In theory, artificial intelligence (AI) should be able to analyze a company’s performance and issue equally or even more predictive reports than a real, human stock analyst.

But Rotman School of Management professor of accounting Hai Lu thinks there’s a lot more human touch involved.

“Most analysts have some level of social skills. They’re not just computer nerds,” says Lu.

While the job is, on the surface, all about research, social skills help in various ways, says Lu. Those who are personable can develop strong relationships with upper management at the companies they’re monitoring, gaining better information access, including site visits and airtime during earnings calls. Socially skilled analysts should be able to collaborate well with co-workers and be savvy enough to read social cues — maybe even detect when someone is exaggerating or lying — both of which can lead to better insights.

All of this should feed into more well-rounded, accurate forecasts that sociable analysts share effectively with investors, says Lu.

“We don't have any solid scientific evidence to support and quantify how your social skills could affect your job performance,” says Lu, adding that without good data, researchers are unable to assess the effect of social skills.

It’s a challenge to measure social skills objectively. However, academics working in psychology have shown that the number of connections someone has correlates to their sociability.

“When you think about it, it makes sense. When someone has a lot of friends, they must be sociable. So we think that the number of connections you have on social media can be a proxy for social skills,” says Lu. 

Over a working dinner, Lu and assistant professors of accounting Congcong Li of Duquesne University and An-Ping Lin of Singapore Management University decided to leverage this approach to measuring social skills to understand the output of stock analysts.

Their findings in what is the first large-sample study looking at social skills and the work of financial analysts were published earlier this year in Contemporary Accounting Research.

The team looked at the earnings forecasts of U.S. analysts throughout 2014 and 2015, finding 2,280 analysts researching 2,767 public companies. They counted how many connections these professionals had on LinkedIn, then validated that measurement with other social-skills predictors, such as having an MBA (the idea being that this would be an environment in which individuals can learn those skills, or individuals with such social skills will self-select to take the degree to advance into management positions), participation in earnings conference calls, and working at a company that hires for social skills.

The researchers compared this information to the earnings forecast error (AFE) - or the difference between the analyst’s forecasted earnings of a company’s shares and the actual earnings - of the analysts.

Lu says the team’s theory that sociable analysts are better at their jobs proved true. Analysts with strong social skills produced forecasts with 3.2 per cent lower AFE than the mean. In other words, they were more accurate. “It’s not a lot, but it’s significant,” says Lu.

Sociable analysts also issued more profitable stock recommendations — 5.3 per cent over the mean. Their “buy” and “sell” recommendations elicited stronger market reactions. As professionals, these analysts were more likely to be voted as All-Star Analysts by the Wall Street Journal.

Lu says the study could inspire new hiring approaches that screen for so-called soft skills and emotional intelligence. Organizations should also consider on-the-job training on developing relationships and fine-tuning communication skills for new analysts.

Lu thinks the model he and his team created for measuring social skills for analysts could be leveraged to understand what’s needed in other industries as well. “If a future study wanted to look at social skills, they could use what we did.”

For now, as AI makes inroads across industries, Lu hopes the research will help protect against bringing in too much automation around market analysis. “We don’t want machines taking over all of our jobs,” he says.

He’s all for a future where we look to evidence to understand the vital roles technology and people play in getting the best results. “There are some technical things that machines can do better than humans,” he says. “But they can’t develop social skills.”