Can Artificial Intelligent Predict Market Recession?

The machine learning algorithms have spotted a possible snag. According to Allianz Global Investors’ chief officer for systematic equities, Michael Heldmann, “a very high degree of recession risk” is indicated by the firm’s statistical algorithms that employ AI and ML to anticipate recessions.

Contracts, emails, websites, audio, and video recordings, and computers are all sources of the zettabytes of data that contemporary businesses must analyze. However, no human mind is capable of processing so much information. Financial Services Assistant Vice President at frog, Toshi Mogi, said, “We don’t yet have a shared knowledge” of the complicated economic structure. As we improve our AI toolkits, amass more data, and build more nuanced models, we’ll be able to take on tougher challenges and put in place more cutting-edge systems.

Several companies have created cutting-edge solutions, which is good news. For example, the Sensai team in Palo Alto uses “deep learning” to analyze and comprehend unstructured data such as business papers, transcripts, and social media to expedite the analyst’s process. Palantir, which the military had previously utilized for counterintelligence purposes, has recently earned notoriety on Wall Street due to its innovative use of artificial intelligence to prevent fraud. Companies like these are a driving force behind the widespread use of AI in the banking industry, especially in a few key areas. Economists have established various economic indicators to monitor the economy’s health, including gross domestic product, unemployment rates, and industrial production.

Bybee, a professor of finance at Yale SOM, together with Bryan Kelly and two other writers, have lately looked into news stories as a potential new data point. To be more specific, they discovered that the volume of media coverage given to recessions was a very reliable predictor of subsequent economic indicators. To put it another way, bad times followed periods of increased media coverage of the economy. Bybee urges us to take the news seriously because “it might save our lives.” Bybee’s team reasoned that economic forecasting information would likely be featured in the news since editors are responsible for keeping their readers informed about the economy. If you want to know about GDP, certain sector news, people’s anxieties about economic issues, and expert viewpoints all in one spot, then the Wall Street Journal is where you need to be. According to Bybee, the economic news is prioritized by editors because of financial motivations.

Bybee, Kelly, and his team, which included Asaf Manela, Louis, and Dacheng Xiu, worked hard to turn the company data into a new economic indicator. The researchers used a computer to look at more than 763,000 WSJ stories from 1984 to 2017 to determine how often certain words and phrases were used. The writings were then looked at by a machine learning system that looked for repeated groups of words, or "themes," to find words that were used more than once. Each set of semantically related phrases was individually examined and assigned labels. For example, when the word "recession" was talked about in the news, production and employment grew more slowly. Just as a day’s worth of news may shed light on the economy, so can the market’s performance.

Articles may report on events as they happen, but research has only looked at the correlation between news coverage and examined indicators. The researchers pondered whether the newly developed indicators might be used for economic forecasting. So, they examined how the media portrayed the “recession” and how it affected economic production and job development in the three years following.

Rather than relying on the metrics above, maybe a more accurate indicator of future trends is the degree to which individuals follow the news. The study found that a five-percentile increase in "recession" attention test scores was linked to a 1.9 percent drop in manufacturing production 17 months later and a 0.9 percent drop in employment 20 months after that.

The short-term forecasts were spot-on, with industrial production falling by 0.3% about two months after the increased media attention. Even though data on stock market prices is publicly available, the “recession” emphasis indicator aimed to increase predictive ability.

To sum up, the level of AI we have now has given economists a great tool for predicting the market. The research done thus far is only the tip of the iceberg. If more data and signs are given to the AI system, it may be able to give a more detailed analysis of how the market is doing. This expected change doesn't change how logarithms have been used in the field of artificial intelligence so far. AI logarithms have become more accurate and can now be used for data intelligence. To improve AI's ability to predict, more effort will be put into collecting high-quality data and choosing the most important indicators.