At a macro level, the Bank monitors industry trends, newsfeeds, and policy changes, which may have a bearing on the portfolio quality. To that effect, the Bank has subscribed to various information databases from external vendors. Using this information, detailed analysis/risks are identified, which collectively form a part of Level 1 EWS framework, covering macroeconomic risk analysis, and are reported internally at regular intervals.

Detailed Key Risk Indicators (Level 2 triggers) are defined to assess the performance on various parameters such as Incremental Overdue, PAR30+, PAR90+, Early Delinquencies, Quick Mortality, Non-Starters, and Collection Performance, among others. These metrics are analysed at various levels and dimensions, such as branch, state, and product, among others. These help identify underperforming segments and pockets for corrective actions early on. The portfolio level triggers were standardised during the year for base-level internal reporting using advanced visualisation tools, and these reports are increasingly automated for ease of reference.

In the previous year, as part of Level 3 EWS triggers at a borrower level, the Bank leveraged its internal capabilities in automating early warning signals based on unique behavioural patterns exhibited by the borrower. These triggers/ exception lists were developed in compliance with the RBI Guidelines on EWS. The automated exception reports enable real-time monitoring of problematic accounts and take necessary course corrections early on. The Bank intends to enhance its monitoring mechanism of these triggers through a system-based solution. This mechanism enables the Bank in the classification of Accounts Under Watch (AUW)/Red Flagged Accounts (RFA) and aids in the classification of frauds. These triggers were carefully evaluated and back-tested at periodic intervals to explain the default propensity. During the year, the Bank worked towards implementing an application to capture the feedback from field officers on such borrowers. The feedback was also evaluated to identify commonalities and key risk drivers. Using these insights, the Bank intends to strengthen its negative profiling framework to restrict exposure build-up in high-risk cases.

For its Institutional Lending portfolio, the Bank has an EWS framework in place which factors data from financial statements, trends in compliance with loan covenants, external rating reports and newsfeed, among others. During the year, the Bank enhanced its approach through the bifurcation of the EWS and AUW frameworks. The EWS assessment for its Institutional Lending portfolio is also expected to be undertaken through a software platform.

The Bank’s rating model/scorecard landscape have evolved through the years since its inception. Concerted efforts were made in analysing the Bank’s historical data to identify patterns and trends that can be used in business decision-making. To this effect, the Bank has developed application scorecards and behavioural models that can be used in decision-making and the computation of default probability. The priority with which scorecards and behavioural models were deployed was on the basis of the relative vintage and materiality of the credit segment.

On the Bank’s MFI/Group Loan portfolio, a new statistical scorecard (regression-based) was developed using the post-pandemic data during the year. This scorecard was subject to Out of Sample (OOS) and Out of Time (OOT) validation and was found to be statistically significant. This scorecard will be embedded in the Bank’s Loan Origination System for further testing and stability. The Bank intends to explore the usage of these scorecard outputs in pricing and in fixing delegation at the senior level for high-risk cases initially. The transition will, however, be undertaken only after multiple rounds of out-of-time validation.

Along similar lines, a rating model for an Individual Loans portfolio has been designed using internal data encompassing attributes like demographics, repayment trends and bureau-related variables. Variables were selected on the basis of statistical significance, which best explains the propensity to default. This scorecard was also subject to annual validation and was found to be satisfactory. The scorecard is embedded in the Bank’s LOS platform and continues to be used in credit decision-making and pricing. During the year, the Bank commenced the development of challenger scorecards for the IL portfolio in a bid to improve accuracy.

For Vehicle finance, statistical scorecards using the Bank’s data were developed during the year, separately for New-to-Credit and New-to-Bank/Existing-to-Bank customers. However, since these scorecards were developed using limited internal information, the Bank additionally procured bureau scorecards in a bid to strengthen the credit appraisal process. Decisions on credit underwriting and pricing are taken on the basis of a composite score generated by these scorecards. The Bureau scorecards will retire gradually, along with internal scorecards, as and when data is available.

The pandemic resulted in an acute distortion of data at Bank and Bureau level preventing any meaningful modelling in normal circumstances. Since the MSME and Housing Loan variants are disbursed over a longer time horizon, the inability to capture the performance over longer time horizons in the current portfolio, coupled with the issue of fewer data points, required modelling activity to be postponed up to FY 2026 for these loan segments. In lieu of statistical scorecards, the Bank continues to use credit rating scorecards developed using expert judgement at a product level for MSME, and on the basis of customer segment level in Housing Loans. These scorecards will aid in building a dataset on critical parameters and demographic details which can be used in statistical modelling in the future.

