AI will disrupt equity research from the bottom up
Maybe the pyramid structure of investment banks will start to look more like a diamond
AI technologies will change working practices across many white-collar professions, including investment banking. Tasks at the bottom of traditional career pyramids are likely to be automated first. Those towards the top will take much longer.
Many years ago, I started as a graduate trainee in the equity research department of a long-forgotten London stockbroker. My first jobs were to collect information, update models, format charts, prepare presentations and get the coffee. By doing these basics, I freed up my seniors to do more important tasks such as idea generation or meetings with companies and investors.
To climb from the bottom of this career pyramid, I had to prove competence at more advanced tasks — analyse information, set up models, create interesting charts, construct presentations and contribute to written research. I learnt these skills from colleagues further up the pyramid, more by osmosis than formal training.
I could even see a path to the top. To climb up there I needed to accumulate further analytical skills while learning how to originate ideas, present them to investors and build networks. This wasn’t easy. On the way up, I saw many talented colleagues fall by the wayside. Other white-collar careers such as consultancy, law, academia, even software coding, have similar pyramid structures.
My profession, sell-side equity research, peaked in the late 1990s bull market and has been fading ever since. Funding has been cut and headcounts reduced, but the job has changed remarkably little over my career. When I finally departed two years ago, I wasn’t doing anything much different to my very first boss. My juniors were doing a job similar to mine 34 years earlier. Sure the technology and access to data improved, allowing us to cover more stocks and markets with fewer staff, but the skills required to climb the pyramid were the same.
I was lucky to fall into a profession which changed little through my career. My hard-learnt skills did not become redundant overnight. But many think that artificial intelligence will change all that. Finally, here is a new technology that will disrupt the cosy white-collar professions. They could suffer the same fate as UK factory workers in the 1980s.
My default instinct is to be sceptical. So, having retired from equity research, I decided to see for myself. I’ve been working with a London-based start-up to identify opportunities to apply AI technology in the financial services industry. Apparently, I have useful “domain knowledge”.
There have been some wild claims about AI use cases in my old profession. Just ask ChatGPT to collect information, build company models, pick favourite stocks and write reports. Job done, no need for expensive analysts.
But that misunderstands the research career pyramid. The current technology is best suited to tasks which I performed in my beginner years. Information collection, model updates and presentation formatting can all be automated, even if getting the coffee cannot. However, as research tasks become increasingly complex and bespoke further up the pyramid, automation opportunities are more elusive. Right now, AI is better suited to being a research assistant than a research analyst. Or, to use the tech industry jargon, the scale opportunities are currently greater at the bottom of the pyramid than the top.
Of course, the AI will get better but it will have to prove itself before taking on more difficult tasks, just as I did all those years ago.
More focused work and AI models seem to be the way forward. One use case we have found is in content distribution. AI can allow customers to ask complex questions of a research department’s output, be it numbers or words — it’s almost like speaking to the actual analyst. This is a complex task, given the wide range of reports that analysts publish. For example, the shelf life of a two-page quarterly earnings preview is much shorter than an 80-pager published when an analyst initiates coverage of a stock. The generic LLMs struggle with these subtleties.
Given what I know about AI now, it’s tempting to say that I would never do my old job the way I used to. But it really depends on which old job. The new technology is most useful, and perhaps threatening, to those at the bottom of the pyramid. My old job at the top seems less vulnerable. AI will probably chip away at the traditional white-collar career path from the bottom upwards. Maybe the pyramid starts to look more like a diamond. But that raises an obvious question — where will the next leaders start their careers?