学术论文

人们的生活作息规律是由生理时钟和社交时钟共同构建而成。不过，人们的社交时钟经常偏离生理时钟（例如：为了工作或学习，比睡到自然醒早起；为了夜班工作或赶上项目完成期限而熬夜）— 此现象被称为「社交时差」。社交时差如何影响消费者行为？研究的工作数据和实验结果显示，由于处于社交时差的消费者往往对社交互动不太感兴趣，因此社交时差会减少炫耀性的消费行为。然而，此效应在以下情况会有所减弱：当消费者与熟悉的人而非陌生人进行社交互动；当他们的炫耀性消费行为不会引起社交圈子的关注；以及当他们倾向于私人场合中使用奢侈品。

虽然先进的机器学习算法大大提升了金融欺诈检测研究的发展，然而，在构建有效辨识欺诈行为特征（或属性）方面仍然存在挑战。近年，在人们可以于没有金融中介的情况下向他人贷款的P2P 借贷平台上，出现了一种新型欺诈，因为信息不对称的问题在这些借贷市场更加严重。我们以舞弊三角论及其引伸的理论为基础，利用设计科学研究方法，除了参考借款人和贷款请求的基本特征以外，更直接以P2P 借贷交易数据构建了五类新的行为特征。这些行为特征，以研究欺诈者的贷款请求和还款记录、人际网络、投标过程的特征和活动次序为基础，并旨在评估其进行欺诈的能力、诚信和欺诈机会。我们使用涵盖中国两个大型 P2P 借贷平台的真实用户的数据集进行评估。研究结果显示，结合新的行为特征与基本特征可显著提升金融欺诈检测的表现。是项设计科学研究为金融欺诈检测的文献和实践带来了新见解。

我们调查了一个评论平台上用专家评论来补充用户评论的编辑评论项目。具体来说，我们研究了编辑评论是否以及如何影响后续用户评论（那些由非专业编辑撰写的评论）。基于一个在亚洲领先的评论平台上发生的准实验，我们通过计量分析和自然语言处理技术，发现了一些实证证据证明编辑评论对后续用户评论有一个整体的正面影响。首先， 收到编辑评论的餐厅可以吸引更多用户评论。此外，这些增加的用户评论在保持对不同主题探讨的同时，还会更多的讨论实质性主题。这增加了后续用户评论内容的长度和多样性。在情感分析的维度上，这些新增的用户评论更加中立且给出的评分也较低。我们的机制分析显示， 收到编辑评论的餐厅收到的后续用户评论， 不论在主题，情感或评分，篇幅长度和可读性等各方面，均与编辑评论变得更加相似。我们指出羊群效应是编辑评论影响后续用户评论行为的主要机制。聚焦长期存在于这个评论平台的用户， 我们进一步证实了羊群效应的存在。我们的研究结果表明评论平台可以通过编辑评论项目来提高用户评论的质与量。

This study examines how the market share of dark venues changes at earnings announcements. Our analysis shows a statistically significant increase in dark market share in the weeks prior to, during, and following the earnings announcement. We also predict and find evidence that increases in dark market share around earnings announcements are higher for firms with high quality accounting information. In addition, we find a positive relation between the change in dark market share and the speed of resolution of investor disagreement-a key dimension of informational efficiency, which suggests that dark trading is associated with an improvement in market quality. How market fragmentation changes around news events, the role accounting information plays in market fragmentation, and how changes in market fragmentation relate to market quality can help provide insights to securities regulators.

This paper presents an economic framework to study strategic interactions along the analyst-auditor-owner disciplinary chain, in which the auditor examines the financial reports prepared by the owner, and the analyst uncovers financial misreporting as well as audit failure. We find that although analyst scrutiny ex post detects misreporting, it ex ante aggravates the owner's misreporting behavior and further impairs financial statement reliability if the legal penalties for the auditor and the owner are small. We also show how the effects of a regulation depend on its target's disciplinarian(s). Specifically, (i) although enhancing the auditor's legal liability always increases audit quality and financial statement reliability, it decreases investment efficiency if and only if the analyst is highly independent; and (ii) increasing the owner's misreporting penalty decreases investment efficiency if and only if either of (but not both) the regulations on the auditor and the analyst is strict.