During the year, the Bank also put in place a credit risk model development and validation policy to provide an overall governance framework for model development, monitoring and validation aspects. This policy is a landmark milestone, wherein the learnings from the development and validation of existing scorecards, coupled with best practices in modelling as per BCBS guidelines, are incorporated to standardise all future model-related activities within the Bank.

The Bank has in place a detailed framework for computing Expected Credit Loss (ECL) as per the Ind-AS requirements and also for internal reporting purposes. A specialised set of models is in place to compute key risk factors such as forward-looking Probability of Default (PD) estimates, Loss Given Default (LGD) and Exposure at Default (EAD) for our MicroBanking and Housing Loan variants. For all other loan segments, a combination of simple transition matrix/roll forward rates or industry benchmarking is used in the absence of sufficient data.

It is pertinent to take note of the Discussion Paper released by RBI on the ECL framework for provisioning in January 16, 2023. While a mandate to transition from the current IRAC-based provisioning to an ECL-based framework is yet to be notified for commercial banks, the Bank has nevertheless worked towards preparedness to make the transition efficiently. The existing models are observed to be complying to the minimum requirements of the Regulator. Additionally, the Bank has commenced automating our PD and LGD models using internal capabilities. During the year, the automation project was completed for all the credit variants and is currently under testing for accuracy and scalability. It is expected to be moved to production in the ensuing year. The Bank, as part of good governance, continues to report ECL outputs internally.

A key learning from the pandemic, especially for unsecured lending products, was the need to monitor collection-related KRIs at a granular level. As a monitoring mechanism, the Bank introduced productivity trackers at an officer level against predefined benchmarks. These trackers serve as an important tool in tracking low performance at various levels such as officer, district, state, and loan quantum, among others. In the ensuing year, the Bank intends to leverage our internal capabilities in automating the exception reports/triggers for real-time tracking.

In a bid to further enhance risk oversight and identify lapses in credit appraisal standards, the Risk team put in place an independent Credit Risk Monitoring Unit (CRMU) in FY 2022-23. The initial focus of CRMU was restricted to reviewing high-ticket loans, identifying and analysing quick mortality cases and the root causes thereof, and providing early insights into vigilance to initiate investigations from a fraudulent angle. Much of the scope was restricted to the review of MSME credit proposals.

During the year, the Bank furthered the Terms of Reference for CRMU to include institutional lending, review of repayment rates, and Early Warning Signals review, among others. Findings of CRMU are analysed for adequacy in policy controls or to identify areas of weaknesses. Post deliberations at CRMU level, policy changes and proposals are made at CRMC for adoption.

Continuous analysis of all loan portfolios to identify potential areas of stress on the basis of geography, ticket size, branch/clusters, among others. Such analysis has helped the Bank set mitigants in the form of limits and caps on exposure. The Bank has also subscribed to various industry dashboards to benchmark portfolio performance.

All new products and processes (including enhancements) are subject to a mandatory comprehensive review. We continuously review and enhance our key processes to adapt to industry best practices. During the year under review, the Operational Risk Management Unit reviewed and approved 58 process documents spanning diverse areas.

The Bank performs User Acceptance Testing (UAT) to identify gaps in the actual deliverable versus the one proposed in the Business Requirement Document (BRD). These gaps are further addressed and closed during the Functional Specification Documentation (FSD) stage before moving to production. During the year under review, 125 BRDs and FSDs were reviewed and UAT was performed for 70 new developments/changes/fixes provided by the Bank’s IT team.

The Bank has initiated Risk and Control Self-Assessment (RCSA) in all business processes to identify inherent and residual risks. These tools also help identify the design and effectiveness of the controls introduced. The outcome of RCSA activities has helped provide insights into known and potential Operational Risk areas in various processes and business lines.

At an entity level, the Bank continues to monitor 40 Key Risk Indicators (KRIs) for strategic impact. While strategic KRIs are intended to be evaluated for operational resilience, there are functional KRIs defined specifically for business units and support functions for tactical management. In FY 2023-24, customised KRIs were introduced in Branch Banking, MicroBanking, Housing Loans and Digital Banking units. These KRIs are analysed on a monthly basis and comprehensive reports are placed to the management and Board at quarterly intervals with an action plan for the closure of open issues.

