本研究开发了一种创新方法来预测中国上市公司的"漂绿"行为。通过将大型语言模型BERT整合到机器学习框架中，我们构建了一个先进的漂绿预测模型。这种方法能够捕捉企业社会责任报告和年度报告的环境披露中微妙的语言线索和语义细节，显著提高了识别漂绿的精确度。研究采用了多种机器学习模型，包括支持向量机（SVM）、随机森林（RandomForest）和随机欠采样算法（RUSBoost），并在三种不同的数据集上进行了测试：基础财务数据集、扩展的Word2Vec环境披露数据集，以及BERT优化的环境披露数据集。结果表明，RUSBoost算法结合BERT调整的环境披露数据在各项评估指标上表现最佳，凸显了先进自然语言处理技术在分析环境披露文本方面的优势。此外，我们的研究发现预测的漂绿指标与ESG评级机构间的评级分歧显著相关，验证了本研究所开发的漂绿变量。本研究为识别和预测企业漂绿行为提供了一个创新的、基于文本的方法。这一工具对投资者、监管者和政策制定者具有重要价值，有助于捕捉公司的欺骗性环境披露。

本文探究了长尾市场上的数字信贷利率形成问题，基于Hotelling模型从信息传输的视角提出该市场存在一种新型的垄断，并且动态地分析了这种垄断的加剧过程。本文主要发现有：第一，银行决策的最优信贷利率与其获得的借款人信息量负相关。第二，如果数字空间存在信息传输损耗，那么均衡利率高于Bertrand利率水平，为一种信息垄断利率，垄断溢价、垄断利润均与损耗系数正相关。第三，银行与借款人之间的交易次数越多，这种溢价也会越大，而且借款人之间的利率不平等问题也会加重；第四，部分借款人可能还会遭受歧视性定价。针对这些问题，本文提出了三项政策设计讨论，包括因信息垄断造成的垄断指标测度、风险指标测度以及最优的数字普惠考核政策，以破除数字鸿沟、缓解信息垄断，从而降低长尾市场均衡利率，保障数字金融的普惠性。本文最后对理论模型进行仿真，模拟主要结论。

We examine whether large language models (LLMs) can extract contextualized representation of Chinese public news articles to predict stock returns. Based on representativeness and influences, we consider seven LLMs: BERT, RoBERTa, FinBERT, Baichuan, ChatGLM, InternLM, and their ensemble model. We show that news tones and return forecasts extracted by LLMs from Chinese news significantly predict future returns. The value-weighted long-minus-short portfolios yield annualized returns between 35% and 67%, depending on the model. Building on the return predictive power of LLM signals, we further investigate its implications for information efficiency. The LLM signals contain firm fundamental information, and it takes two days for LLM signals to be incorporated into stock prices. The predictive power of the LLM signals is stronger for firms with more information frictions, more retail holdings and for more complex news. Interestingly, many investors trade in opposite directions of LLM signals upon news releases, and can benefit from the LLM signals. These findings suggest LLMs can be helpful in processing public news, and thus contribute to overall market efficiency.

This article applies natural language processing (NLP) to extract and quan- tify textual information to predict stock performance. Leveraging an exten- sive dataset of Chinese analyst reports and employing a customized BERT deep learning model for Chinese text, this study categorizes the sentiment of the reports as positive, neutral, or negative. The findings underscore the predictive capacity of this sentiment indicator for stock volatility, excess re- turns, and trading volume. Specifically, analyst reports with strong positive sentiment will increase excess return and intraday volatility, and vice versa, reports with strong negative sentiment also increase volatility and trading volume, but decrease future excess return. The magnitude of this effect is greater for positive sentiment reports than for negative sentiment reports. This article contributes to the empirical literature exploring sentiment anal- ysis and the response of the stock market to news on the Chinese stock market.

投资者对宏观经济风险的评估如何影响资产价格一直是实证资产定价的难点之一。已有研究指出新闻文本会改变投资者对宏观经济的判断和预期进而影响股价，但如何有效提取与宏观经济风险相关的文本叙事信息来解释或预测资产价格变动，尚未达成共识。本文基于2007-2021年中国七家专业财经媒体的新闻报道数据，首次结合人工智能前沿领域的BERT大语言模型来测度新闻叙事注意力信息，然后利用稀疏工具主成分（Sparse IPCA）估计影响基本面的状态变量和影响资产价格的叙事定价因子。基于A股市场的实证检验发现：第一，本文利用新闻文本估计的状态变量对于消费、产出、国债收益率等宏观经济指标具有显著的预测效果，这表明新闻叙事蕴含着影响经济运行的信息。第二，相比CAPM、三因子等基准模型，基于新闻文本构建的叙事因子模型能更好地解释资产错误定价现象，并对未来资产价格的变化有更强的预测能力。第三，与基于关键词的文本分析方法（如LDA主题模型）相比，利用BERT提取文本信息可在保证因子模型简约性的基础上获得更优异的定价效果。本文的研究结论对于解释资产横截面收益差异有新的启示，同时为应用大语言模型于经济金融学研究抛砖引玉。

