Can AI make economic forecasts by reading the newspaper?

Can AI make economic forecasts by reading the newspaper?

Researchers have developed many indicators to track various aspects of the economy, such as GDP, employment, and industrial production. But extracting a clear picture from these signals and predicting whether the country is headed for a recession remains a challenge.

“We have a data problem,” says Leland Bybee, a doctoral student in financial economics at Yale SOM. “It’s very difficult to get really good, high-quality data to understand what’s going on in the economy at any given time.”

In a recent study, Bybee, Yale SOM finance professor Bryan Kelly, and their co-authors explored another possible source of information: news articles. The researchers ran software to sift through hundreds of thousands of the wall street journal stories and quantifying the media attention given to specific topics.

They found that the prevalence of recession-related articles, in particular, predicted certain measures of economic performance well months or years later. In other words, when journalists wrote more about this topic, troubled economic times were more likely to follow.

“We should pay attention to information in news texts,” says Bybee.

Of course, the decision makers have already read the the wall street journal to get this type of information. But if they’re advocating for a new policy, they need rigorous quantification of news content rather than simply relying on the latest headlines, suggests Bybee. A policymaker wouldn’t say, “I read that the market isn’t doing so well, so we should start pumping out a bunch of stimulus checks,” he says. “You want to have some sort of statistical basis for this.”

Bybee’s team felt the news could contain valuable predictive data because editors’ job is to give readers information about the state of the economy. The the wall street journal is meant to be a “one-stop-shop” for learning about big issues like GDP, industry-specific news, people’s concerns about economic issues, and expert opinions.

The publishers’ incentives are “to report the news that matters to people who care about the economy,” Bybee says. “As a subscriber to the the wall street journalthat’s what I pay.

Bybee, Kelly and their team – which also included Asaf Manela from Washington University in St. Louis and Dacheng Xiu from the University of Chicago – wondered if they could quantify this information and turn it into a new economic indicator. . Policymakers could use this metric as another piece of evidence to make decisions, such as whether to take action to stimulate the economy.

The researchers collected approximately 763,000 WSJ articles, published from 1984 to 2017, and ran software to count the number of times specific one- or two-word terms were used in each article. A machine learning algorithm then identified broad “topics”: groups of terms that often appeared together. The team manually reviewed the word groups and tagged each topic.

For example, a set of words included “Greenspan”, “Yellen”, “federal funds rate”, “rate of increase”, etc. ; the subject identified by the software was clearly the Federal Reserve. Other topics included health insurance, China, natural disasters, airlines and elections.

Next, the researchers measured the amount of “news attention” given to each topic, defined as the percentage of words in the WSJ on this subject every month. They could analyze how attention increased and decreased over time.

News attention metrics seemed to track well with existing economic indicators. For example, when attention to the “recession” theme increased, industrial production growth and employment growth tended to decline.

And surprisingly, media attention to topics such as “recession” and “problems” (a general category that included terms such as “big problem”, “major problem”, “mess”, “debacle etc.) could explain 25% of the variation in stock market returns. In contrast, a set of 101 other economic measures could only explain 9%.

“What’s happening with the market and what’s happening with the the wall street journal are very similar. These are two ways of aggregating an extremely rich set of information.

The reason could be that “what is happening with the market and what is happening with the the wall street journal are very similar,” says Bybee. “These are two ways of aggregating an extremely rich set of information.” In other words, market returns on any given day say a lot about the state of the economy, as does the content of the newspaper.

But these analyzes only checked whether attention to the news was correlated with other indicators at the same time; articles could simply describe recent events. The researchers wanted to know if their new measures could help predict the future performance of the economy.

They therefore analyzed whether news attention to the “recession” was related to changes in industrial production and employment over the next three years. The team monitored these two indicators and other standard measures such as the S&P 500 index and the Federal Reserve funds rate. In other words, could news attention predict changes beyond what these existing indicators provided?

They found that an increase in the “recession” attention measure, from the 5th to the 95th percentile, was correlated with a 1.99% drop in industrial output 17 months later and a 0.92% drop employment 20 months later. Shorter-term forecasts also worked; for example, two months after the surge in media attention, industrial production fell by around 0.3%.

“There are things in the news beyond what the market picks up,” says Bybee.

The team also devised a way to identify the most critical articles that decision makers should read. When the forecasting software makes a prediction – for example, that employment will fall over the next few months – it also extracts the WSJ articles that have devoted the most attention to the theme of “recession”. Decision makers who are overwhelmed by massive amounts of information could hone in on these stories to get the most relevant details.

This method filters out noise to extract “what really matters,” says Bybee. “It gives you a tool to process that information.”


#economic #forecasts #reading #newspaper

Leave a Comment

Your email address will not be published. Required fields are marked *