Henri Wajsblat, Anaplan’s Head of Financial Services Solutions, interviewed Chappuis Halder Director Vincent Wiencek about implementation challenges of banking stress tests, what banks should expect from technology, and the benefits of the Anaplan platform.
1. Among all banking regulations, stress tests seem to be a major concern. Why are stress tests a real challenge for the banks?Stress testing has always been used by financial institutions as an efficient way to understand their sensitivity to external or internal risk factors. In addition, Basel regulations on capital requirements have enforced the use of specific stress testing (ST) approaches on credit, market, or liquidity risks and global approaches for measuring business risks. Since 2014, the European Banking Authority (EBA) has designed global stress testing approaches, which encompass all risk factors that financial institutions may be subject to. These include financial risks (liquidity, rate, net income), credit, and market, and also non-financial risks (reputation, compliance, or operational risks). Given the significance of the stress tests for senior management, banks have developed their own internal global stress testing approach, which may differ from the methodology defined by EBA, to tackle business model specificities. Internal stress testing is now a major tool to understand the strengths and weaknesses of a bank’s business model, and then adjust its strategic business plan or hedging policies. Now they have acquired sufficient maturity, banks need to redesign (or design) their stress testing production framework to achieve the following benefits:
- Reduce operational risk caused by largely manual processes
- Reduce time-to-market stress test production to improve decision-making capabilities and relevance
- Optimize the workload of teams in charge of production and analysis
- Improve the precision and robustness of the exercise (reduce the number of proxies, improve granularity of information, etc.)
2. In your view, what should banks do right now and how does it relate to technology?Our conviction at Chappuis Halder and Lionpoint Group is to take advantage of the absence of the EBA stress testing exercise in 2019 and start ambitious and transversal programs to improve the stress testing framework. Upcoming improvements rely on the following key pillars:
- Reviewing the global production and coordination process The stress testing exercise may be one of the most transversal exercises ran in a bank. It involves the support functions in charge of balance sheets and income statements (management control, asset and liability management, treasury, and risk) and the lines of business in charge of activity forecasting. As a consequence, the level of operational risk linked to this heavy coordination process between multiple teams may be very high. The first major benefit of industrializing stress tests in dedicated tools is a more resilient and consistent workflow management between different contributors leveraging versioning, audit trail, and collaboration functionalities.
- Strengthening the methodology used to forecast and stress different items of the balance sheet and income statement
Recently, regulators have encouraged banks to strengthen their approach for business forecasting. As a result, financial institutions have required significant investments to either develop or upgrade their modeling capabilities:
- Risk departments, under capital requirement directives, already have strong and consistent models for the asset side of the balance sheet; they will only have to extend their models to cover distressed situations (Probability of Default scales may not be the same under stressed situations)
- Asset and Liability Management (ALM) departments have a strong culture of model development, notably through behavioral models on the balance sheet (early repayments, embedded options, etc.). Because liquidity is a key metric in stress testing outcomes, ALM departments have engaged in review and benchmark exercises of their models
- As stress testing covers both the balance sheet and income statement, banks need to improve P&L forecasting capabilities, especially for net trading income, fees and commissions, or even cost evolution models. Business-as-usual budgeting exercises will benefit from these models
- Significantly improving data quality Data quality really is a pain point for the financial institutions and patches cannot be an adequate answer. A strong data quality framework, as highlighted by Basel Principles for effective risk data aggregation and risk reporting (BCBS 239) is required for stress testing approaches. Stress tests are highly sensitive to the quality of input information (both data and models).
- Improving underlying systems Different implementation strategies may be considered for stress testing implementation:
- A “micro service” strategy that defines common and shared data models and principles and relies on a strong autonomy of data and model providers
- A “centralized ST” strategy that consists of the centralization of data and models in a dedicated stress testing tool
3. For the last decades, finance and risk architectures have been quite siloed. Do you think current stress testing requirements will drive more synergies between business-as-usual (BAU) finance processes and stress tests and more collaboration between finance and risk teams?Absolutely. In the past, most financial institutions designed their systems in silos for various understandable reasons:
- Risk systems were designed to comply with capital requirement directives
- Accounting systems improved to comply with accounting standards (IAS, IFRS)
- ALM systems addressed liquidity issues (and sometimes rate risk management)
- Marketing and front-office systems were strictly focused on sales efficiency purposes
4. What core functionalities would you expect from technology to integrate BAU forecasts and stress tests in a same platform?Technology is surely a core component of a strong stress testing framework. An effective stress testing system will rely, in our view, on the following core capabilities:
- A strong workflow management including user level permissions to engage finance, risk management, and ALM teams on a single stress testing platform while preserving their respective areas of ownership on data, models and engines implementation.
- Data management functionalities that allow interfacing with various data sources (databases, csv files, REST API, etc.) and to interoperate with statistical libraries or systems such as python, Pandas, numPy, TensorfFow, SAS or R
- End-to-end modeling from data inputs to synthetic stress tests outputs
- An agile platform providing real-time recalculation of the data based on a change of assumptions or data inputs and what-if analyses leveraging scenario modelling and simulations.