This article applies natural language processing (NLP) to extract and quan- tify textual information to predict stock performance. Leveraging an exten- sive dataset of Chinese analyst reports and employing a customized BERT deep learning model for Chinese text, this study categorizes the sentiment of the reports as positive, neutral, or negative. The findings underscore the predictive capacity of this sentiment indicator for stock volatility, excess re- turns, and trading volume. Specifically, analyst reports with strong positive sentiment will increase excess return and intraday volatility, and vice versa, reports with strong negative sentiment also increase volatility and trading volume, but decrease future excess return. The magnitude of this effect is greater for positive sentiment reports than for negative sentiment reports. This article contributes to the empirical literature exploring sentiment anal- ysis and the response of the stock market to news on the Chinese stock market.

本文基于不完全信息Stackelberg 寡占模型，首次将博弈研究用于分析Shibor 存款利率定价。 在假设部分报价行之间存在合谋定价的基础上，分别分析三种市场战略结构中银行的报价行为。研究显示： 在Bertrand 双寡头市场结构中，报价行间合谋定价操纵Shibor 存款利率的行为不可能实现。而在Stackelberg 寡占市场结构中，参与合谋报价的银行利润可以显著的提高。由此得知，尽管群体操纵Shibor 存款利率定 价的行为不会存在，但个别报价行确实存在通过控制Shibor 存款利率定价操纵市场的机会。此外，参与合 谋的报价行拥有存款供给对Shibor 存款利率调整敏感性的信息优势，通过影响未参加合谋报价行预期的策 略，可以获得更多的均衡利润，挤占未参加合谋报价行的市场份额。有鉴于此，文章最后提出了预防并限 制报价行违规行为的若干对策。

本文通过比较双头寡占下价格竞争的纯粹寡占模型和混合寡占模型，讨论了存款保险和产权结构对银行竞争的影响。研究发现：(1)外部风险控制不能代替银行内部风险控制。(2)风险调整的存款保险制度依然可能会引起道德风险。但风险调整存款保险制度与固定费率存款保险制度相比，会降低道德风险。(3)固定费率存款保险制度中风险对利率的敏感性小于风险调整存款保险制度。(4)当银行贷款利率和存款保险制度设计给定时，产权结构不影响银行的风险选择。

Based on the background of the banking industry in China, this paper establishes Cournot, Bertrand and Stackelberg mixed oligopoly competition models with deposit and interest rates as strategic variables between a representative state-owned bank and a representative foreign bank. We discuss and compare the equilibrium deposits, interest rates and profits in different market structures. More importantly, considering the endogenous timing setup and taking the extended game with observable delay as the basic model, we analyze the competition results of the mixed duopoly at different market structures and make numerical simulations in order to get the outcomes of the extended game. It is found that, in the deposit competition with endogenous timing, if two banks have the same deposit return rate, the simultaneous move could become the final result; If the deposit return of the foreign bank is more than twice that of the state-owned bank, all SPNEs of the extended game are sequential, i.e one bank will lead, the other will follow. If the state-owned bank’s profit in Cournot is larger than that of the state-owned bank as the Stackelberg follower, the only SPNE will be (E, L), i.e. the state-owned bank will lead and the foreign bank will follow. Additionally, in the interest rate competition with endogenous timing, if the degree of nationalization is no larger than 75% as well as two banks have the same deposit return rate, the final SPNEs of the extended game are sequential and the simultaneous move can not become the SPNE.

Based on the background of the banking industry in China, this paper establishes Cournot, Bertrand and Stackelberg mixed oligopoly competition models with deposit and interest rates as strategic variables between a representative state-owned bank and a representative foreign bank. We discuss and compare the equilibrium deposits, interest rates and profits in different market structures. More importantly, considering the endogenous timing setup and taking the extended game with observable delay as the basic model, we analyze the competition results of the mixed duopoly at different market structures and make numerical simulations in order to get the outcomes of the extended game. It is found that, under the scenario of endogenous timing, and 1) the assumption that the foreign bank’s deposit return rate is more than twice that of the state-owned bank and 2) the degree of nationalization of state-owned bank is no less that 1/4, the SPNE of deposit extended game is (L, L), i.e. both banks will choose to move later, neither player has so called “first mover advantage” which leads to the Cournot outcome and payoffs. When the degree of privatization is more than 3/4 and both banks have the same deposit return rate, the SPNE of interest rate extended game is (E, L), i.e. the state-owned bank will lead and the foreign bank will follow.