Artificial intelligence is picking up pace in financial services and the effect on jobs could be devastating.
230,000 jobs could disappear by 2025, according to research from consulting firm Opimas, and if you think this is confined to easily-automated back office roles, think again. High-paying front office jobs could just as easily be replaced by a machine – we’ve broken out which jobs and why below.
Asset managers have been struggling to justify their high fees anyway, particularly as ETFs and other passive investment strategies hoover up more investors. What’s more, human portfolio managers have been unable to maintain an edge as hedge funds and long-only asset managers alike draw on increasingly large external data-sets to inform investment decisions.
“AI will just see a reinforcement of the trend away from gut feeling and human portfolio managers and towards a quant-approach using large datasets,” says Axel Pierron, co-founder of Opimas, and author of the report.
Steve Cohen is already adopting an AI approach at Point72, ramping up its internal quant fund and scrutinising the “DNA of trades” in order to codify and replicate an elite PM using a machine. Raffaele Savi, head of developed markets at BlackRock’s scientific active equities team, believes that the machines are already beating humans in asset management and that the sector is set to tip into AI in a big way.
“Unlike quant algorithms, which tend to follow similar strategies, AI will add a competitive edge because it will draw from various different external datasets,” says Pierron. “Human judgement is still needed to interpret the data, or simply to unplug the AI when it fails to deal with market events. But the ratio of humans to machines will reduce dramatically.”
The key with machine learning, of course, is that it's not just about building an algorithm and plugging it in - it will learn from its mistakes and evolve.
Back office processing is an obvious target for using machines to take-over manual processes which “add little value”, says Pierron. The cost-income ratios of securities services firms are “very bad”, he says, and they’ll look to cut costs wherever possible.
AI is likely to be used for everything in the trading process from order generation, order routing, pricing and quoting and trade execution over the next eight years, suggests Pierron. What’s more, traders are subject to the same forces as portfolio managers on the buy-side – namely, firms using machine learning to crunch huge external data-sets in the hunt for alpha. The traders who survive need to understand both programming and Big Data, he says.
For the sales teams, there’s a different proposition. Banks are trying to do more with less, says Pierron, and this means empowering sales staff with complex customer relationship management systems powered by cognitive analytics. In a nutshell, this means that a machine will make recommendations to clients based on data around their previous behaviour.
“What this means is that sales staff will still be needed, albeit in smaller numbers, but that each employee will need to take on a greater number of clients,” says Pierron.
The obvious reasons for wealth management jobs going up in smoke are robo-advisers, which many predict will be able to offer better financial advice than human employees in the not too distant future. Pierron says that this will be the main reason for thinning out the ranks of private banks, but it also presents an opportunity for some.
“You have to believe that ultra high net worth individuals will want customised services and want to interact with a human. This could be a selling point for some firms,” he says.
Automation can come to investment banking advisory roles, seemingly immune to the onslaught of technology. But there are things that don’t need to be done by a human being. Goldman Sachs has mapped out 146 distinct steps to every IPO, and not every one of these needs to be carried out manually, according to a recent speech by CIO-turned-CFO Marty Chavez. But Pierron’s research suggests that AI could do financial modelling more efficiently than people.
“You have bright, well-trained people just doing number crunching, and it’s not the best use of their time. A computer could take away much of the grunt work, but this will mean fewer people are needed,” he says.
Photo: Getty Images