In this paper, I study the effectiveness of personalized nudges that are sent on a money management app from a major commercial bank in Canada. These fintech nudges are personalized in a sense that they are sent to the users of the app when their spending on the previous day exceeds twice their own daily average spending over the past 12 months. In a nutshell, the paper compares the behavior of message recipients to that of non-recipients in the subsequent period.
I find that recipients of an overspending message are estimated to spend C$8.15, which is 5.4% of daily average spending, less than non-recipients, on the next day of receipt. They most significantly reduced spending on the shopping category, which is the most evident form of discretionary spending. The messages resulted in permanent reduction in cumulative spending, and the effects of messages are more pronounced for individuals who are relatively older, have higher liquid wealth, are more financially savvy, are new to the app experience, or reside in a city with higher education levels.
In addition, I document suggestive evidence that there is spillover of fintech nudges from one user to another in the same household. In other words, two individuals in a family inform each other of their financial situations once either spouse receives a message, as a result of which they reduce spending together. The magnitude of the spillover effects is almost as great as that of direct effects, which suggests that literature on nudges may have underestimated the total effect of nudges, as the previous papers only investigate the effect on individuals who have received direct intervention.
However, overspending messages resulted in one unintended result. Message recipients are less likely than non-recipients to log in to the app in the future. This tendency to neglect adverse financial information is known as the ostrich effect. Hence, one policy implication is that the types of unintended behavior of the participants must be taken into consideration as far as designing effective nudges is concerned.
Read the full paper here.