Why Model Risk Management Matters in Asset Management
Why Model Risk Management Matters in Asset Management
In a financial industry which has embraced quantitative models and advanced analytics along its entire business and operations, and therefore sharply raised its dependency on them, it should come as no surprise that the risk of model failures has become an increased point of attention for both regulators and market participants.
In asset management, models are ubiquitous in day-to-day business, such as in supporting or even taking decisions in the investment, portfolio management, risk management, and finance functions. Therefore, it is only a logical consequence for asset managers to manage the potentially dire consequences of erroneous models and to come to grips with a constantly-shifting model landscape they are held accountable for.
An asset manager faces model risk - the risk of loss in earnings, capital, liquidity, or reputation - whenever their models have deficiencies or implementation errors, or are misused in decision-making. That is why it is so important to get predictive modelling right the first time around.
Today’s Challenges of Model Risk Management for Asset Managers
Lack of Insight provided by Traditional Models
Models put in place in the past often are not fit for purpose today. The market environment has changed enormously, and at an ever-increasing pace. There are several factors at work: geopolitical, social, and environmental, to name a few. As such, traditional models may lead to inadequate results if not reviewed and validated frequently. Outdated frameworks simply do not apply to the unchartered waters of today’s financial landscape and products.
Asset managers are further exposed to model risk thanks to a multiplicity of client interfaces, service providers, data, systems, asset types and modelling techniques. Add a lack of clear oversight and an increase in organisational fragmentation, and you are seeing unprecedented complexities in the model landscape and related impact.
Both in North America and in Europe, regulatory oversight is becoming more stringent, notably in models used for liquidity risks measurement and mitigation, as well as for derivatives in counterparty risk assessment.
For bank-holding asset managers and banks with a US footprint, regulatory demands placed on Model Risk Management (MRM) are growing. Asset Managers have to find the right balance between meeting the regulatory requirements imposed by the bank-holding and the added value of a tailored MRM framework. Other jurisdictions are expected to follow in the same direction.
Evolving Client Expectations
Client expectations today are very different from what they were ten years ago. Whereas clients used to go into a bank or an organisation’s office, they are instead now interacting online or via mobile. They expect to have advisory services that provide clarity about what they should invest in. They also expect new levels of transparency. It is not enough anymore for the financial industry to have a ‘black box’ behind predicted outcomes. Clients want to be informed on the value the industry is providing in return for the fees they pay to financial services providers. This heightened level of scrutiny and transparency demanded by investors (both retail and institutional) puts additional requirements on explanatory features of models applied.
One of the main sources of model risk is a model deficiency. Deficiencies may happen if you use an inappropriate methodology, or base a model on wrong assumptions, fail to code and implement the model according to specifications, fail to test a model for robustness and soundness, use faulty technology, feed a model with faulty data, or simply use a model for an ill-suited purpose.
Not only do such deficiencies lead to wrong decision-making and judgement of risks, they can have a significant reputational impact.
Methodology Model Risk Management in Asset Management Vs. Banking
Although best practices and components of the MRM framework are universal, MRM in asset management faces its own peculiar challenges.
Firstly, asset managers have their own structural set-up with asset management companies, fund management entities and funds using internal and external managers, vendors and service providers (custodians, administrators, research, etc.), which add to the complexity of the total model landscape.
Secondly, different asset management ownership structures, i.e. bank-owned, insurance-owned, or standalone, result in fairly different regulatory face-offs. The asset manager might be subject to internal Group policies that are driven by banking, insurance, or targeted asset management regulations.
Also, large asset managers are often global in nature, investing or serving investors in numerous jurisdictions, which can each raise regulatory requirements to be fulfilled.
Finally, contrary to banks and insurers, asset managers are mostly exposed to non-financial risks. They do not face direct financial losses derived from their investment decisions. However, would fund performance suffer, negative impact on redemptions can follow, affecting the asset manager’s earnings and reputation too.
The Changing Role of Analytics and Quantitative Information
To manage new risk and regulatory themes and compete for growth, Asset Managers are increasingly reliant on more diverse and comprehensive data (“Big Data”), advanced analytics and quantitative tools to support client-facing activities and investment decisions. There are several factors to be taken into account when developing advanced analytics (such as machine learning, deep learning, or artificial intelligence), including:
- Regulatory and accounting reforms
- Change in the landscape of assets and risks
- Change in financial markets infrastructure
- Evolving customer expectations and behaviour, and
- Digital disruption.
On balance, models provide insight, but also inherent exposure in a complete reliance on data as well as advanced analytics techniques. The more such models are incorporated into investment risk estimation and management, the more potential there is to misuse this information and techniques if they are not properly understood and integrated in a well-defined MRM framework.
Figure 1: Risk Dynamics end-to-end Model Risk Management Framework
Approaching Model Risk Management Today
The First Practical Steps towards Model Risk Management
While the specifics of model risk may very much depend on the business model of an asset manager - such as relative importance of internal models vs. vendor models or the use of models at asset management company and fund management levels - the overarching building blocks of MRM remain the same.
Establishing a model inventory
Maintaining a clear overview into all analytics and quantitative tools used across the institution is a prerequisite to managing model risk. An asset manager will also gain efficiency in identifying which models matter most to management decisions. Moreover, understanding the interdependencies between different models is crucial, and will help to identify models that are centerpieces and provide inputs to other models.
For example, you might have several yield curve models to estimate interest rate in the future, linking differently between other models.
Whatever the model purpose, tiering model risk is another important initial step in model risk management. Using both quantitative and qualitative criteria, models can be classified by materiality in business decisions. Model tiers will then enable the tailoring of the model risk strategy and mitigation, as well as prioritising resources in model validation.
The materiality of a model is driven by several criteria, among which the exposure, regulatory requirements, model usage, external stakeholders, and model complexity rank highest. The models of highest priority which end up in the top tier will be subject to thorough and frequent validation reviews, whereas both depth and frequency of validations will be gradually reduced as you move to lower tiers.
Then, once you have mastered these first steps, at least for the main models and quantitative tools which can affect your financial or non-financial performance, or your regulatory standing, you can start to establish the next building blocks of a robust MRM framework: strategy, governance, processes, evaluation, management and mitigation tools, etc, always keeping in mind the structural differences of the asset management industry vs banking and insurance.
About the Author:
Maribel Tejada is Senior Consultant at Risk Dynamics. Within Risk Dynamics Maribel focuses on Risk Appetite advisory and Non-Financial risks, mainly franchise risk and reputational risk. Click here to email Maribel.