Navigating IFRS 9 PMAs

Banks face a constantly evolving risk landscape, due to factors such as the COVID-19 pandemic, energy supply risks, geopolitical instability coupled with trade wars, high interest rates, inflation and climate change. Traditional modelling frameworks, which rely on historical data, are sometimes inadequate to address these emerging risks. IFRS 9 requires banks to set aside loan loss provisions based on future risks. This helps protect capital and manage emerging financial threats.

Under IFRS 9, banks must categorise financial assets into various stages based on their credit risk and calculate Expected Credit Losses (ECL) considering anticipated economic conditions and their prospective impact on the creditworthiness of customers. Sometimes the underlying models in IFRS 9 fail to capture evolving risks, particularly those that are new or changing rapidly due to inefficiencies in the model or the absence of relevant historical data at the time of model development. To mitigate such inefficiencies, banks use post-model adjustments (PMAs) in addition to model-driven ECL estimates.

The application of PMAs, sometimes referred to as 'management adjustments' due to the absence of a formal definition of 'overlay' under IFRS 9, remains a feature of credit risk modelling within the UK banking sector – however, their use has declined markedly since 2022, as illustrated below.

While the reduction in PMAs reflects decreased uncertainty driven by positive developments such as lower inflation, falling interest rates and strong borrower credit performance contributing to economic stability, it does not necessarily indicate reduced risk.

While the outlook may appear more stable in the short term, it is essential for banks to exercise prudent risk management and ongoing vigilance. They should continually adjust their ECL models to reflect emerging risks and ensure that provisions remain appropriate.

UK banks have heightened their focus on PMA unwinding mechanisms which are critical for mitigating balance sheet volatility and help to ensure that they can manage their financial positions effectively as they navigate the evolving economic landscape.

PMAs are essential to ensure accurate, comprehensive risk assessments and financial provisions. They can enhance model accuracy by incorporating expert judgement to bridge the gap between model predictions and real-world complexities, and to mitigate or address the following deficiencies:

Model deficiencies and data limitations

• Institutions struggle with data availability, consistency and quality, leading to potential inaccuracies in their ECL estimations. This is especially true for smaller institutions or those lacking robust data infrastructure.
• The ECL model itself is complex, requiring sophisticated statistical techniques and significant expertise to implement and calibrate models correctly. Incorrect calibration can lead to significant misstatements in impairment estimates. The complexity also increases the risk of errors and biases in the model's outputs.
• PMAs are needed to address model deficiencies primarily related to model inaccuracies (actual vs. predicted), events for which the model is not trained (e.g., high interest rates, geopolitical conflicts, COVID-19), and data issues (e.g., COVID-19 data, limited history, low defaults, new portfolios).
• PMAs are also needed to address model limitations due to lack of necessary granularity to identify sector-specific vulnerabilities, making it challenging to address the unique risks faced by different sectors.

Internal policy changes impacting modelled provisions

• Changes in a firm’s internal risk policies can alter the assumptions and methodologies used in models, leading to discrepancies between the models' output and the actual risk profile of the portfolio. PMAs can be introduced to refine ECL calculations and ensure they accurately reflect the firm's forward-looking assessment of credit risk.
• Policy changes that could potentially impact provision estimates would include those related to credit risk management, asset allocation, credit risk assessment methodologies, data collection, etc.

Macro-economic factors not captured by modelled provisions

• Models struggle to predict the impact of sudden and significant geopolitical shifts, such as wars, major political upheavals or unexpected international sanctions. These events can drastically alter market conditions and economic forecasts, leading to unforeseen losses or gains that are not reflected in initial provisions.
• The increasing frequency and severity of extreme weather events, driven by climate change, pose significant risks to businesses across various sectors. The long-term impacts of climate change are difficult to model precisely, and the short-term effects can be highly unpredictable.
• Shifts in consumer preferences, driven by factors including social trends, environmental concerns or changing demographics can significantly impact demand and profitability. These shifts are often unpredictable and difficult to model accurately.

Qualitative factors

• Some credit risks are difficult to quantify using purely quantitative models. Qualitative factors, such as management quality or governance issues, can significantly impact creditworthiness. A PMA can incorporate expert judgement on qualitative factors to adjust the ECL estimate.

Strategies for managing risks effectively include a combination of quantitative and qualitative PMAs, together with continuous monitoring and oversight to ensure that adjustments remain relevant and effective. Enhancement of models to capture risks better and address limitations is crucial, with solutions including robust governance frameworks, independent reviews, advanced monitoring systems, proactive risk management and dedicated teams with ongoing training.

To reflect accurately the heightened affordability and refinancing risks inherent in the current uncertain interest rate environment, banks should consider refining their PMA quantification methodologies. This may require a shift from less precise approaches, such as portfolio-level scalars or aggregate ECL level adjustments, in favour of a more granular, component-level analysis.

