WIXLIB

January 2019Early warning signals in adigital era: A proposed PwCframework to manage creditrisk better

Moving towards global standards ofcredit monitoringAccording to RBI data, on 31 August 2018, 48.3% ofthe loan book of some of the large Indian scheduledcommercial banks (SCBs), taken together, consisted ofcorporate assets. The remaining 51.7% is spread betweenmicro, small and medium enterprises (MSMEs), retailcustomers and others (Figure 1). With this distributionof asset classes, the RBI estimates that the grossnon-performing assets (GNPA) of SCBs can rise from11.6% in March 2018 to 12.2% by March 2019. Themajor reasons for this is the inability of the banks toanticipate the incipient stress in accounts that are likely todefault. Simultaneously, the RBI’s regulations have beenevolving from rule-based monitoring to an early warningidentification approach over the last two decades (fromIncome Recognition and Asset Classification (IRAC) normsin the Nineties to monitoring specific signals in 2017). Thishas led investors to question the way loan accounts aremonitored by Indian banks in recent years.According to market participants, there is a dire needto revisit current practices and rework them accordingto globally accepted and advanced approaches used indeveloped economies. Banking organisations in India needto move from an essentially compliance-driven post factomechanism to a proactive control-based digital system tomonitor credit risk.To achieve this, lenders will have to constantly monitorpublicly available news on borrower- and industry-specificperformances. Analysing this information on advancedanalytical platforms will help lenders to take timelydecisions. Borrowers can then be grouped into three orfive watch list categories. Assuming a three-colour scale(green, amber and red), borrowers in the amber regionshould be monitored closely (here we are assumingcustomers lying in the red category have already turnedbad and giving them an early warning digitally may not2 PwCFigure 1 - Breakup of loan book of SCBs inIndia (Q1 FY19)Breakup of Loan BookRetail/MSME/OtherExposure 51.7%Corporate/Industry/Wholesale Exposure 48.3%create any significant value). Additionally, it may be notedthat as credit quality-related information takes some timeto impact the borrower, the system can be designed torun via batch processes, maybe daily or weekly, insteadof making it on an absolute real time basis. The initialinvestments to make it work on an absolute real time basiscan be very high.

Need to be proactive, rather than reactivein monitoring and managing credit riskPast experience has taught us that proactive controlscan help banks limit their exposure in bad segments andlead to reduction in overall losses (Figure 2). We havedivided Indian banks into two categories—Category A andCategory B, with Category A banks following traditionalpractices and Category B banks adopting an advancedmonitoring approach. The chart indicates that CategoryB banks have been successful in identifying bad assetsbetter than their counterparts, and hence, their gross NPAlevel and provisioning numbers are comparatively low.Therefore, we propose a digital credit risk monitoringframework to capture such early warning signals. Let usfirst try to understand the differences between the two.Figure 2 - Based on past experiences banks have been categorised (Category A – traditional andcategory B – advanced)16Category A Banks - Gross NPACategory A Banks - Loan loss provisionCategory B Banks - Gross NPACategory B Banks - Loan loss provision141210Δ GNPA levelof category Ais 1.85 timesof category -122010-112009-102008-092007-0802006-072Loan lossprovisioning forcategory A ismore than doublethan that ofcategory BComparison between the proposed framework and traditional practicesTraditional monitoring systemDigital monitoring system to capture early warning signalsApproachTraditional monitoring systems work on a postfacto basis, primarily based on past eventsand selective financial projections, i.e. they arereactive in nature.Digital monitoring early warning signals help financialinstitutions capture information in advance and takeprecautionary action before an account turns bad, i.e. it isproactive in nature.IntegrationFront-end business, underwriters and themonitoring authority would work in silos inthe past, which resulted in discrepancies ininformation.Seamless coordination of monitoring, underwriting and businessteams is crucial for the success of a system.EmphasisThe primary emphasis was on high-risk areasand the focus was limited to decisions basedon sample performances.The entire population-related data is analysed and thenaccounts are categorised according to a three-colour scale(green, amber and red). Customers in the amber region shouldbe under highest vigilance, assuming that those in the redarea have already become NPAs. This makes the system morecomprehensive in nature.PeriodicityAccounts were usually reviewed on a semiannual basis, or at the most, quarterly for highrisk accounts.Based on the capability of a system, assessments can beperformed in batch processes (on a daily or weekly basis).CoverageThe traditional system had a much narrowviewing angle. It failed to take a consolidatedview of individual loan accounts.Early warning signals (EWS) collates data from multiple sources,which helps the bank to have a consolidated view of individuals’loan accounts and decision-making becomes easier.UsageTraditionally, provisioning was based onregulatory guidelines (RBI’s IRAC Norms) andaccounting requirements (INDAS109).Systems can be upgraded if such signals are incorporated intothe INDAS109 staging decision process.OutputManual cleaning and analysis of data was theprimary option available in most cases.MIS dashboards can be prepared automatically and userinterfaces can be customised, based on simple drill-downs.How to manage credit risk better with early warning signals: A proposed PwC framework 3

Potential benefits of adopting a digital frameworkfor monitoring and managing credit risk Reduction of default rate at portfolio level: The rateof defaults at the portfolio level can be reduced byrestrictive exposure to particular customer segmentswhere warning signs have been observed. This will helpto control the probability of default (PD). Tightening of covenants: Lenders need to interveneearly by tightening the covenants and increasingcollateral levels to maximise recovery. Once a loanaccount comes into the watch list, regular supervisionwill help to maximise collateral-related requirements andreduce losses in the event of actual default. This willminimise loss given default (LGD). Securitisation of assets: Lenders must try to proactivelysecuritise or sell assets that have warning signalsassociated with them and where chances of revival arelow. There are certain non-banking financial companies(NBFCs) and asset reconstruction companies (ARCs) inIndia that may be interested in taking over such assets.This will help companies get bad loans off their booksand get a better price for them. Decision on IFRS9 staging-related aspects: The IFRS9framework can be integrated with the INDAS109 ECLmodelling framework. It will enable banks to quicklygauge the actual health of a borrower and accordinglydecide on the ‘stage’ of the account for provisioningpurposes. Identification of trends and opportunities in themarket: New trends can be identified in differentindustries and banks can readjust their portfolio mix.For example, it could have helped banks identify thetelecom sector as a whole turning adverse and slowlyreduce their exposure to it. Decisions taken on composition of portfolios:Monitoring at the portfolio level can help to reducebanks’ exposure to threatened sectors, and evenat individual levels reduce exposure by decreasingcommitted lines of credit, i.e., the credit conversionfactor. Modification of the credit policies of banks: High levelcredit policy-related decisions can be taken by banks,based on information provided by the system aboutsectoral regulations, performance and news.A segmental approach to reducefalse positives Large and medium companies: As these are high-valueaccounts, the ideal approach should be to assess therisk of bad loans at an account level. Based on theassessment, industry variables can be identified. Inanother approach, market data, movement in the spreadof commercial papers (CPs) can be used as a goodindicator for this segment, since many of these entitiesissue CPs, bonds and non-convertible debentures(NCDs), which are subscribed by banks. SME segment: Banks can create industry andgeographic clusters or a mix of both in this segment.They can also use already existing geographic clusters. Retail assets: Banks can create pools of homogeneousaccounts for designing purposes. Moreover, socioeconomic and geographic classification can be helpful. Agriculture portfolio: The biggest problems in thissegment include extremely small average agriculturalland holdings and variable repayment schedules. It isalso necessary to take into account factors that have asignificant impact on crop production and repaymentconditions, e.g., rainfall and political scenario.4 PwC