Information transmission: evidence from investor site visits and analyst forecasts

Abstract

The thesis focuses on the process of information transmission from different perspectives. The chapter 2 explores the information transmission between insiders and outsiders by detecting the investor site visit’s effect on reducing information asymmetries between insiders and outsiders based on the text-based analysis. This chapter classified site-visit Q&As (questions put by participants to management and answers given by management) from 2012 to 2019 into 15 categories using a machine-learning approach. The results indicate that the contents discussed in site visits can predict future corporate finance decisions (mergers and acquisitions (M&A), seasoned equity offering (SEO), and managerial turnovers) and can predict the performance of the company affected by these decisions in the next one-to-three years, and these significant relations remain in various evaluation periods.
The chapter 3 explores the information transmission between management and analysts by detecting the quality of the information released from the site visit from the angle of speaking language standard level since the standard level of voice-based information affects the extent to which outsiders observe the information. The chapter 3 examines the role of the standard level of pronunciation of the chairman by using speech scores derived from evaluating the audio of speeches using the iFLYTEK Open Platform. The results confirm the positive relationship between the standard level of pronunciation corresponding to the company’s chairman, who participates in the site visit, and the accuracy of this company’s earnings forecast by analysts participating in this site visit. The difficulty of understanding information delivered from the site visit intensifies the effect of the standard level of pronunciation.
The chapter 4 explores the information transmission by detecting the extent to which mutual funds rely on financial analysts’ recommendations when they digest information. The results illustrate that mutual funds do not rely heavily on analysts’ recommendations before the company releases the earnings preannouncement and therefore reduce their shareholdings in the companies. However, mutual funds have good performance before the pre-earning announcement date. The implied mechanism is profit-taking due to risk-averse. Mutual funds rely more heavily on analysts’ recommendations issued by the securities companies having at least one star analyst. Therefore, there is no profit-taking in these mutual funds before the date of the pre-earning announcement, which makes these mutual funds have further positive returns after the date of earnings preannouncement.

Keywords: Site visits; Machine learning; M&A, SEO, Turnovers; Standard level of pronunciation; Analyst forecast accuracy; longest common subsequence (LCS); Mutual fund; Analysts’ recommendations; Earnings preannouncement; Shareholding

Date of Award	Mar 2024
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Shuai Yuan (Supervisor) & Xiaogang Bi (Supervisor)