Financial risk management has witnessed a revolution in advanced analytics over the last 20 years. New analytic techniques combined with quantum leaps in computing power provide risk managers with an array of tools to better measure and manage risks of various types. Integrating analytic capabilities offers great potential for risk managers and their firms; however, it remains an elusive goal for much of the financial industry.
The natural evolution of risk models sprung from the need to improve the measurement of individual risk types such as credit, market and operational risk over time. Such efforts typically materialized from specific regulatory requirements such as Basel capital or business mandates. Depending on the asset class, risk models are further differentiated by type of risk, such as interest rate or credit risk. The increased range and complexity of models has created issues with integration success. Differences in obligor type, e.g. consumer, commercial or governmental entity greatly influence the type of data available to build analytic tools. Changes in the availability and collection of such data has expanded the variety of risk models used at financial institutions today, but added more challenges to integration. In addition, institutions that have grown by way of acquisition may also experience integration issues in their risk models by way of separate pre-merger development across acquired business units. Key questions for risk managers to consider then are what do integrated analytics mean for your firm, why are they important, and how can we move toward an environment where analytic methods are harnessed in an integrated manner?
At the enterprise level, risk components should be modeled, evaluated and aggregated in a consistent manner in order to present a comprehensive, integrated view of risk exposure for the firm. Moreover, risk should be represented consistently across the asset’s lifecycle. For instance, the assessment that is made at origination should align with a firm’s loan loss reserving processes, pricing, portfolio and capital management, asset valuation and default and collections strategies. Due in part to data limitations, risk type, business unit or model differences, forming an integrated view of risk that meets the consistency standard is challenging but achievable. Risks are often not stand alone. As a result, in some instances, integrated views are also complicated by the competing nature of risk.
To illustrate the competing risk issue and its importance to an integrated analytics framework, consider mortgages either held on-balance sheet as whole loans or as assets underlying a mortgage-backed security that are common to most commercial banks.
Such assets effectively present the borrower with the option to either default or refinance (prepay), hence posing both credit and interest rate risk to the bank. In this case credit and prepayment risk compete somewhat against each other and for that reason it is crucial to have a linked view of these risks.
Individual borrowers have the choice to either default or prepay, but not both. As such, the borrower’s ability to default or prepay is conditioned on his prior exercising of these options. Therefore, projecting default propensity without consideration to the competing risk of prepayment will result in separate estimates that when combined will not accurately reflect the realized cash flows of the loan pool over time. Moreover, common modeling limitations such as using different data sets to estimate defaults versus prepayments will assuredly compound the inaccuracy. Integrated modeling should serve to reduce error and bias in risk management.
Reinforcing this point, the figure below represents how credit score, a key driver of credit and prepayment risk, affects risk-adjusted return on capital (RaRoC). Borrowers with lower credit scores, all things equal, tend to have higher default risk. Alternatively, these borrowers have a lower chance of prepaying. Borrowers with a greater likelihood of refinancing their mortgage tend to have higher credit scores and also lower credit risk. Focusing on managing each risk in isolation of the other can result in suboptimal financial performance as depicted in the figure.
The current economic climate accentuates the need for an integrated view of analytics as well. The prospect for higher interest rates going forward after a decades-long trend of lower rates is as high as it has been in years in part due to the current Administration’s stated expansionary fiscal policy agenda.
This raises the potential for extension risk in bank portfolios. In fact, over the last several years, many community banks have witnessed an expansion of longer duration assets as a proportion of their balance sheet while longer-term liabilities have not grown at a similar pace1. Regulators such as the Office of the Comptroller of the Currency and Office of Financial Research have flagged rising interest rate risk as an issue to watch in the coming years.
Prior to the recent Great Recession, modeling consumer credit risk was challenged by the lack of data relating to prior similar periods of risk. Long periods of expansion and limited loss experiences but for a small number of prior events (thrift crisis, Great Depression, etc.) dramatically limited the data available to accurately project credit losses as markets turned. Similarly, one problem facing risk managers trying to forecast and manage for the potential for rising interest rate risk today is the limited historical data on long periods of rising interest rates. Creating and applying interest rate risk models from time periods that may not be representative of current and prospective credit conditions, or don’t consider regime changes or other structural differences in markets, will require a great deal of judgment to be applied alongside the modeling. To be truly useful, such efforts demand an integrated approach to risk modeling and analytics.
The benefits from pursuing an integrated risk analytics framework are many. Lifecycle asset management is critical to effective risk management. An integrated risk analytics platform enables the business to dynamically manage current and future business better by integrating views of risk at the front-end with middle- and back-end finance and risk management activities. Asset pricing improves by forming a more accurate view of the combined risk of the asset. Further integrating risk models within and across asset types offers the opportunity to optimize portfolio risk-adjusted returns and hence improve capital allocation across the firm.
The quest for integrated risk analytics starts with an understanding of how the inventory of models used by your firm relate to each other. For example, are different models used for measuring credit risk in the asset-liability management and loss forecasting processes? If there are multiple businesses originating similar assets, understanding differences in models used to measure similar risk will be important, as well as ensuring alignment of models consistently across the firm. Adopting an integrated risk analytics philosophy across the organization is an essential ingredient to successful enterprise risk management surveillance and risk mitigation.
1FDIC reported that the proportion of assets with maturities over 5 years on community bank balance sheets was significantly higher than for other banks and that their holdings of these assets had risen steadily in recent years. Source: FDIC, Quarterly Banking Profile, Third Quarter 2015.