The quantification of PMAs requires careful consideration of several factors to ensure the resulting impairment estimates are both reliable and compliant:

• Data availability and quality: this includes historical data on macroeconomic variables, as well as forward-looking information from economic forecasts and expert judgment. The quality and reliability of this data directly impact the accuracy of the PMA.
• The methodology for quantifying the PMA: several methodologies can be used to quantify the PMA, including:
  • Scenario analysis: developing various scenarios based on potential future macroeconomic conditions and assessing their impact on ECL.
  • Expert judgement: leveraging the expertise of credit risk professionals to estimate the impact of unmodelled factors by using non-quantitative factors to adjust model outputs, especially when quantitative data is lacking or when addressing emerging risks and trends.
  • Statistical methods: improving the mathematical and statistical components of risk models to increase their accuracy and reliability involves refining model inputs, assumptions and methodologies to reflect better the true credit risk within a portfolio.
  • Sensitivity analysis: plays a vital role in evaluating the robustness of PMAs and addressing potential biases resulting from the continued use of simplified models across diverse economic contexts. It helps to investigate the effects of changes in any key assumptions made on the derived PMA values.
• Documentation and transparency: the methodology used to quantify a PMA, along with the underlying assumptions and data, must be clearly documented to ensure transparency and auditability. The rationale for choosing a specific methodology and the justification for any key assumptions should be explained clearly.
• Validation and review: to meet regulatory expectations outlined in SS1/23 and address the increasing governance demands on Post Model Adjustments (PMAs), quantified PMAs should undergo rigorous validation and review processes to ensure accuracy and reliability. This may involve independent review by internal audit or external experts.

Model risk remains elevated and UK financial institutions are expected to establish robust policies and procedures for applying PMAs in line with PRA supervisory statement SS1/23. Adjustments should be documented within the governance framework.

As banks have continued to refine their PMAs, they have encountered various challenges throughout the process, including:

• Lack of standardisation: inconsistent application and varying criteria across teams can lead to discrepancies in risk assessments and provisioning estimates, underscoring the need for banks to refine PMA quantification to capture risks better by transitioning from broad, portfolio-level methods to precise, component-level approaches.
• Inadequate oversight: weak governance structures and lack of accountability can cause errors and misjudgements in risk assessments, highlighting the need for oversight of strategic redevelopment plans to ensure enhanced capabilities for better risk capture and reduced reliance on significant PMAs.
• Poor monitoring: ineffective monitoring and reporting systems can hinder the ability to track PMAs over time, affecting the transparency and reliability of financial statements. Improving internal reporting to offer deeper insights into loans or segments most sensitive to changes in recovery strategy can help in the targeted application of PMAs.
• Delayed response: slow responses to emerging risks and a reactive approach to risk management can increase vulnerability to novel risks. Proactive risk management strategies are essential to mitigate these vulnerabilities and ensure timely interventions.
• Resource constraints: operational inefficiencies can strain a bank’s ability to allocate sufficient time and effort to PMA processes, potentially damaging stakeholder trust and the firm’s reputation
• One size fits all approach: the application of uniform policies and procedures across diverse models and scenarios can lead to suboptimal risk management. Tailoring PMAs to the specific characteristics and requirements of each model is essential to accurately address their unique limitations and risks.

Banks should evaluate regularly whether their models need redevelopment to remain effective and aligned with current risk environments. KPMG’s modelling specialists can assist in estimating, validating and preparing robust frameworks for monitoring the evolution and eventual unwinding of PMAs. Support could include:

• PMA frameworks: advice on the development and implementation of robust PMA frameworks, encompassing governance structures, comprehensive documentation, and processes for regular validation and ongoing monitoring.
• Identification of PMAs: helping to identify PMAs through data analysis, model validation and process reviews – pinpointing discrepancies and improving the accuracy of relevant models and reporting where required, including identifying potential sources of PMA and recommending remediation strategies.
• Implementation of PMAs: assisting with the execution of quantitative and qualitative model adjustments. This could encompass a range of techniques including Significant Increase in Credit Risk (SICR), rigorous statistical testing, comprehensive sensitivity analysis, seasonality adjustments, precise model parameter refinements and informed qualitative PMAs.
• Alignment of IRB and IFRS 9 models through PMA: assistance with aligning Internal Ratings-Based (IRB) and IFRS 9 models by identifying discrepancies, developing consistent frameworks, reconciling methodologies and ensuring data quality, ultimately streamlining the adjustment process and enhancing regulatory compliance.
• Climate risk integration: supporting banks in integrating climate risk into IFRS 9 provisioning and PMA through the collection, utilisation, integration and analysis of environmental data. 

For more information, please reach out to: Andrew Fulton, James Philpott, Ayan Haldar or Himanshu Bagga.