transfers. Among major theories of reciprocity, altruism with an endogenous reference group was shown to be most promising in explaining the experimental results.