Abstract: The recent decades have seen a growing number of dualmode platforms such as Amazon and Google Dualmode platforms may favor their own products in the downstream market through search result display，algorithmic recommendations，etc，and discriminate against downstream market competitors Thus，selfpreferencing has become an important policy concern in the antitrust of the platform economy At present，antitrust economics research on the selfpreferencing behavior of dualmode platforms has drawn different views Therefore，how to scientifically recognize the competitive effects of selfpreferencing and how to adopt optimal antitrust remedy policies have become the focus of the issue
This paper focus on exploring the behavioral incentives and competitive effects of digital platforms selfpreferencing via algorithmic recommendations and examining the effectiveness of policies that restrict selfpreferencing The main findings are as follows A monopoly platform will only have the incentive to engage in selfpreferencing when the increase in sales revenue of its own products due to selfpreferencing exceeds the loss in commission An important market structural condition for digital platforms to implement selfpreferencing is that they have strong relative market power on the seller side Selfpreferencing can increase digital platforms market power in downstream markets and induce market foreclosure to competitors However，its impact on consumer surplus and social welfare is uncertain，mainly affected by search costs and commission rates In terms of the effects of different remedy policies，structural separation，neutrality regulation and data isolation all can promote market competition，but their implications on consumer surplus and social welfare are significantly different，and all have certain economic costs 
The policy implications of this paper are： First，the antitrust regulation of digital platforms selfpreferencing via algorithmic recommendations should use reasonable principles to analyze case by case Second，regulation of selfpreferencing should first be subject to a market structure review，with a focus on the platforms market power in the upstream intermediary service market Third，the antitrust review should adopt a threestep approach of “capability  incentive  competition damage” Fourth，the antitrust remedy policy should be selected based on the economic facts of individual cases Fifth，when designing antitrust policies for digital platforms selfpreferencing in China，it is necessary to establish an organic connection between exante regulatory policies and expost enforcement policies and to achieve the best combination of scientific review and effective remedy

Key words: digital platform, algorithmic recommendations, self-preferencing, antitrust policy, remedy policy

摘要： 双重模式数字平台利用上游市场垄断势力优待下游自有产品或服务的自我优待行为是数字平台反垄断尚不明确的新问题，如何认识自我优待行为的竞争效应以及如何采取最优的反垄断救济政策成为问题的焦点。本文重点分析数字平台实施算法推荐自我优待的行为激励、竞争效应及最优反垄断救济政策选择。分析结果显示，在商家侧具有较强的相对市场势力是数字平台实施自我优待的重要市场结构条件，垄断数字平台实施自我优待会提高其在下游市场的市场势力，并产生对竞争对手的市场封锁，但其对消费者福利和社会总福利的影响是不确定的，主要受到搜寻成本和佣金率的影响。从政策的效果来说，结构分离政策、中立政策和数据隔离政策都有助于维护市场竞争，但他们对消费者剩余和社会总福利的影响存在明显差异，并且都具有一定的政策实施经济代价。因此，对算法自我优待行为的反垄断审查应基于个案的经济事实采用合理推定原则进行分析，并考虑不同救济政策的适用条件来选择最佳救济政策。

关键词: 数字平台, 算法推荐, 自我优待, 反垄断政策, 救济政策