Researcher ORCID Identifier

https://orcid.org/0000-0003-1839-9059

Graduation Year

2021

Date of Submission

5-2021

Document Type

Campus Only Senior Thesis

Degree Name

Bachelor of Arts

Department

Economics

Reader 1

Professor Darren Filson

Terms of Use & License Information

Terms of Use for work posted in Scholarship@Claremont.

Abstract

In this paper, I investigate the relationship between sentiment on Twitter in tweets relating to certain initial public offerings (IPOs), and the first day returns of these IPOs. Twitter, specifically, is a very simple interface and widespread across many groups of people, making it the perfect data source to gather peoples’ opinions and generate sentiment data from these opinions. Given the presence of social media in everyday life as a source of information/news, this study can potentially by used to create a live trading algorithm or prediction product on the information sell-side of trading securities.

This paper explores 129 companies that IPOed between April 20th 2020 and March 29th 2021. I collect data on these IPOs from Bloomberg, and Twitter data from a Claremont McKenna College database, containing all geolocated tweets from the last three years. In order to calculate sentiment, I broke the tweets down by word, assigned a value based on a dataset in the R packages “textdata” and “tidytext”, and then recompiled the data with aggregated sentiment scores. I measure sentiment with a variable that is either positive or negative, to measure the direction of the sentiment. I also measure the intensity of the sentiment, which is a value between zero and four, to measure how far off neutral the opinions in these tweets are. I also used one model in which I interacted the two variables to show intensity’s different effect based on whether the tweet is positive or negative.

I use linear regression to estimate the effects of these variables using seven different models. I hypothesize that favorable sentiment will increase the first day returns of IPOs: if the average sentiment in tweets relating to a firm in the 7 days prior to its IPO is positive, then, on average, the firm will have more positive first day returns. I also hypothesize that more intense sentiment will increase the first day returns of IPOs: if the people who are tweeting positively about these IPOs have “stronger” opinions, then this will lead to higher first day returns. My results somewhat support my hypotheses, but due to data collection methodologies and the difficulty in predicting short term returns, I could not confidently draw conclusions from the results with statistical significance.

This thesis is restricted to the Claremont Colleges current faculty, students, and staff.

Share

COinS