Operational Decision-Making for Reach Maximization of Incentive Programs that Influence Consumer Energy-Saving Behavior

Suny At Buffalo, Amherst NY

Investigators

Abstract

Environmental programs that exploit informational social influence have recently found success in fostering energy-conscious consumer behavior. By disseminating Home Energy Reports that compare individual households' energy spending patterns to those of their neighbors, these programs capitalize on the fact that people are more likely to adjust their behavior when simple information on its negative impacts on the environment is preceded by ''social proof'' that referent others have already started to behave pro-environmentally. While exploiting the power of social influence for the societal benefit is a breakthrough in itself, further research is required to boost the impact of such programs with direct incentives, so that they can be proactively controlled and their impact accelerated in an organized manner. This award supports fundamental research into the design and planning of randomized incentive programs -- social lotteries -- that inform consumers of their peers saving energy while being rewarded for doing so. The researched prescriptive methodologies enable calculated maximization of the ''reach'' of a social influence program or policy, and are expected to be applicable in economics, healthcare, education, etc., for increasing individual awareness of issues of societal concern. This project will further support the educational mission of the NSF by involving members of underrepresented student groups, senior design project advising, and conducting undergraduate student seminars. The research plan encompasses the modeling efforts to quantitatively capture the connection between the informational social influence and the impact of the randomized energy-saving incentive program; the development of mixed-integer programming, dynamic programming and decentralized network optimization algorithms for reach maximization under budgetary constraints; and the investigations into operational program planning under limited information about a targeted heterogeneous population. This project will produce new classes of problems and fundamental insights into operational planning of interventions exploiting social networks. Its products will allow for modeling of the mechanisms of social influence driven not directly by diffusion, as previously done, but by reach. The developed designs, recipes and tools will be evaluated with the data and advice provided by a collaborating energy provider. The improvements resulting from the introduction of the researched formalisms and algorithms will be characterized by gathering theoretical evidence, en route to preparing a pilot implementation.