In this Asia risk webinar, artificial intelligence experts (AI) and machine learning examined the growing applications seen in the field and their merit in credit risk modeling
During a panel moderated by S&P Global market intelligence and Asia risk to Japan risk conference held in June, Hiroyuki Yoshizawa, Executive Director and Head of Product, Data, Measurement and Analytics at S&P Global Market Intelligence, and Yutaka Sakurai, Head, PR technology at the AI Finance Application Research Institute, examined the current state and future of holistic credit risk assessment.
Yoshizawa began the session by asking questions about the growing role of AI and quantum computing analysis in this area.
The role of AI and machine learning in credit risk modeling
Sakurai explained how AI and machine learning are compatible with credit modeling. In an example comparing the fields of trading and asset management, he noted that with trading, you can’t always progress by following the same pattern. It is important to change your approach and strategy and to adapt your methods according to the strategy and behavior of the other party. While machine learning can help you uncover statistical information about a specific point, ever-changing information means it can be difficult for humans to adapt in real time. Asset management is much more stable and therefore better suited to machine learning. Sakurai added that many people have tried to apply machine learning to trading but inevitably end up losing money and giving up.
Sakurai also highlighted the importance of alternative data in credit risk management, as it has been a growing trend in recent years. Machine learning applies to this alternative data that is not typically included in financial statements and can provide a higher degree of accuracy in managing credit risk. He added that there had been significant progress on this front in North America and Europe.
On the retail credit front, some transactions in Japan, such as deposits, have been studied with reasonably consistent results. However, alternative data is still currently limited in Japan, and Sakurai says little research is being done to advance this industry.
Yoshizawa reiterated Sakurai’s point that, compared to market risk projections, credit risk modeling is more compatible with machine learning. Following this, he predicts that AICredit risk analysis for individual businesses will soon become a growing trend, which is an exciting prospect.
Linking the discussion to his work, Yoshizawa explained how S&P Global Market Intelligence offers data to global financial institutions such as hedge funds and asset management firms, while noting that clients often perform credit risk modeling and investment decisions to move forward . S&P is currently focused on how it can deliver data in a way that provides the greatest value to its users. It is for this reason that data accuracy, data transparency, and raw data disclosure and data cataloging are extremely important.
Yoshizawa added that with a strict focus on credit risk, the most discussed challenges relate to the ability to present historical and current data in a uniform format that works for most businesses as well as the ability to maintain ‘hall. When looking at a single company, it is useful to map the company’s historical data, such as the evolution of different categories of companies and various equity links such as spin-offs, mergers and acquisitions. This mapping greatly facilitates the monitoring of any development.
Finally, Yoshizawa explained the link between historical financial data and real-time market data – in particular, the bridge between quarterly data and real-time data often used in models. For both debt and equity, the data can indicate a company’s volatility or a company’s likelihood of bankruptcy and financial viability. He then raised the question of how machine learning can use all available input data to help with business decision making.
Sakurai and Yoshizawa agreed that market data can be easily accessed. That said, Sakurai emphasized the need to keep in mind the many types of market data and methods of combining. Japan is relatively behind the game in terms of data collection, but it needs to start now. Once the data is collected correctly, there are endless options for combining and sorting it. Sakurai therefore explains that improving data matching and use will inevitably lead to better results. Both Sakurai and Yoshizawa point out that there is significant room for improvement as well as increased dialogue in this data space now to offer new suggestions for data processing.
Modeling and ethics
Yoshizawa wondered if modeling could help implement a more ethical decision-making process and if we could discuss ways to adjust it to account for a range of variables.
Sakurai noted that people usually ask the opposite: whether modeling will harm ethics. He points out that ethics can vary from country to country, noting that the WE and Europe have set the ethical tone for the rest of the world, particularly in terms of equality and non-discrimination. To apply AI and machine learning of data can often lead to discrimination and therefore is already being addressed in some countries including the WE.
Sakurai presented the example of retail credit, where machine learning can produce higher credit risk for ethnic minorities, despite the same set of input data. He noted that when this happens, it’s possible to get different results based on gender, age or ethnicity. If we can define this problem as discrimination, then it becomes clear that steps must be taken to remedy the situation. That said, this issue is complex.
Yoshizawa agreed that there is currently a lot of debate around the ethics of minority and gender data, and that it may be worth looking at simulations to understand the potential market and regulatory implications. . Yoshizawa believes the market is doing a good job of reflecting data obtained through scenario simulations, and that as data selection and modeling improves, many existing ethical issues can be alleviated.
On the broader topic of environment, social and governance, Sakurai added that detrimental results can appear from the data because the areas or standards chosen tend to provide more favorable or negative results depending on the industry. to which they are applied. He believes that people who make decisions based on these criteria need to understand that these discriminatory results can occur and therefore cannot be entirely relied upon. Models can be improved but will never be perfect, so Sakurai wondered if there was a need to consider additional standards for trading and investing decisions.
Yoshizawa further added that when adapting models, a very careful approach should be taken. When it comes to capital markets, the consensus has been that market participants need more and more data to generate profits. However, we are now at a point where there is an abundance of data, underscoring the focus on quality over quantity and on honing data intelligence.
Further clarifying what he means by “a conservative approach,” Yoshizawa noted that this approach should include various types of data, a clear relationship between data and output, and the coding and technology used by data scientists. to make sense of it all.
Sakurai added to the notion of a cautious approach, suggesting that a quick turnaround can’t be the only approach because you run the risk of creating a situation with more clutter and less accurate data. Bringing the discussion back to the use of alternative data, Sakurai noted that alternative data has a much shorter history compared to traditional data, so it is difficult to conduct research consistently.
Yoshizawa ended this section by noting that bulk and model fit will remain an important part of this discussion for some time to come. Although alternative data is not widely used in financial markets at present, the industry should find more applications for market data, as there is still a lot of data to be mined.
Forgotten Elements in Credit Risk Assessment
The webinar ended with a discussion on overlooked elements in credit risk assessment. Sakurai noted that, particularly in Japan, this is how higher interest rates affect credit risk. While in the WE and in Europe, more and more people are seriously considering this pressing issue, Japan has not experienced high interest rates for over 30 years, and therefore the correlation between interest rates and risk credit is not a priority. That said, Sakurai explained that higher interest rates influence credit risk in multiple ways: banks become less willing to lend and funds don’t flow as smoothly, borrowers face higher interest payments higher and spreads in commodity markets can increase rapidly. He added that the time had come to start seriously considering this issue in Japan.
Yoshizawa concluded that data can be much more useful when the following elements are well understood: data consistency, a general idea behind inputs and outputs, and the technological contribution of data scientists.
Finally, Sakurai noted that now that we understand the critical importance of data, it’s time to become more efficient in data collection and be more creative. The most important thing is to understand that this is a long-term game and the only way to progress in this area is to continue testing and accumulating more experience with the data.