RiskTech Forum

SAS: The Challenge of New Financial Standards

Posted: 3 July 2017  |  Author: Tom Kimner  |  Source: SAS

The Financial Accounting Standards Board has released its Accounting Standards Update on Financial Instruments – Credit Losses (Topic 326). This was the second phase of a global accounting standards change that began when the International Accounting Standards Board published their new standard, known as IFRS 9 Financial Instruments (a replacement for IAS 39), to the rest of the world in 2014.

The updated guidance relating to the measurement of credit losses (using a current expected credit loss estimation) represents a fairly significant change from the previous incurred loss model, which was in place for some time. For example, the incurred loss model recognized losses only when they have reached a probable threshold of loss. Many analysts have suggested this model had a negative impact during the financial crisis because potential future losses were recognized and provisioned far too late.

While the US standard deviates in a few significant ways from the international IFRS 9 standard, both revised accounting standards share an important feature: The calculation of the expected loss is now based on the life of the loan. This change increases the credit impairment over the affected assets (with some estimates being as high as a 35 percent), resulting in higher provisioning expenses and a negative impact to capital.

Working with uncertainty

IFRS 9 and CECL are primarily principle-based. As such, the implementation guidelines will likely change as consensus is built and clearer guidance is provided. Given the evolutionary nature of these standards, institutions may spend more time iterating model development cycles before transitioning to IFRS 9 and CECL reserving.

Because IFRS 9 and CECL development will be concurrent with ongoing stress-testing activities, it is likely that stress tests will need to include preliminary allowance models that are missing a full set of reviews. In a decentralized environment, additional model risk may emerge as methodologies and models used in stress test production become out of sync with the IFRS 9 or CECL models. Institutions must ensure that the allowance models used for stress testing keep pace with accounting standards development and that assumptions remain consistent between testing activities.

Additionally, lifetime loss models may rely on data elements not previously required for stress testing. As a recent report, Stress Testing: A View from the Trenches, revealed, only 17 percent of banks surveyed had a unified data repository established for stress testing. This suggests many banks are not prepared for the ever-increasing integration of finance and risk data required for IFRS 9 or CECL and stress testing. This may create data integrity risk, as analysts obtain the required data through ad hoc channels in order to meet their stress-testing deadlines. Institutions must ensure that the data used for stress testing ties with data used for IFRS 9 and CECL model development.

Meeting the challenges

To meet the changing demands of regulatory compliance and financial standards, organizations must look at solving these issues collectively and comprehensively across several areas. Otherwise, the overall costs of compliance can become burdensome, especially if various parts of the organization find themselves creating redundant or overlapping processes.

Consolidation of finance and risk data provides a number of benefits across the institution. A common data repository for stress testing and allowance estimation greatly reduces audit and reconciliation issues and model risk. It also lessens the overhead of managing multiple platforms.

An organization’s loss modeling and stress testing processes must be robust, yet flexible. They must be able to accommodate dynamic changes in allowance modeling while maintaining sufficient controls to withstand the regulatory scrutiny of capital adequacy assessment under stress. Institutions will find that consolidation of their stress test and allowance estimation platforms provides the best foundation to accomplish this goal.

A centralized model library provides a structure to maintain governance over the stress-test process as allowance model development and testing progress. With proper model versioning, management can maintain oversight and make informed decisions about new model integration into the stress-test process.

In addition, a flexible and modular model library facilitates sensitivity testing. This becomes an increasingly important tool to understand the larger impacts of changing models and assumptions over the IFRS 9 and CECL development cycles.

By iteratively swapping out model components, comparisons can be performed between model versions. Institutions can also quantify the overall effect of model refinements on their balance sheets while under stress. Sensitivity testing is also an effective tool for trend attribution analysis – a key component of stress testing. Prior stress tests can be rerun with updated model components, or vice versa, to quantify the effects of the ongoing model development on stress test results.

The changes in regulatory stress testing and in accounting standards have necessitated the need for historically independent risk and finance divisions to further integrate and collaborate. The importance of tight cooperation and sharing of data will only intensify as the new accounting standards begin to take effect. Departmental lines are blurring more and more with regard to regulatory and accounting compliance, and the interdependence of the various functions will only intensify.

The transition to IFRS 9 and CECL presents many financial and operational challenges – compounded by the numerous implementation details that continue to be subject to varying interpretations. Institutions must address these challenges holistically with flexibility to adapt over time. Regulators and investors need to know banks are effectively assessing and managing the risks in their portfolio. Proactively implementing a well-governed approach to managing data and models instills confidence that processes are in place to manage these critical issues now and in the future.