Governments could soon use artificial intelligence (AI) in predicting if a bank bailout will save taxpayer money in the long term or not. The AI tool is developed by researchers from the University College London and the Queen Mary University of London.
The AI tool described in the paper "An AI Approach for Managing Financial Systemic Risk via Bank Bailouts by Taxpayers," which was published in Nature Communications, assesses whether a bank bailout is the best strategy for taxpayers and also suggests how much investment should they get as well as identify the bank or banks the government should bail out any time.
Governments Can Now Use an Artificial Intelligence Tool To Assess Whether or Not To Bail Out a Bank in Crisis
Predicting the Future and Reviewing Past Bailouts Using the AI Tool
The authors of the paper tested the algorithm using data from the European Banking Authority, which consists of 35 European financial institutions declared as the most important to the global financial system. But national banks can use and calibrate it using detailed proprietary data unavailable to the public, SciTech Daily reports.
The paper's corresponding author Dr. Neofytos Rodosthenous from UCL Mathematics said that government bank bailouts are complex decisions with not only financial implications but also social and political. Through using the AI tool, governments could assess the financial implications of whether it would be in the best interest of taxpayers.
Professor Vito Latora, a co-author of the paper from the Queen Mary University of London, added that governments can also use the AI tool to review past crises and gain valuable learnings to inform future choices. For example, they could review the UK government's bailout during the 2007-9 financial crisis.
A Mathematical Framework To Compare Bailout Strategies
Government investment in a bank increases during a bailout and reduces the risk of default. In their study, researchers created a mathematical framework to compare different strategies of bank bailouts in terms of predicted losses to taxpayers. Using the mathematical control process called Markov Decision Process the researchers incorporated the effect of government intervention at any time.
Then they developed the AI tool to assess the optimal bailout strategies and compared it to no intervention to different types of intervention. They found that a government bank bailout would be optimal only if the stakes of taxpayers in the banks are greater than some critical threshold determined in the model. It also drastically changes once the percentage loss has gone above that threshold.
Furthermore, Science Daily reports that government bank bailouts tend to be more favorable the greater the network's distress, the longer the crisis lasted, and the bigger the exposure of banks to other banks.
They also found that the best strategy for taxpayers was if the government will continue to invest in that bank to prevent default once that bank received a bailout. Such a move will result in a lack of incentive for the rescued bank to protect against risk and potentially increase risk-taking.
Dr. Daniele Petrone, the lead author of the study, noted that banks have weathered the current economic storm triggered by the COVId-19 pandemic. Regulatory measures issued during the 2007-9 crisis helped them to be resilient to avoid bankruptcies across industries. But no one can predict the effect of the financial system and so bailouts are still a possibility.
Check out more news and information on Artificial Intelligence in Science Times.
