Model validation: essential for the bank and for society

“How great is the risk that clients won’t pay back their loans? What losses can we expect? And what is the maximum amount we can lose? Those questions are essential for the bank. And they can only be answered by sorting through countless data and combining them in the right way. That’s exactly what credit models do – and much faster than humans could do it. The information from those models helps the bank to determine the amount of loans, whether or not to accept clients, and to determine how much money they should put into a special safety buffer.”

Internal checkpoint

“The safety buffer keeps banks from collapsing during a financial crisis, and prevents financial problems for millions of companies and private individuals. So it’s extremely important. The European Central Bank has determined that each bank’s buffer must be one tenth of its total expected losses. The same ECB checks the validity of the credit models the bank uses to calculate that expected loss. If the model isn’t valid, then the makers need to go back to the drawing table and come up with a new model. As a bank, you want to prevent that kind of rejection, because it costs a lot of time and money.”

“For that reason, and because Rabobank wants to accept and serve the right customers, we first check the credit models we’ve developed internally before the ECB takes a look at them. My department, Credit Model Validation, is responsible for conducting that check. We analyze every model to see how it obtains information from data. If we notice something missing, we’ll find out what could go wrong as a result of the missing information and how bad that would be for the bank. It doesn’t always need to be perfect, but a model should work properly. If that’s not the case, then we think of things the developers could do to improve it.”

Good feedback

“Our team consists of around 30 people. They’re an important reason why I enjoy going to work in the morning: they’re open, helpful and extremely sociable. Those are important factors in teamwork, including our cooperation with the Credit Model Development team. We share a focus on different types of credit models and the use of programming languages Python and SQL. Where the individual validator examines multiple models, the developer concentrates on a single type of model for a single type of portfolio. So they know more about that specific model, and they can sometimes get irritated when we suggest corrections. Fortunately, everyone understands that we’re all part of the same team, share the same goals, and have to work together.”

“How we as validators give feedback also makes a big difference. Technical skills alone aren’t enough in this job; social skills are at least as important. A friendly, collegial dialogue motivates people more to make corrections than a reproachful tone. That kind of dialogue also makes it easier to find out if the correction is really necessary, or if better options are available. We’re sharper when we work together. And we have to be sharpest in the validation phase. There’s a lot of pressure to get it right: it’s the last check before the credit models go to the ECB.”

Great work culture

“I always thought that working overtime was the norm in the financial sector. But our team is actually expected to avoid overtime. Because working long hours makes you tired, and that increases the chance of mistakes. So managers encourage us to have a healthy work-life balance. I think that’s one of the great things about the culture at Rabobank. Just like the fact that everyone has plenty of opportunities to develop. Every employee has an annual budget for training courses. Rabobank understands how important it is for employees to continue growing and stay satisfied, both for them and for the bank itself.”

“I love creating analyses, coming up with solutions and discussing the solutions with other specialists. Those are all things I get to do in my team. Plus, you regularly have the opportunity to take on extra challenges here. For example, until recently we didn’t have our own Credit Model Validation tooling with integrated statistical tests. I had the chance to lead the team that developed the tooling. Now our whole team uses it, which speeds up our checks and helps us make better conclusions. Projects like those make it even more fun to work at this department.”

Exciting step

“When I came to work at the bank in 2018, I wasn’t really a data expert. I was almost finished with my PhD in Applied Mathematics, but I didn’t see myself staying in the academic world. I thought it was too abstract, and with too much of a focus on new solutions when existing solutions are often just fine. So I applied for something else: a Young Professional traineeship at Rabobank with a focus on Advanced Analytics, with training in subjects like data analytics, machine learning and risk management. And to my surprise, I was accepted.”

“I’d done some reading beforehand, of course. I was drawn to the fact that Rabobank works to help society and the environment. I also recognized the importance of a strong bank for the economy and society at large. But I wasn’t sure what to expect from the traineeship. Or if I’d be suited to a job in the financial sector. Finance isn’t the same as engineering or mathematics, although the bank uses models and mathematical calculations. So it was an exciting step outside my comfort zone. But that’s what makes it so rewarding. And now I’m grateful every day that I took that step.”