Software uses facial expressions to 'predict' bank's policies

Software (IOB: 0NJS.IL - news) can be used to analyse the facial expressions of central bank governors and predict policy changes, researchers have claimed.

Two artificial intelligence researchers in Japan have been using technology to measure the momentary expressions on the face of Haruhiko Kuroda, the governor of the Bank of Japan, during press conferences.

The research uses computer vision technology to detect minuscule changes which humans might not pick up on, and which could otherwise be subjectively interpreted.

The technology was developed by Microsoft (Euronext: MSF.NX - news) as part of its cognitive services products, and uses an algorithm to analyse an image and return a confidence rating for eight emotions: anger, contempt, disgust, fear, happiness, neutral, sadness and surprise.

Mr Kuroda's expression was seen as neutral most of the time, but there were subtle expressions of other emotions as well, the study found.

The governor's face showed tiny signs of "anger" and "disgust" at press conferences preceding major policy changes, according to the researchers from Nomura Securities and Microsoft.

Specifically, Mr Kuroda's face betrayed the Bank's decisions in January 2016 to introduce negative interest rates and the September 2016 decision to adopt a "yield curve control".

According to the study, Mr Kuroda was sensing the constraints of existing policies more than a month before the central bank's board decided to change them.

The researchers, who presented their study to the Japanese Society for Artificial Intelligence, said that Mr Kuroda's face showed less "sadness" in briefings after the changes were announced.

If the technology worked, the effects its predictions could have on the market are enormous and may allow investors to adapt to a dynamic financial environment before their competitors.

The study's lead author, Yoshiyuki Suimon, a researcher at Nomura's Financial & Economic Research Centre, said the technology could become a powerful tool in predicting policy changes.

He hopes to analyse the facial expressions of US Federal Reserve chair Janet Yellen and European Central Bank governor Mario Draghi next.

The study offers "a fresh approach to analysing policymakers' inner world and minds", said Professor Kiyoshi Izumi, a specialist in financial data mining.

"I think it's possible to build up data on other key people globally, quantify it to predict what comes next," he added.

A Bank of Japan spokesman told Reuters that the bank was not in a position to comment on the study.