Comparative study of deep learning models for analyzing online restaurant reviews in the era of the COVID-19 pandemic

Online reviews remain important during the COVID-19 pandemic as they help customers make safe dining decisions. To help restaurants better understand customers’ needs and sustain their business under current circumstance, this study extracts restaurant features that are cared for by customers in current circumstance. This study also introduces deep learning methods to examine customers’ opinions about restaurant features and to detect reviews with mismatched ratings. By analyzing 112,412 restaurant reviews posted during January-June 2020 on Yelp.com, four frequently mentioned restaurant features (e.g., service, food, place, and experience) along with their associated sentiment scores were identified. Findings also show that deep learning algorithms (i.e., Bidirectional LSTM and Simple Embedding + Average Pooling) outperform traditional machine learning algorithms in sentiment classification and review rating prediction. This study strengthens the extant literature by empirically analyzing restaurant reviews posted during the COVID-19 pandemic and discovering suitable deep learning algorithms for different text mining tasks.     

个人、机构投资者情绪与证券分析师的乐观偏差——来自中国A股市场的证据

证券分析师的盈利预测存在着显著的乐观偏差,且这种乐观偏差会因投资者情绪的变化而忽高忽低。基于开户数量构建个人、机构投资者情绪指标,实证研究投资者情绪对证券分析师乐观偏差的影响。研究发现,个人、机构投资者情绪越乐观,证券分析师的乐观偏差越大;机构投资者情绪的波动比个人投资者更为剧烈,且对证券分析师的影响也大于个人投资者;在考虑了公司的截面特征后,结论依然稳健。研究有助于进一步理解新兴市场中证券分析师的盈利预测行为,并为投资者使用证券分析师的研究报告提供决策参考。     

企业生命周期、融资方式与融资约束——基于投资者情绪调节效应的研究

本文放松企业资金需求同质性假设,基于企业不同生命周期的组织特征、融资需求、融资能力差异,以不同生命周期企业融资约束差异作为切入点,融入投资者非理性情绪对企业外部融资环境的影响,动态考察投资者情绪变化对不同生命周期企业融资选择及融资约束缓解效应的影响。研究发现:不同生命周期企业融资约束状态不同,利用高涨投资者情绪缓解融资约束的程度及途径均存有差异:成长期企业融资约束程度最大,利用投资者情绪择时融资以缓解融资约束的程度也最强,衰退期企业次之,成熟期企业最小;高涨的投资者情绪改变了企业的外部融资环境和相对成本,各生命周期企业均会利用投资者情绪变化选取最适宜自身的融资方式:处于生命周期各阶段的企业均会利用投资者的高涨情绪进行信贷融资以缓解融资约束,而相对衰退期企业,成长期企业更偏好利用股权融资缓解融资约束,成熟期企业更偏好利用债券融资缓解融资约束。因此,投资者情绪不仅对生命周期影响企业融资约束具有调节效应,而且对企业生命周期影响融资方式的选择也具有调节效应。     

本地偏好、投资者情绪与股票收益率:来自网络论坛的经验证据

本文以投资者在东方财富网股吧针对创业板上市公司发表的90多万条帖子为研究对象,通过IP地址识别,构建本地关注指标;并利用计算机文本挖掘技术,提取网络发帖所体现的情绪倾向,构建投资者情绪指标。实证结果显示,本地关注对股票收益率的影响取决于投资者情绪,当投资者持积极情绪时,本地关注对股票收益率有显著的正向影响;当投资者持消极情绪时,该影响显著为负。在积极情绪和消极情绪下本地关注对股票交易量都有显著正向影响,但积极情绪下的影响程度比消极情绪下更大。此外,本地关注与投资者情绪的交叉效应在上市公司样本数量较多的北京、广东、江苏、上海、浙江这五个区域都存在,并与整体样本的特征基本一致,但在样本数量较少的其他区域并不完全一致。     

Drivers of helpfulness of online hotel reviews: A sentiment and emotion mining approach

