近日，9999js金沙老品牌百人计划研究员袁哲关于量化研究数据价值、算法经济影响的两篇前沿研究论文，分别在国际公认的管理学顶级期刊《Management Science》和《Information Systems Research》发表。其中袁哲研究员作为通讯作者，与长江商学院科技与运营教授孙天澍、阿里巴巴集团李春晓、徐俊、张凯夫等关于数据价值与数据隐私的合作论文“The Value of Personal Data in Internet Commerce: A High-Stakes Field Experiment on Data Regulation Policy” 发表于《Management Science》，袁哲研究员作为第一作者，与英属哥伦比亚大学陈塬、阿里巴巴集团王一同、长江商学院科技与运营教授孙天澍等合作研究算法影响消费者行为的文章“How Recommendation Affects Customer Search: A Field Experiment”发表于《Information Systems Research》。

“The Value of Personal Data in Internet Commerce: A High-stake Field Experiment on Data Regulation Policy” 论文摘要 

Personal data has become a key input in Internet Commerce, facilitating the matching between millions of customers and merchants. Recent data regulations in China, Europe and US restrict Internet platforms' ability to collect and use personal data for personalized recommendation and may fundamentally impact Internet Commerce. In collaboration with the largest E-commerce platform in China, we conduct a large-scale field experiment to measure the potential impact of data regulation policy, and to understand the value of personal data in Internet Commerce. For a random subset of 555,800 customers on Alibaba platform, we simulate the regulation by banning the use of personal data in the homepage recommendation algorithm and record the matching process and outcomes between these customers and merchants. Compared to the control group with personal data, we observe a significant higher concentration in the algorithmic recommendation of products in the treatment group and a very sharp decrease in the matching outcomes as measured by both customer engagement (clickthrough rate and product browsing) and market transaction (sales volume and amount). The negative effect is disproportionate and more pronounced for niche merchants and customers who would benefit more from E-commerce. We discuss the potential economic impact of data regulation on Internet Commerce, as well as the role of personal data in generating value and fostering long-tail innovations.

个人数据已成为互联网商务的关键要素，促进了数百万客户和商家之间的匹配。中国、欧洲和美国最近的数据法规限制了互联网平台收集和使用个人数据进行个性化推荐的能力，并可能从根本上影响互联网商务。我们与中国最大的电子商务平台合作，进行了大规模的现场实验，以衡量数据监管政策的潜在影响，并了解个人数据在互联网商务中的价值。基于阿里巴巴平台上的555,800 个客户的随机样本，我们通过禁止在主页推荐算法中使用个人数据来模拟监管，并记录这些客户和商家之间的匹配过程和结果。与拥有个人数据的对照组相比，我们观察到治疗组中产品的算法推荐集中度显着更高，并且通过客户参与度（点击率和产品浏览）和市场交易衡量的匹配结果急剧下降（销量和金额）。对于那些能从电子商务中受益更多的长尾商家和客户来说，产生了更大的负面影响。我们讨论数据监管对互联网商务的潜在经济影响，以及个人数据在创造价值和促进长尾创新方面的作用。

Link: https://pubsonline.informs.org/doi/10.1287/mnsc.2023.4828

“How Recommendation Affects Customer Search: A Field Experiment”论文摘要

Product recommendation and search are two technology-mediated channels through which E-commerce platforms can help customers find products. However, the relationship between the two channels and the underlying mechanisms and implications for platform design are not well understood. We leverage a randomized field experiment with 555,800 customers on a large E-commerce platform to investigate how product recommendation affects customer search. We vary the relevance of the recommendation that users experience upon arriving at the homepage of the platform and find that a decrease in recommendation relevance leads to a significant increase in consumers’ use of the search channel, indicating a (partial) substitution effect between the two at the aggregate level. We find substantial heterogeneity across product categories, propose a conceptual framework, and theorize how different states of customer demand—demand fulfillment and demand formation—may drive such heterogeneity. The results are aligned with our framework and provide evidence that both demand formation and fulfillment are at work in the channel interactions between recommendation and search. Specifically, when customers receive more product recommendations in a category, they search more in that category with generic query words, which indicates complementarity between recommendation and search. However, when customers receive fewer product recommendations in a category of interest, they compensate for this reduction by searching more in that category with long-tail query words, which indicates a substitution between recommendation and search. This experimental study is among the first to examine the causal relationship between the recommendation channel and search channel and offers implications for the design of E-commerce platforms.

产品推荐和搜索是电子商务平台帮助客户找到产品的两个以技术为媒介的渠道。然而，这两个渠道之间的关系以及底层机制和对平台设计的影响尚不清楚。我们利用大型电子商务平台上555,800 名客户的随机现场实验来研究产品推荐如何影响客户搜索。我们改变了用户到达平台首页时体验到的商品推荐的相关性，发现推荐相关性的降低导致消费者对搜索渠道的使用显着增加，表明搜索和推荐两个渠道总体上存在替代效应。我们发现这个影响在不同产品类别之间存在显着的异质性，并且提出了一个概念框架，并对客户需求的不同状态（需求满足和需求形成）如何驱动这种异质性进行了理论分析。实证结果与我们的框架一致，并提供证据表明需求的形成和满足在推荐和搜索之间的渠道交互中起作用。具体来说，当客户在某个类别中收到更多的产品推荐时，他们会更多的使用通用的搜索词在该类别中搜索，这表明推荐和搜索之间存在互补性。然而，当客户在他们原本感兴趣的类别中收到较少的产品推荐时，他们会通过更多的使用长尾搜索词在该类别中搜索来补偿这种减少，这表明推荐和搜索之间的替代性。这项实验研究是第一个检验推荐渠道和搜索渠道之间因果关系的实验研究，并为电子商务平台的设计提供了启示。

Link: https://pubsonline.informs.org/doi/10.1287/isre.2022.0294

作者简介：

袁哲博士，9999js金沙老品牌百人计划研究员，博士生导师。袁哲毕业于多伦多大学（博士）和北京大学（本科），哈佛大学、威斯康辛大学访问学者。袁哲的研究领域是算法与人工智能的经济影响、平台经济和产业经济。他的论文发表于American Economic Journal: Microeconomics, Management Science，Information Systems Research等国际顶级刊物。他主持国家青年自科项目，浙大文科青年交叉创新团队，参与多项国家自科重大和重大专项项目。他在多个国际顶级人工智能和平台相关的会议上宣讲，并获得奖项。