AP’s computer-generated earnings stories increase trading

Computer-generated earnings stories published by the Associated Press increases trading in the companies that are the subject of the articles, according to a study led by two accounting professors.

The study examined trading in companies before the AP began publishing computer-generated earnings stories in 2014 and then afterward.

It focused on more than 2,000 companies that did not receive earnings news coverage before the AP began using software from Automated Insights that now had stories written about their financial performance.

Trading rose by 11 percent in those companies, according to the study, which was published in the Review of Accounting Studies.

“We find compelling evidence that automated articles increase firms’ trading volume and liquidity,” concluded the authors. “The effects are most likely driven by retail traders. We find no evidence that the articles improve or impede the speed of price discovery. Our study provides novel evidence on the impact of pure synthesis and dissemination of public information in capital markets and initial insights into the implications of automated journalism for market efficiency.”

The study does not answer whether the market interprets inferences from algorithmically-generated articles differently from similar articles by humans.

The study was done by professors Ed deHaan of the University of Washington and Elizabeth Blankespoor of the Stanford University Graduate School of Business, and Christina Zhu, a graduate student at Stanford.

The study can be found here.