Director of Model Development, Retail Risk Models and Global Risk Management
Bringing risk management in-house
Scotiabank builds a retail risk management modeling practice to control costs and speed development
Outsourcing some business processes can have benefits, such as lowering overhead and driving volume efficiencies. But it isn’t always the right answer.
That’s what the Bank of Nova Scotia – commonly known as Scotiabank – found when it looked at its risk-modeling process for retail lending. Scotiabank was spending a fortune using a consulting firm to develop and modify risk management scorecards, but still had to perform the data preparation work in-house.
Dina Duhon, Director of Model Development, Retail Risk Models and Global Risk Management for Scotiabank, estimates about 70 percent of risk modeling work is in the data preparation. “The most interesting work – actual scorecard development – was being done outside the bank,” Duhon says.
It wasn’t just expensive and inefficient, it was also a deterrent to attracting top analytical talent.
The most interesting work – actual scorecard development – was being done outside the bank.
“In our industry, it is essential to have strong analytical skills and by not doing the modeling in-house it was challenging to retain and find qualified employees,” Duhon says.
Scotiabank selected Duhon, who has years of financial experience and is President of the Toronto Data Mining Forum, to build a team of more than a dozen retail risk modelers in 2011. With talent on board and a deep foundation in SAS® analytics products, the bank brought risk management model development in-house.
The domestic retail risk model-development team handles Basel models, as well as all other risk models that are required throughout the customer life cycle for products such as mortgages, lines of credit, auto loans, etc.
Duhon says the bank evaluated a number of risk management modeling tools, but Credit Scoring for SAS® Enterprise Miner™ was a clear best choice. Data preparation and management were already being performed using SAS. Choosing SAS® Enterprise Miner™ allowed the bank to use a single tool, the same language and a standardized methodology to create risk models and scorecards.
Seeing the real bang for the buck
But it’s not just about risk scoring. The new process also introduces more transparency into governance requirements. Come audit time, every process can be replicated exactly. That’s where much of the return on investment comes from. “There are tangible and intangible sources of ROI,” Duhon says. When making the business case, the biggest benefits are intangible in the form of transparency for auditing and governance.
But there are also bottom-line benefits to bringing the retail risk management process in-house. A suite of models developed by consultants can cost up to $1 million. And the process is inflexible; iterative changes cost money and take weeks. “A vendor doesn’t know your data,” Duhon says. In-house, where the data is prepped, you’ve got a clearer view into the process. In addition, keeping model development in-house develops greater understanding of customer trends and behavior while promoting institutional knowledge.
SAS Enterprise Miner gives Scotiabank the capability to constantly improve existing models and build new ones to keep up with changing regulatory requirements, a competitive financial market and dynamic customer needs.
Given the volume of work on retail risk modeling for Scotiabank, there is substantial time savings from being able to build in-house and iteratively update the models.
“Compared to outsourcing, it costs less to do it in-house with an automated tool like SAS Enterprise Miner,” says Duhon. “Additional benefits include increased flexibility around model development processes, regulatory transparency and talent retention. With the efficiency that automation drives and in-house talent, we can be on the forefront of best modeling practices in our highly regulated and competitive environment.”
Reducing the cost and time of creating retail credit-risk models.
- Faster model development and iterative changes.
- More transparent regulatory compliance.
- Lower costs.
Scotiabank is one of Canada’s largest banks and one of the top 50 worldwide financial institutions.