For the ensuing year, the Bank intends to expand the scope of customised KRIs by expanding the ambit to include secure Agri-banking, vehicle finance, MSME and operations.

Loss Data Management is in place to record material incidents to learn from errors and strengthen existing controls. Incidents are recorded as operational losses and near-miss events. This is followed by a Root Cause Analysis (RCA) of critical incidents. The Bank records instances along the RBI-defined lines of Operational Risk events and process enhancements are tabled at various committees for further action. During the year, the Bank engaged a vendor to put in place a system to capture, report and analyse incidents and losses thereof. This system is expected to be rolled out by Q1 FY 2024-25.

During the year under review, the Bank completed a periodic risk assessment of key outsourced vendors. It helped ensure that these vendors comply with the minimum requirements prescribed by RBI and business continuity aspects as per internal norms. The outsourcing policy was also enhanced to adhere to the RBI guidelines on IT outsourcing. Gaps, if any, are regularly analysed as part of the Risk Assessment (RA). Detailed notes are recorded on each vendor through visits, and also, placed at various forums and committees for further action. During the year, the annual assessment for 31 vendors was reviewed and completed.

The Bank along with concerned stakeholders, prepares and enhances Risk and Control Matrices (RCMs). Design, implementation, and operating effectiveness of all the controls mentioned in the RCMs (i.e., financial and operational controls) are tested annually by the Operational Risk team by selecting samples based on risk categories covering different regions and across the review period. The critical gaps observed during such testing are discussed with concerned functions for upgrading controls, which may include automation of the controls.

The Bank monitors various risk and loss limits such as duration, PV01, stop loss, counterparty exposure, and borrowing and lending limits regularly. Any isolated instance(s) of breach of the limit are immediately brought to the notice of stakeholders, and remedial measures are taken. During the year, the Bank commenced investing in non-SLR securities encompassing Certificate of Deposits, Commercial Papers and Liquid Mutual Funds.

During the year, RBI revised its guidelines on classification, valuation and operation of the Investment portfolio vide circular dated September 12, 2023. As part of these guidelines, policy and system-level changes were warranted in areas such as portfolio classification, revised carrying costs, accounting norms and additional disclosures. The Bank reviewed our preparedness to comply with these new directions w.e.f. April 1, 2024, and necessary changes were implemented in all aspects.

During the year, we updated our behavioural analysis of maturity, pre-maturity and rollover trends in Retail and Bulk deposits, revolving credit facilities and SMA collections. Basis the outcome, the impact is factored into the preparation of liquidity statements and broader liquidity management.

During the year, the Bank continued to perform comprehensive stress tests on the cash flow statements under various levels of stress scenarios. The outcome of the same was assessed against the liquidity buffers available and linked to the Contingency Funding Plan.

The Bank computes and monitors key ratios such as Liquidity Coverage Ratio, Net Stable Funding Ratio, Structural Liquidity Statement, and Cash/Funding Gap Analysis among others. Thresholds are monitored on a daily/fortnightly basis as applicable and breaches are escalated as per internal policy norms for corrective actions.

The Bank computes our Risk Weighted Assets as per RBI guidelines. The impact on the Capital to Risk (Weighted) Assets Ratio (CRAR) is computed and assessed along with providing a forward guidance to the capital adequacy through its Internal Capital Adequacy and Assessment Process (ICAAP). While SFBs are mandated to assess capital adequacy only for its credit risks, the Bank, as part of good governance, also assesses our capital buffers after factoring all Pillar I risks viz, credit risk, operational risk and market risk. CRAR, after factoring Pillar I risks, is also disclosed by the Bank in our Pillar III disclosures. As part of ICAAP, economic capital requirements are computed and reviewed against availability for adequacy assessment.

A key imperative for the ERM unit is to analyse and identify capital requirements for those risks that are not covered/are partially covered under the Pillar I risks. The scope of capital adequacy assessment for Pillar II risks includes liquidity risk, interest rate risk in banking books, concentration risks and underestimation in credit risks, among others. The ERM unit actively engages with the silo-based risk units throughout the year to identify the capital needs for unexpected losses.

