Since the crisis, stress testing has been a central driver of enhanced global standards for advanced risk management. The use of scenario-based modelling that began with estimations of credit losses and interest rate risk capital levels under defined macroeconomic deterioration has now shifted toward pre-provision net revenue modelling (PPNR).
Stress testing is an expensive element of regulatory compliance, with the largest US banks spending well over $100mm annually. So as banks in the United Kingdom are pushed to improve their PPNR competency, identifying areas where new regulatory spend can be leveraged for improved business outcomes is a prudent move.
The US’s leadership in this regard is notable. Since 2009, the Federal Reserve’s Comprehensive Capital Analysis and Review (CCAR) has required US financial institutions to master elements of loss modelling with steady progression into enhanced PPNR estimation. In Australia and Canada, greater regulatory scrutiny of deposit and portfolio analytics has generated similar progressions. This evolution is summarised by country/region in the following table.
In our view, no global financial institution has reached a point where PPNR modelling is seamlessly integrated into business management. In the most advanced institutions, PPNR models are being used to support capital management and allocations, but these models are only now being integrated into earnings forecasting and strategy development.
A Way Forward
For all banks, the skill set and experience necessary to move from Level I to Level IV does not come quickly or easily – large US banks have been at it for several years. But a set of practical steps can allow banks both to progress quickly and to avoid many of the pitfalls that ensnared earlier movers.
We see six dimensions that distinguish an institution’s progression from undeveloped capabilities to the point where PPNR modelling is driving the corporation. These dimensions are summarized in the following table and further described below.
- Organize appropriately. Moving from rule-of- thumb to driver-based models that withstand statistical scrutiny requires close coordination among finance, treasury, risk and business lines. Determining the roles and responsibilities for model ownership, validation, and execution requires flexing some new organisational muscles. Advanced banks have formed permanent modelling teams which combine statistical skills with strong knowledge of underlying business drivers. These modelling teams are mirrored by a validation team which sets standards for and evaluates progress toward effective model development.
- Improve data quality/granularity. The historical centrality of credit loss and interest rate risk modelling has helped create large pools of data to support the measurement of these risks. However, for PPNR modelling, long-term historical balance and rate data is typically more difficult to extract. Advanced banks have established data “ecosystems” to capture information in ways that allow for models to best meet business needs and support non-regulatory-generated explorations of past performance
- Raise statistical rigor. Oftentimes statistical approaches used to create rule-of-thumb relationships between a single variable and a balance sheet or rate element fail the tests for statistical validity that internal validation groups and regulators require. Banks take time to learn the intricacies of applying these tests in the PPNR context, which can be unfamiliar to those focussed on credit modelling.
- Broaden model drivers. Robust PPNR models incorporate management actions (like pricing, marketing and resource deployment) often have significant impact on results. Unlike credit modelling, PPNR outcomes are not foreordained at time of origination –factors like pricing and marketing spend affect an institution’s ability to acquire and retain relationships and maintain spreads.
- Incorporate business engagement. For many banks, the earliest PPNR estimates are driven by management assertions. As banks move from assertion to quantitative modelling, a central goal is to ensure that businesses remains closely involved in hypothesis generation and results review –providing a critical “push and pull” where business insight can drive model development forward while statistical findings sometimes provides the business new information about key drivers.
- Integrate into bank management. Regulatory scrutiny in the US and elsewhere has recently shifted toward ensuring that models used to drive capital management are connected to day-to-day business decisions. Banks are finding that they can improve the insights related to corporate forecasting, interest rate risk management, and pricing by using PPNR models to generate insights in each of these other areas. The results are improved business decisions, risk management and more prudent capital management.
PPNR modelling evolution at global financial institutions continues to evolve and improve. Although there are variations in pace, the overall direction is clear and the pace has stepped up. Banks in the early stages of PPNR modelling have three surprises in store:
- It’s a lot of work. Significant effort is required to build, validate, document and maintain these models which require major investment for even the most advanced institutions.
- It’s not credit modelling. There are more cross-currents of drivers than for credit risk, management actions have greater impact on results, and “worst case” scenarios are not always those scenarios where GDP is going down.
- It pays off. For institutions that have invested in these capabilities, there is a clear payoff outside of regulatory compliance. New data are supporting development of other management information applications, model insights are affecting business decisions, and new skill sets are being propagated into other business functions.
PPNR modelling investments are creating a world of information “haves” and “have nots”. Global institutions, often led by their regulators, are moving in one direction. Staying in front of regulations helps contain the sort of “emergency” efforts that can distract institutions from their day-to-day business.
Andrew Frisbie and Steve Turner are managing directors in Novantas, an industry leader in analytic advisory and solution services for financial institutions with presence in New York, Chicago, Toronto, London, and Sydney.