AI: The new frontier for credit risk modelling
19 May, 2022
With all the disruption stemming from COVID-19 over the past two years, how sound are credit risk models? This was one of the questions we sought to find the answer to with a global research study that surveyed 100 industry decision makers in LATAM. The results were more than a little unsettling – only 21% of fintechs and financial services organizations believe their credit risk models are accurate at least 76% of the time.
This state of great uncertainty in credit risk modelling is exposing the shortcomings of legacy approaches for credit risk decisioning that leverage limited data, workflow and automation – often in separate systems. To really level-up decisioning, organizations need more data, more automation, more sophisticated processes, more forward-looking predictions and greater speed-to-decisioning. And to this end, they need AI, machine learning, and alternative data.
Our survey underscored the growing appetite for AI predictive analytics and machine learning, data integration, and use of alternative data as the means to improve credit risk decisioning. Real-time credit risk decisioning was respondents’ No. 1 planned investment area in 2022, as organization’s work to resolve today’s “financial fault line” in credit risk decisioning.
Financial services executives see AI-enabled risk decisioning as the cornerstone to improvements in many areas, including fraud prevention (59%), improving cost savings and efficiency (52%) and improving accuracy of credit risk profiles (45%).
However, many companies struggle with mounting the resources needed to support their AI initiatives; it can take long time to develop and implement AI, and it can be prohibitively expensive. Only 39% of financial services organizations begin to see a return on investment from AI initiatives within 120 days.
Read next: A Mexican WeChat? Baz expands its superapp
AI for fraud prevention and financial inclusion
Sixty-one percent of decision makers in our survey indicated they recognize the importance of alternative data in credit risk analysis for improved fraud detection. Additionally, 58% recognize its importance in supporting financial inclusion. Alternative data is a more varied way for lenders to detect fraud before it happens and evaluate those individuals with a thin (or no) credit file by putting together a more holistic, comprehensive view of an individual’s risk
For unbanked and underbanked consumers, AI gives organizations the opportunity to support those consumers’ financial journeys. Financial services organizations typically struggle to support these consumers because they don’t come with a history of data that is understandable by traditional decisioning methods. However, because AI can identify patterns in a wide variety of alternative and traditional data, it can power highly accurate decisioning, even for no-file or thin-file consumers. This vastly benefits those who can’t be easily scored via traditional methods, while also benefitting financial institutions, by expanding their total addressable market.
By deploying AI and machine learning technologies, and embracing alternative data, organizations are on their way to improved agility and confidence in credit risk modelling. In doing so, they will be more prepared to react to changes moving forward, while also supporting critical industry imperatives such as fraud prevention and inclusive finance.
The era of AI is here – just in time for organizations to come to terms with and resolve the financial fault line in credit risk decisioning.
About the Author
Jose Vargas is General Manager and Executive Vice President, LatAm, at Provenir, which helps fintechs and financial services providers make smarter decisions faster with its AI-Powered Risk Decisioning Platform. Provenir works with disruptive financial services organizations in more than 50 countries and processes more than 3 billion transactions annually.