Although online hotel reviews (OHR) help consumers in better decision–making, and service providers in better service design and delivery, they are hard to manage due to their high volume, velocity, and veracity. This paper focuses on the drivers of helpfulness of textual OHR, for which we have used text-mining techniques to find the sentiment content, polarity, and emotions; we have also used econometric and machine learning techniques to explain and predict its helpfulness. We found that content and title polarity lead to OHRs being less helpful, whereby this negative relationship gets accentuated with higher sentiment content. On the other hand, while negative emotion with low arousal makes OHR helpful, high arousal makes it less helpful. It has also been noted that after controlling for polarity, sentiment, and emotions, longer reviews are less helpful. Higher quantitative rating, recency of OHR and a reviewer’s past expertise make a review more helpful. Additionally, machine-learning techniques have been found to predict ‘review’ helpfulness marginally better than econometric techniques. This study contributes to OHR literature in terms of its performance, and would also help decision makers in OHR management strategy.     

Expectile CAPM

Conventional wisdom suggests that the uncertainty of uninformed noise-traders’ sentiment deters rational traders’ arbitrage activities. However, nowadays, social media have made the public sentiment highly predictable, whereas the CAPM-motivated beta-return relation still does not hold in practice. This study advances an argument that the sentiment can also be brought about by rational, sophisticated investors’ use of psychological insight; resultantly, the arbitrage activities are demotivated by their own sentiment, rather than deterred by noise-traders’ sentiment risk. The proposed expectile CAPM provides a parsimonious way to account for this claim, and leads to a sentiment-based functional form of pricing kernel.     

The discrete and differential impact of monetary policy

This study exploits a unique feature of the Australian monetary policy environment to determine whether economic recovery can be stimulated via central bank communications. This study finds that unexpected monetary policy announcements and communications have a significant and comparable impact on the value and volatility of the Australian foreign exchange market, suggesting that they can be used interchangeably to stimulate economic recovery. However, further analysis reveals that the state of the economy influences this impact. Specifically, during poor economic states, monetary policy actions speak louder than words, an adage that in this context provides actionable information for central bank regulators.     

Listening to the voice of the guest: A framework to improve decision-making processes with text data

The amount of texts available on the web is growing continuously and making sense of this unstructured data efficiently and effectively, therefore, poses a demanding challenge for organizations. Although computer science community has developed many techniques, there is ample room for improvement on organizational utilization of such text data, especially when referring to decision-making support. In this article, we propose and validate a framework towards an effective use of text data inside hotel industry, bringing tourism sector to this discussion. We combined three text mining techniques for text classification, sentiment analysis and topic modeling in a novelty way to allows managers to analyze guests’ comments and compares competitors in hospitality industry based on SERVQUAL. Our objective is to present an automatized process involving text data collection and analysis, improving decision-making process.     

ESO compensation: The roles of default risk, employee sentiment, and insider information

This paper derives a pricing model for employee stock options (ESO) that includes default risk and considers employee sentiment. Using ESO data from 1992 to 2004, the study finds that the average executive's subjective value is about 55% of the Black-Scholes value. Only employees who over-estimate firm returns (or insiders who know that the firm is under-valued) by about 10% per annum will prefer ESOs over cash compensation. Our model also shows that work incentives offered by ESOs may be far lower than those implied by Black-Scholes but that ESOs may induce less risk-taking behavior, contrary to typical moral hazard arguments. Findings may impact relevant accounting regulations as well as compensation decisions.     

Sovereign CDS spread determinants and spill-over effects during financial crisis: A panel VAR approach

This paper examines the determinants of CDS spreads and potential spillover effects for Eurozone countries during the recent financial crisis in the EU. We employ a Panel Vector Autoregressive (PVAR) model which combines the advantages of traditional VAR modelling with those of a panel-data approach. In addition to variables that proxy for global and financial market spread determinants we also employ variables that proxy for behavioral determinants. We find that the determinants of CDS variance are neither uniform nor stable during different periods and different countries. For instance, as we move from 2008 to 2014 the impact of the slope of the term structure on CDS spread variance is increasing for peripheral countries such as Spain, Portugal, Italy, Greece, Ireland, and decreasing for core countries such as Germany, France, Netherlands, Belgium and Austria. Other findings indicate that investor sentiment was an important CDS spread determinant during the subprime crisis, along with other factors, while spillover effects run from larger peripheral economies such as Spain and Italy to core countries; spillover effects from Portugal, Greece, and Ireland are of minor importance.