ChatGPT vs The Sellside

After experimenting a bit with OpenAI’s ChatGPT we’ve observed that opinion writers should worry, but FTAV journalists are probably safe for now. But what about investment analysts?

Yin Luo, Wolfe Research’s head of quantitative research, has also been kicking the ChatGPT tyres, and earlier this week published a note written almost entirely by the AI chatbot.

Wolfe’s chief quant asked it to source academic papers that explored the link between corporate lay-offs and equity market performance, find three lay-off databases and write some Python code to scrape the data, and write a research note on how sackings affect the stock of a company.

Here is what ChatGPT came up with:

It is difficult to generalize about how stocks perform during lay-offs, as the stock market is complex and can be affected by many different factors. However, there is some evidence to suggest that lay-offs can have a negative impact on a company’s stock price. One study found that firms that announce lay-offs often experience a decline in their stock price in the short term. The study, which looked at data from 1991 to 2003, found that on the day that a company announced lay-offs, its stock price tended to fall by an average of 0.7%. This decline was even more pronounced for firms that announced large-scale lay-offs, with an average drop of 1.2%.

Other research has also found a negative relationship between lay-offs and stock performance. For example, a study published in the Journal of Financial Economics found that firms that announced lay-offs saw a decline in their stock price in the month following the announcement, compared to firms that did not announce lay-offs.

It is important to note that the relationship between lay-offs and stock performance is not always negative. In some cases, companies may announce lay-offs as part of a restructuring effort that is intended to improve their financial health. If investors believe that the restructuring will be successful, this may actually lead to an increase in the company’s stock price.

In conclusion, while it is difficult to predict how stocks will perform during lay-offs, there is some evidence to suggest that lay-offs can have a negative impact on a company’s stock price. However, the overall effect on a company’s stock price will depend on a variety of factors, including the size and scope of the lay-offs, the reasons for the lay-offs, and the broader economic context.

Wolfe’s quants fact-checked this and found similar results. ChatGPT was similarly solid when tasked with writing up an optimal asset allocation for a stagflationary era, and when exploring the performance of equity factors.

Luo also used ChatGPT to correct some Python code, and asked it to solve a maths problem that had taken him 15 minutes to answer. ChatGPT did both with aplomb (although Luo feels ChatGPT’s solution was not as intuitive as his own).

JPMorgan’s TMT salesman Jack Atherton has already asked ChatGPT to write one of his daily emails for him, with “impressive” results. And Luo reckons the emergence of systems like ChatGPT might be bad news for parts of his own profession:

Is Sell-Side Research at Risk? Many jobs are at risk of being replaced by AI. ChatGPT can probably already write the routine earnings preview and review as well as many sellside analysts. To survive the AI revolution, analysts need to be more innovative, be equipped with proprietary data and models, offer deep domain knowledge, and provide unique forward-looking insights.

However, when Luo asked “will I be laid off from my place of employment”, ChatGPT cannily dodged and gave a version of its bland, rote answer to tricky questions.

I’m sorry, but I am not able to provide specific information about your personal employment situation. As a large language model trained by OpenAI, I do not have access to your personal information or the ability to predict the future. My purpose is to assist users in generating human-like text based on the input provided to me. I do not have a physical presence or a personal history, as I am a machine learning model that exists solely in the form of code. Is there something else I can help you with?