Revealing what separates credit risk leaders from the laggards
Revealing what separates the credit risk leaders from the laggards
Executing an optimal credit risk model that simultaneously protects against loan default and supports business growth is a significant challenge for credit-providing organizations of all sizes. In the modern financial services landscape, it is arguably more difficult than ever. Organisations are under increasing market pressure to progress with digital transformation and now face unprecedented levels of competition, particularly from digital-first firms and traditional rivals speeding up their own digitalisation.
Credit risk is a key area for credit providers to lower costs and compete with digitally intelligent and agile counterparts. However, it is a function that can be difficult to execute well. Recently, an EXL client wanted to understand how it fared against its peers and competition, and what the driving credit risk analytics factors separating the leaders from the laggards were. EXL carried out a study on credit risk analytics, benchmarking the aforementioned client against 14 financial organizations across Europe, the United States, and Africa.
Key takeaway: organizational size does not dictate maturity
EXL assessed each of the 14 report participants in seven separate areas. These were:
1. Operating model within credit risk analytics
2. Credit risk modelling for customer acquisition
3. Data capabilities
4. Credit risk strategy for existing customer management
5. Credit modelling strategy for existing customer management
6. Model governance and management
7. Validation and implementation agility
The study results show that the two leading banks have the most mature credit risk analytics maturity level, while the five other large banks generally vary somewhere in the middle of the spectrum between “leader” and “laggard” under the “peer” category. Only two of the five large banks achieved “leader” status, respectively in one and two of the seven assessed areas, while all five were categorized as “laggard” in at least one area. The medium-sized banks were by and large laggards across the seven capabilities assessed, with the odd outlier category exception of “peer” or “leader”. What may surprise is that the three mature start-ups and one early-stage start-up ranked just after the two leader banks in terms of advancement, taking advantage of their ability to respond with agility to pace of market change.
There are three driving areas that the leading two large banks implement to a high degree of excellence, which enables them to achieve optimal credit risk performance. These are data and its use, speed of implementation, and machine learning models and their management.
The Elements driving a leading (or laggard) credit risk strategy
There are notable things that the leading organisations do to achieve high credit risk analytics performance. Conversely, the laggards also display a series of common traits.
What makes a credit risk leader?
The leading organisations generally display accomplished performance in modelling and strategy.
- Advanced gradient boosting machine (GBM) model implementation for existing customer management, leveraging machine learning technology and contributing 10- 15% against logistic regression models.
- Frequent, dynamic CLIs.
- Loss model development with short term risk as the target, as opposed to traditional model targets.
- GBM model variables limited to between 15 and 20.
- Limit optimisation implementation at account level and a margin optimisation model.
- Granular test-learn and clean test processes in place, which enables linking MI with strategy outcomes.
- Live data testing.
- Early warning risk indicator modelling and vintage loss forecasting in operation to measure strategy performance.
- Risk models and innovations in risk assessment are planned with implementation and business outcomes (profit and loss, risk etc.) in mind.
- Culturally, senior management across credit and business units align in viewing proposals quickly.
What makes a credit risk laggard?
The laggard organisations demonstrate multiple areas which require improvement in order to enjoy improved credit risk outcomes.
- Significant legacy technology issues, including poor data latency, slow query performance, and loading errors.
- Failure to transition away from logistic regression models to more advanced modelling.
- CLD and CLI are based only on bureau scoring, with an absence of robust internal scores.
- Lack of dedicated modelling structures for each credit risk program, resulting in a reliance on makeshift arrangements.
- Key metric inconsistencies as a result of multiple versions of definition truth by different teams.
- Strategies not based on a single customer view.
- Limited strategy capability as a result of inefficient and ineffective data analysis.
- Absence of operational automation, leading to higher costs.
- Very high weightage to rules-based segmentation, rather than detailed statistical optimisation.
- Consistent regulatory risks due to inconsistent historical data, especially concerning multiple portfolio acquisition.
EXL was able to identify the key differentiators between what makes a leading credit risk analytics organisation and a laggard as well as develop a detailed strategy and execution roadmap for the client to upgrade their own credit risk analytics function. With these study takeaways, EXL’s client, in partnership with EXL, was able to chart a clear path forward on how to improve its credit risk analytics function performance.