Robots to replace complaining junior bankers

Imagine, the pitch will undoubtedly go, a third year associate who never goes out clubbing and therefore never comes in late with a hangover.  An analyst who works ninety hours a week or more and doesn’t moan to the newspapers.  Imagine a junior banker who doesn’t celebrate Christmas, Thanksgiving or any other holiday and has never heard of weekends.  Imagine … the robo-analyst.

This prediction – and indeed this pitch – has been made plenty of times in the past, and as Liam Proud of Breakingviews acknowledges, it’s always fallen flat on its face, and always for the same reason.  That reason being that when senior bankers hear the words “software that gathers data automatically and pulls it into live pitchbook templates”, they immediately think “numbers pulled from a database”, and that tends to trigger a horrible memory of a time that they did that and ended up embarrassing themselves in front of a client. 

The problem is that companies are all unique individual entities, who have to report numbers that fit into a set of uniform categories.  This means that lots of compromises have to be made by data providers, many of which have to be dealt with by adjustments later.  Not only that, but the choice of which compromises and adjustments to make will often depend on what purpose you’re trying to use the information for – everyone who has ever been told “you can mainly reuse stuff from the last pitchbook” will be aware of the surprising extent to which this isn’t true.

And although the work which juniors put into documents is almost entirely pointless, it’s only almost entirely pointless.  One time in a hundred or more, even checking that all the fonts are exactly the same will reveal that something has been cut and pasted from the wrong source.  It’s useful to think of analysts on all-nighters as doing something similar to a concert promoter removing all the brown M&Ms from a bowl of candies before a Van Halen concert – it’s not the work itself that the client necessarily cares about, it’s the assurance that no corners have been cut which might result in something important being missed.

So, in order to fulfil the dream, any robot that automates the pitch book tasks and leaves bankers to do, as Dan Dees of Goldman Sachs puts it, “more of the meaningful and less of the menial” is going to have to be so good that even paranoid and perfectionist banking MDs are happy to trust it.  Is that possible?

Well, never say never. As well as Goldman, Moelis and Barclays have ongoing automation projects, and industry analyst Mike Mayo thinks it’s an inevitability.  But it seems that the level of general, adaptable artificial intelligence that would be needed is much more difficult than the high speed number crunching which has decimated headcount in sales and trading.  Just as senior bankers aren’t at risk until they make a machine that can schmooze the clients, junior bankers probably won’t be obsolete until they make a machine that you can get angry with when it makes a mistake.

 Daniel Davies –


Leave a Comment

Your email address will not be published.