Biased Recommendation with Superior Knowledge and Implications of Data Compliance

Description

By learning superior consumer information (preference knowledge that exceeds consumers' own prior) from both behavioral and personal data, an intermediary platform such as Netflix can facilitate product matching by personalizing product recommendations. However, the recommender may have biased incentives to steer consumers into accepting a less relevant but more profitable product, which may induce consumer suspicion that leads to rejection and deadweight welfare loss. In this project, we aim to examine the non-pricing economic implications of superior knowledge by building the micro-foundation of personalized recommendations with three building blocks: cheap-talk model, biased incentives, and consumer profiling. The consumers' intrinsic preference match value is correlated with both the behavioral data and personal data. But the consumers only learn their personal data, and the behavioral data is accessable only by the firm. Specifically, we plan to examine three compliance rules, under which the recommender is permitted to personalize recommendations using (1) only the personal data, (2) only the behavioral data, or (3) both personal and behavioral data. We will analyze the optimal recommendation strategy under each information structure and then make static comparisons to examine the welfare implications of the compliance rule. Our preliminary results show that in equilibrium, more consumer data may not necessarily imply more accurate personalization. Therefore, the optimal recommendation strategies require formal examination. This project may make theoretical contributions by extending the classic signaling games to partial information exchange with superior knowledge. In addition, this project aims to help consumer advocates and regulators better understand the welfare implications of the compliance rules, and thus help characterise the optimal data usage policies by rules of reasons.