Economic bubbles and financial crises: Detecting outliers generated by self-feedback and crowd behavior

Research

Economic bubbles and financial crises
Detecting outliers generated by self-feedback and crowd behavior

Big data reveals the reality of bubble economies

A bubble economy is defined as a state where the prices of things are higher than they should be, but nobody knows what their true value is.

However, almost exactly the same situation exists in the world. In particular, there are many such situations in economic big data.

As a rule, if items are the same then they cost the same, and every item has its price. Therefore if there are two items that appear to be the same, but one is very expensive, then the cost of this item is likely to have been inflated by a bubble economy.

Endogenous mechanisms driving asset markets

If prices continue to rise because nobody knows what the price should actually be, then this gives people the impression that the intrinsic value must be higher, resulting in calls for even more investment.

By chance, purchasing is concentrated more on certain brands than on others, or more on real estate in certain areas than in others. Whatever the initial cause, this concentration creates an infinite loop that leads to more concentration.

This tendency is strong when there is a glut of money, and bubbles can form from the expansion of slight differences, even in financial products whose prices remained the same during a recession.

Immediately spotting articles that affect the market

Out of more than 90 million global news stories, only a few had an effect on the market.

To detect such articles as soon as possible, it is essential that we constantly measure the novelty of articles (how many words they use that did not appear in earlier articles) and their newsworthiness (if the story was picked up by multiple competing media outlets in a short period of time).

References

[1] T. Mizuno (2018) Best Paper Award at Transdisciplinary Federation of Science and Technology the 2018 Oukan Conference

[2] T. Mizuno, T. Ohnishi, T. Watanabe (2017) EPJ Data Science 6, 26.

[3] T. Mizuno, T. Ohnishi, T. Watanabe (2020) Proceedings of the 23rd Asia Pacific Symposium on Intelligent and Evolutionary Systems, 194-202.

Research topics

Research Publications

Research