Banks with Ontology and Knowledge Graph Capabilities well-positioned to meet the new ‘CECL’ Accounting Standard

What is the new FASB standard for ‘Current Expected Credit Losses’ (CECL)?

The 2008 financial crisis exposed many weaknesses in the global financial system, including institutions holding inadequate reserves for losses absorbed in a stressed economic environment. Beginning in 2020, public business entities that register with the SEC will need to comply with a 2016 Financial Accounting Standards Board (FASB) credit loss accounting standard, known as “current expected credit losses” (CECL). The new standard will have a lasting impact on capital requirements and business strategy, requiring that firms estimate potential losses for each additional debt instrument in their books, regularly scan the horizon for shifts in credit risk, and set aside anticipatory capital corresponding to expected losses – a fundamental shift from the current backward-looking standard. Reserves are likely to grow (more than 50% by some estimates) and gain volatility. Under CECL, sound judgment – based on thorough analysis and robust data – will be a primary basis for capital reserve estimates.

Prepare for resulting data challenges

Financial institutions must use deeper and broader data to support CECL estimates than the current incurred loss framework requires. As an example, to estimate impairment based on historical performance within segmented groups and origination years, a bank may use loan-level vintage analysis; to do so, it must have access to reliable loan-level data such as historical balances, loan-to-value, risk ratings, structure, and charge-off and recovery data – much of which may not be easily accessible or high quality. A bank will need to acquire and maintain economic data to model defensible economic forecasts – perhaps including local or regional forecasted employment, consumer data and housing price data, for example.

Recent surveys find that many institutions are still in the planning stage for CECL, and that obtaining data necessary to support model development is among their most pressing tasks.

In the recent past certain banks have met CCAR or EU GDPR (General Data Protection Regulation) requirements with the use of graph-based technologies to connect inconsistent and fragmented data sources – key to establishing a comprehensive view of enterprise-wide data. Graph technologies are not new. Facebook, Google and LinkedIn are among firms established leveraging knowledge graph technology, empowering – in conjunction with machine readable ontologies – agile access to data across disparate sources. However many financial institutions are still leveraging traditional approaches to establish a central view on enterprise-wide data.

Start with a comprehensive data readiness assessment

Firms that fail to adequately organize, validate and ensure tight control of their data put themselves at risk during audit and with regulators given the reality of these new requirements, potentially to significant shareholder detriment – irrespective of the opportunity cost to business strategy from failing to bolster general analytical capacity.

It follows, then, that establishing a clear understanding of data requirements, availability, quality, accessibility and control via a readiness assessment should be an early and high-priority CECL transition activity to prevent such negative impact to financial statements when CECL becomes effective.

Organizations should also evaluate the use of modern data technologies – markedly advanced in recent years – as an option to connect and analyze data required by CECL as part of their readiness assessment.

In planning a data readiness assessment activity, the following critical steps should be included:

  • Gather data reflective of the firm portfolio. What data is required for CECL? This will be intrinsically tied to the methodology chosen. Building “reasonable and supportable” estimates necessitates loan-level data going back at least one credit cycle, current loan-level performance, and economic forecast data relating to credit risk. These core data requirements additionally influence model and technology decisions.
  • Define the data consumption target state and evaluate the accessibility of all required data. Include the controls, processes, architecture, roles and responsibilities across relevant business units and corporate functions. Crucially, consider how to least intrusively connect fragmented loan, mortgage, and external data sources. Assess how existing data access layers and semantic ecosystems may be leveraged to simplify this challenge.
  • Identify disparities between the current data state and the target for CECL. This exercise clarifies the necessary actions to acquire, manage, and leverage the data to meet the CECL standard. Gaps within data coverage, architecture, and quality are typically identified and require resolution for CECL compliance.

Recommended next steps for CECL compliance?

Institutions regularly cite the most significant challenge in complying with CECL is successfully acquiring and managing data from internal and external sources.

  • Partnering with the finance and risk teams within the organization to assess CECL data readiness is recommended as soon as possible.
  • Data management and transformation initiatives to meet other regulatory needs may already be underway and should be considered to support CECL data collection and validation efforts.
  • Establishing a robust and accurate historical data set is a complicated task that usually requires a large investment. With early action, the enterprise data team may proactively prepare to close the gaps and incorporate CECL related tasks into 2018 and 2019 initiatives. This effort is significant, requiring key resources, data procurement, and the build out of new technical and data management capabilities.
  • Finally, institutions already utilizing an ontology and knowledge graph within their data ecosystem will find their hard work greatly simplify CECL compliance. For those not yet leveraging such solutions, engaging IT and architects to do so is highly advised.

Element22’s knowledge in data management best practices and modern data technologies, and FI Consulting’s expertise in expected credit loss modeling are suited to structure a smooth transition to CECL. A common-sense approach simplifies the complexity of CECL and helps uncover ways to leverage and enhance existing data management capabilities to create and execute on a tailored CECL model.

Feel free to contact us here for a free workshop to discuss your strategy for CECL compliance.

 

A market commentary provided by

Thomas Bodenski, PartnerElement22

Mark Jordan, Senior Manager, FI Consulting

The opinions expressed are as of March 2018 and may change as subsequent conditions vary