Challenges and practices of effective existing model landscape maintenance

Challenges and practices of effective existing model landscape maintenance

By Darius Grinvaldas, Head of Credit Risk Modelling 1 (PD Maintenance) / GRM Vilnius Site Leader, Danske Bank A/S

Interview ahead of the Model Risk Management Europe Summit
Can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?

In order to describe my professional profile I would like to put some highlights on my education first of all. From the high school years I was really focused on exact sciences like physics, mathematics, chemistry, and also very strangely music. Physics was the biggest passion and I was constantly showing the best academic record, which I proved with the one of the best national exam scores in the country. If I went back in my memories retrospectively, could openly say that probably learning physics had this highest degree of required creativity and “music” with various numerical problems. Thinking and considering about the professional career choices using this two-dimensional approach (“music with numbers”) after graduating the school encouraged me to choose Financial and Actuarial Mathematics at Vilnius University in Lithuania, as rapidly developing financial industry in the Baltics was attracting with plenty of creative and challenging work opportunities in the Scandinavian banks establishing their Baltic branches.

After master studies (combined with some financial analyst roles in a few local market companies) I dived straight into credit risk management discipline, first of all in Danske Bank A/S Lithuania branch, at a later point moved to Danske Bank Group Risk Management (GRM) which launched the global team in Vilnius, Lithuania, where my professional specialist focus was mainly on IRB modelling matters and various projects. After a number of successful years and various professional contributions in building global organisation I took Vilnius GRM Site leadership role and also started heading the global team responsible for the maintenance of the existing PD models for the whole Danske Bank Group, which involves extensive professional collaboration with internal Model Validation function, Model Owners and regulatory authorities.

What for you are the benefits of attending a conference like the Model Risk Management Summit and what can attendees expect to learn from your session?

The most important benefit is to have a possibility to exchange opinions, knowledge and points of views with the number of European peers working in the industry on various Model Risk Management and Regulatory agenda in risk modelling related topics. Also getting valuable inspiration and highlights for the future improvements on many areas where the organisational maturity and general working practices related with MRM could be improved and elevated to even higher level.

During my session I would like to bring some model developer’s experiences and points of views not from the position of strength in MRM area, but from the position of improvement and many unanswered challenging practical day to day questions when organisation sets very clear direction to strengthen MRM discipline in establishing the new roles, structure and more strict model risk control and governance environment following increasing regulatory demands and general industry trend.

In your opinion, why is working independently from model validation team so important?

First of all as a model developer I might think about the trivial answer to this question- this is simply a regulatory requirement, but risk management is not only the set of minimal regulatory requirements and the ways to comply with them. Risk management is carrying a way more broad measures and views for the financial institution to make the prudent and calculated actions and strategic decisions. In order to be aligned with that, acting independently and ensuring the right level of the challenge and transparency internally is a simple necessity.

Independence in respect of the validation function in practice is usually ensured through direct reporting line to the risk committee structures and the CRO. The validation unit owns the validation process and methodologies. The annual validations of the risk models are reviewed by Risk committees.

However, in practice there might be various professional dilemmas as well, like transition of resources thought different teams, intentions to deliver the modelling products faster and more efficiently by following strategic plans omitting an adequate challenge, knowledge and experience shortage in the different risk department areas. All those dilemmas and even more prudent organisational oversight should be mandated to the third line of defence in the company to maintain the earlier mentioned reasoning.

How can credit model risks be mitigated?

Model risks are usually mitigated by simply tackling the weak and erroneous areas or assumptions of the risk parameters estimation in Pillar I world by considering various different quantitative tools (partial or full redevelopment, calibration, re-estimation, margin of conservatism) in order to compensate the existing degree of uncertainty or risk underestimation.

Another option could be an organisational implementation by educating the model users to adjust and set more conservative outcome due to various modelling uncertainties.

And for the residual model risk the mitigation using additional REA amounts should be quantified and considered (either in Pillar I or Pillar II risk buffers) in order to keep the institution on a minimal level in terms of the model risk exposure.

How can we ensure effective collaboration with model owners?

My understanding of Model Owner concept is that both Model Risk Management and Model Development functions need very clear reference to solid user knowledge perspective.

Model Owners should be on-boarded organisationally to the risk committees with the right level of mandate to support the challenge by MRM regarding various modelling processes, developed final modelling products or components in those products. Model developers should experience the right level of direct responsibility separation from the developed modelling products due to various very natural decision and assumption related biases encountered during the development processes. Model owners should be aware of those biases, subjective modelling decisions, and assumptions, having the right level of technical understanding and technical education related with the modelling processes.

Positive influence to this collaboration could be achieved by keeping maximal level of transparency regarding various model issues and potentially needed mitigation, standard model changes and activities prioritisation, timely status and progress reporting, maintaining the right level in educating on evolving new issues or regulatory constraints, sharing the decision mandate on setting or changing prioritisation, changing mitigation strategies on still open model issues and etc.

How do you see the impact of Model Risk evolving over the next 6-12 months?

I clearly see model risk further evolving to one of the top risks in the financial industry even in this short time as unleashing data driven solutions and various analytics is the strategic priority for the absolutely most of the financial institutions. At the same time the market awareness, understanding and discipline regarding this direction is rapidly evolving and mimicking this trend, so Model Risk related topics, problems, and various solutions are discussed even more and more extensively. So it could be concluded that leveraging decisions and business strategies on models without clear understanding regarding the downside of various potential effects resulting from the model risk impact should not be a discussion point taking into account this time frame.

If we take into account some exogenous effects like current political uncertainties in Europe, then this is a different discussion, and even IRB credit risk models could step into higher model risk uncertainty zones, which for sure were not foreseen and quantified during the developments.

So the subjective view is that the impact should be somewhat neutral or slightly upwards mainly based on mentioned evolving political uncertainties.

Meet Darius Grinvaldas at…

Sign up for your free Risk Insights account: