By Dr. Leslie Kanthan CEO of TurinTech
Recent reports estimate that global AI spending is expected to reach the $500 billion mark in 2023. Moreover, a PwC study reported that by 2030, the potential contribution to the global economy from AI is estimated to be $15.7 trillion.
There is no denying then that AI is here, there, and everywhere. In this article we highlight three areas where AI can deliver real value and help leaders make better financial decisions:
AI helps with Transaction Cost Analysis in trading, payments, etc.
In a recent Refinitiv report, traders put data analysis and data science as the top-ranked skills for the next one to three years across asset classes, ahead of market and market structure knowledge. In particular, when ranking key areas of workflows that needed to be automated, equities traders put transaction cost analysis in third place, ahead of pre-trade price discovery.
One area where AI is useful is calculating market impact costs, which are the additional costs traders have to pay on top of their initial costs due to market changes. When we look at transaction cost analysis, the market impact costs are the only costs that cannot be defined before the trade is initiated. It is here that AI comes in handy by complementing existing market impact models with additional insights. Because traditional nonparametric techniques have no economic intuition for the drivers of price impact, they tend to capture noise rather than relevant information. In addition, they are also unable to distinguish between permanent and temporary market impact.
AI techniques, on the other hand, with their ability to capture nonlinear dynamics, have shown to be especially useful for predicting market impact. In particular, AI has also found its successful application when it comes to estimating the market impact of trades in assets that do not have sufficient historical trading data. In this case, a cluster analysis approach can help solve the issue by identifying comparable assets with similar behaviour and using their historical data instead.
AI helps to detect fraud/online scams
Notably, a Business Insider report identified fraud detection and risk management as one of three main areas of application of AI in banking. According to DarkReading, Visa has invested $500 million (of its USD9 billion cybersecurity investment) on data analytics and AI capabilities which have been embedded in more than 60 different services to identify and block fraud on its networks.
AI can help in three particular areas:
- Detect transaction fraud faster, at scale
- Reduce false alerts in anti-money laundering
- Individualise credit scoring to increase revenue and minimise risk
Let’s take as an example an asset management firm. Despite having systems in place to help run anti-money laundering checks, these often fail to provide accurate results given the fast-evolving nature of the threats. In addition, because these systems use rule-based analytical methods, they can cause the firm to experience even 50% of false positive alerts. Investigating these unnecessary alerts amounts to 42% of the total outlay of AML compliance team’s cost. Not counting the long hours CCOs are probably spending to fill documents to explain a specific risk analysis to comply with regulations.
Applying AI-enabled risk models can help reduce the false positive rate by 98% (from 50% to 1%). In some cases, it can boost the compliance team’s productivity by 300% by transforming data into risk insights and by reducing costs in processing transactions. It can also provide the team with explanations on the key factors in risk assessment to satisfy regulatory requirements.
AI helps to increase business efficiency leading to overall cost savings.
The PwC report mentioned above also identified and rated around 300 use cases where AI would make a significant impact on businesses. It shows the underlying principles of how AI is applied are often the same, but the sectors differ (banking, retail, manufacturing, automotive.)
The result, however, is always improved efficiency, better use of business resources, and cost savings.
Efficient AI and optimised code can assist business leaders in making more accurate and fast-paced financial decisions which is crucial for positive business results.
Insurance claim processing, for example, is an area with a great opportunity for improvement as it is still predominantly manual, and paper-based. Being mostly manual, it has the disadvantages of being highly inefficient and more prone to errors, which leads to high operating costs and poor customer experience.
AI can help insurers analyse vast amounts of customer data quickly and speed up claim settlement processing time – in some cases by 50% – by building models that can predict with more than 90% accuracy. In addition, the models can discern among various claim triages (e.g. claim size and complexity, litigation risk, fraud profitability, and subrogation potential) in real-time. By applying AI technology manual errors can be eliminated and productivity can be improved by looping humans in when it is truly necessary thus reducing costs.
The bottom line
It is clear that the application of AI can bring real value to professionals in the financial services industry – whether traders, compliance and risk officers, or insurers. In order to succeed with AI in your business, companies need to implement efficient AI that can predict quickly and accurately, reduce risk and costs, and improve customer experience. BFSI firms need to look beyond simple automation – whether fixed, linear market impact models or rule-based analytical methods for detecting fraud – and apply innovative AI technologies to bring tangible and measurable benefits to the business.