近年来，随着互联网技术的发展与普及，游戏直播市场发展迅猛并已经成长为了具有相当规模的数字娱乐产业。然而，当前业界与学界对这种新兴数字娱乐市场中主播开播与用户的参与行为影响因素的理解却非常有限。本论文分别从推荐系统和限制未成年人网络游戏政策两个角度对于这一话题进行实证探究。本文的第一部分的研究旨在理解推荐系统的存在如何影响游戏直播市场的供给结构以及其中的影响机制。我们通过观测到一次作用于中国泛娱乐直播行业的外生性政策冲击以及来自直播市场中主播群体的开播行为数据来研究这一问题。研究发现：（1）推荐系统的存在显著降低游戏直播市场的供给集中度。（2）原有供给结构更集中的子类别在推荐系统的影响下，供给集中度将降低得更多。进一步地，（3）推荐系统的存在会使得粉丝订阅数小的主播开播频率提升更多，表明市场供给集中度的降低正是由这些中小主播开播行为的变化所驱动。（4）推荐系统的存在会使得粉丝订阅数小的主播收益提升更高，这一结果同样支持了上述解释。本文的第二部分的研究旨在理解限制未成年人网络游戏政策将对游戏直播市场中的用户参与和主播开播行为产生什么样的影响。我们结合来自主播个体层面的行为数据与一次未成年人被严格限制玩网络游戏的外生政策冲击开展研究。研究发现：（1）限制未成年人网络游戏政策降低了游戏直播市场的整合活跃观众数，这一结果支持了未成年人被限制玩网络游戏后，整体上降低了对游戏的兴趣。（2）政策虽然总体上不显著影响主播的开播频率，但主播体量负向调节政策对主播开播频率的影响。（3）对于主播决定开播的场次而言，政策总体上正向影响主播的每场活跃观众数，而主播体量则正向调节这种影响。此外，我们的研究同样调查了不同日期政策严格程度的异质性对政策影响的调节作用，并进一步围绕弹幕数、打赏额、主播每日直播时长等因素进行了延伸分析。

Online game livestreaming has experienced a phenomenal development and grown into a substantial industry led by the development and wide usage of the internet technology in recent years. However, there is limited understanding of the factors that affect the streamers’ supply and users’ engagement behaviors for this emerging digital entertainment market. This paper empirically studies this issue from the perspectives of the presence of recommender systems and the policy of preventing minors from online gaming.The first part of this thesis focuses on understanding how the presence of recommender systems affects the supply structure of the game livestreaming market and the underlying mechanism. We investigate this issue with an exogenous policy shock and detailed supply behavior data in the game livestreaming market. Our main findings include: (1) The presence of recommender systems significantly reduces the supply concentration. (2) Subcategories with a more concentrated original supply structure are more likely to have a diverse supply structure due to the presence of recommender systems. (3) Furthermore, the presence of recommender systems incentivizes streamers with lower popularity to supply more frequently, suggesting that the impact of the presence of the recommender systems on the supply concentration is driven by these smaller streamers’ activities. (4) The presence of recommender systems also results in higher revenue for streamers with lower popularity, further supporting the explanation above.The second part this thesis aims to understand how the online game restriction policy for minors affects the streamers’ supply and users’ engagement behaviors in the game livestreaming market. In this study, we combine the behavior data of streamers and an exogenous policy shock which restricts minors‘ access to online games. Our main findings include: (1) The restriction policy reduces the aggregate number of active audience in the game livestreaming market, which supports the assumption that minors lose interest in online games after they are prevented from playing online games. (2) While the policy does not significantly affect the overall supply frequency of streamer, the streamer’s size negatively moderates the policy‘s impact on the supply frequency. (3) For the sessions that streamers offered, the policy has a positive effect on the number of active audience per session in general, with the streamer’s size positively moderating this effect. We further investigate the moderating effect brought by the heterogeneity of policy strictness at different dates and also conduct extended analyses on the variables of interest including the number of follow-ups, the amount of tip and livestreaming duration.