This paper studies whether and how mandatory nonfinancial disclosure affects firms’ real decisions. I exploit a disclosure regulation enacted in California, which mandates that firms disclose how they conduct due diligence to address their suppliers’ human rights abuses. I find that treated firms increase their supply chain due diligence, and their suppliers’ human rights performance improves following the regulation. The effects are stronger when firms face greater pressure from non-governmental organizations (NGOs) and socially conscious shareholders, when customers have greater incentives to use the newly disclosed information, and when the regulation leads to a larger increase in information comparability. Collectively, the results suggest that mandatory nonfinancial disclosure can affect firms’ real decisions through market mechanisms and that stakeholder responses play a key role.

为用户提供主动匹配搭建平台的企业在提高匹配效率和保证匹配质量方面面临着巨大的挑战。本文研究了当用户从匹配的垂直属性（即质量）及其水平属性（即异质性匹配）中获得效用时，信息设计如何用于改善匹配结果。我们考虑一个博弈论模型，其中互相竞争的发送者向互相竞争的接收者提出匹配请求，并且两侧的用户在水平和垂直方向上都有差异。我们首先证明用户对垂直属性的偏好会加剧竞争并损害匹配效率，并且为了避免竞争，发送者可能会从近距离接收者切换到远距离接收者，即便对匹配的水平接近度更加看中。其次，我们研究了四种信息设计，每种设计隐去来自市场一侧的一种类型的信息。隐去任何一侧的垂直信息的设计增加了匹配的数量，这其中隐去接收者信息的改进更大。相比之下，隐去任何一侧水平信息的设计可能导致所有请求都集中在一个接收者上，并导致最严重的匹配失败。第三，匹配数量的增加是以牺牲某些用户福利为代价的，因为隐去一侧的垂直信息不仅会伤害双侧的高质量用户，也会伤害另一侧的低质量用户。尽管在某些条件下隐去一侧的水平信息可能会增加匹配量，但隐去一侧的垂直信息的设计可以做到帕累托改进。第四，在配过程中用户的自行策略定价不仅重新分配了用户福利，而且纠正了匹配失真。最后，与没有用户定价的结果对比，当用户可以策略定价时，平台隐去一侧的纵向信息可以使另一侧的所有用户受益，而隐去一侧的横向信息可以使同一侧的所有用户受益。

我们建立一个数据销售模型来研究另类数据对金融市场的影响。投资者需要特别的技术以准确分析购买的原始数据，但建立这项技术成本高同时存在相当大的不确定性。数据供应商透过控制数据样本数量去影响投资者从购买的数据中提取信息的准确性。我们的模型分析发现数据分析技术的成本对资本成本以及资产收益波动率的影响均呈U型关系，但是对市场信息量则呈驼峰型关系。同时，数据分析的技术平均水平和不确定性亦出现类似影响。我们的分析同时发现使用另类数据的基金和数据行业存在着相互促进的关系。

This paper studies how globalization affects the corporate tax policies of U.S. manufacturing firms. Using U.S.-granting China Permanent Normal Trade Relations as a quasi-natural experiment, we find a significant increase in tax reduction activities for firms facing higher exposure to Chinese imports. The effect is more pronounced for firms with higher managerial slack. We also find that the effect is stronger for firms in less diversified products market and faster changing industries. We also show that U.S. firms facing higher Chinese import competition are more likely to engage in other tax-motivated activities: acquisition of subsidiaries in low-tax regions and suspected transfer pricing. Furthermore, we explore the 2017 tax cut and the recent U.S.-China trade dispute and find that firms engage less in tax reduction activities after the 2017 tax cut and after the tariff increase for Chinese imports.