Quantifying the economic cost of political uncertainty is difficult. There are few signs outsiders can use to understand what firms are thinking—and what they’ll do next—during times of perceived risk. But a group of researchers is turning to computational linguistics to crack the code.
The language of earnings calls can help diagnose economic outcomes associated with periods of uncertainty, according to research by Chicago Booth’s Tarek Alexander Hassan, Tilburg University’s Stephan Hollander and Laurence van Lent, and London Business School’s Ahmed Tahoun.
The researchers built a machine-learning algorithm that drew upon two-word combinations previously associated with political risk in outside texts, then used that tool to examine 175,000 earnings-call transcripts. Their results helped them understand how much of the conversation between management and analysts centered on political risk during each conference call.
The findings suggest that when a company spends a greater portion of its call discussing political risk, it’s more likely to make certain financial decisions that might mitigate that risk. “Firms that perceive risks associated with a particular political topic hire less and have less investments,” Hassan says. “These same firms then donate more and lobby more on the topics that they find concerning.”
Analyzing conference calls may be a particularly useful approach to gauging the effects of uncertainty because risk is not always felt industry-wide, and because companies often act to mitigate the perceived risks of new policies well before those policies are in place.
This new use for the language of earnings calls comes at a particularly relevant time. Because of political developments such as Brexit and the election of Donald Trump, political risk “is at historically high levels,” Hassan says.