As part of strategic risk management, the unit developed and deployed advanced risk management tools during the year, as below:

• Bank Performance Index (BPI): A 360-degree performance assessment tool covering strategy, financial, risk & compliance and internal control parameters at the Bank level. During the year, the Bank replaced our erstwhile Bank Stability Maps, which limited our focus to CAMEL parameters, with the BPI, which evaluates performance on various facets. The outputs of the BPI are directly linked to the Risk Appetite framework for ongoing monitoring.
• Risk Adjusted Return on Capital (RAROC): To identify business units adding shareholder value.
• BCG Growth Share Matrix: To identify business units/regions/branches and their positioning in the BCG growth-share matrix - star, cash cows, question marks and dogs.
• Capital Allocation Framework: Allocation of economic capital to business units and integration to Funds Transfer Pricing (FTP) for profitability assessment/recovery of capital costs.
• Coverage ratio: An alternate measure to identify insolvency upon recognition of the true extent of losses.

The Bank undertakes comprehensive stress testing of all loan portfolios using sensitivity and scenario analysis. The impact of adverse events on the PAR%, NPA%, provisions and capital adequacy is assessed at regular intervals. The Bank’s stress testing framework complies with the requirements for ‘Group C’ bank as specified in Stress Testing guidelines dated December 2, 2013. Capital adequacy requirements are evaluated under various levels of stress for inclusion in the ICAAP document.

In scenario analysis, the Bank used advanced statistical techniques to ascertain interrelationships with macroeconomic parameters. To this effect, a Proof of Concept (POC) was developed and deliberated upon for enhancement during the year. These models, after further fine-tuning, are proposed for use in scenario analysis and balance sheet simulations. Since these models are data intensive and industry practices are yet to emerge, the Bank intends to take a transitional approach for linkage to strategic decisions after multiple rounds of backtesting and validation.

On climate risk stress testing, the Bank continued to assess the financial impact of climate-induced physical risks over a short-term period (1 to 3 years). These models were developed by benchmarking scenarios prescribed by the Network for Greening the Financial System (NGFS). Given that climate risk is an emerging area and the best practices are yet to emerge, incremental enhancements to the framework will be undertaken on an on-going basis on a best effort basis.

The Bank has been proactive in its efforts towards climate risk management. Based on expectations laid out by RBI in its Discussion Paper on climate risks dated July 27, 2022, the Bank had relied on the BCBS prescribed TCFD disclosures framework early on to put in place a governance framework, climate risk strategy, risk management and metrics for monitoring. It is pertinent to note that the Bank was the first SFB to publish its TCFD disclosures report for FY 2022-23 in November 2023. On internal review, it was seen that the Bank’s maiden disclosures were already complying with the aspects laid out in the Draft Disclosure Framework on Climate-related Financial Risks, 2024 dated February 28, 2024. Based on materiality, much of the internal policy advocacy was on risk management with respect to physical risks. To this effect, we mapped our micro banking credit exposure, which is vulnerable to floods and cyclones and appropriate exposure limits were included as part of the credit policy. In the ensuing year, the Bank intends to evaluate the posture of transition risks in retail segments, namely vehicle, housing and MSME segments.

As stated above, with the introduction of statistical scorecards, there was a need to put in place a policy to govern the development, maintenance and validation aspects of risk modelling. To that effect, a Board-approved policy for credit risk model development and validation policy was put in place. While the initial policy restricts the scope to application scorecards, behavioural scorecards, and ECL models, it is envisioned to be increased progressively to other modelling activities.

A greater need for a formal structure to evaluate model risks arose with the introduction of pricing models. A key exercise initiated during the quarter was a comprehensive review of pricing strategies adopted in all Strategic Business Units. While RBI guidelines on loan pricing mandate the delineation of spread components and the assignment of benchmarks (MCLR or EBLR) to loan pricing, the Bank undertook an internal exercise to evaluate the reasonableness and effectiveness of loan pricing to meet strategic imperatives. The Bank developed pricing models using internal data estimates (e.g., ECL outputs) and external benchmarking, wherever applicable. Further, these pricing models were carefully evaluated for alignment to business strategy, budgets, system feasibility, adherence to prudential risk management norms, and compliance with RBI guidelines on interest rate management. A salient feature in these pricing models includes the adoption of a differential pricing matrix, wherein pricing of loans to borrowers would be risk-adjusted to reflect the borrower’s creditworthiness. The Bank believes that the introduction of RBP will aid in encouraging and incentivising borrower(s) to maintain a long-term relationship. These pricing models are expected to be made effective by June 1, 2024 and will be subject to regular backtesting or validation to identify model risks.