In a previous post, we reviewed the June 2018 Financial Stability Report (FSR) and made a list of themes and priorities along which the RBI can consider incremental modifications, such as on reforming stress test design, disclosing de-anonymised stress test results, and providing an analysis and a narrative around its chapter on macro-financial risks. In light of the recently released June FSR report, we examine the need for the FSRs to have a general sense of continuity and narrative building, both within the current edition and across earlier versions.
The objective of the Financial Stability Reports is to give an appraisal of the risks to and the resilience of the financial system at a systemic level.
For any such appraisal to be meaningful and relevant, it is necessary that each FSR provides a commentary on the important risks and trends identified in the earlier FSR and how these have evolved, or not, in the intervening period. Providing this continuity would give context to the stress tests and their results. While the latest FSR does compare the results of the stress tests with the earlier FSR, this is done in an ad-hoc manner and without comment. Some specific instances where establishing continuity would have helped are –
- Discontinuity in the thematic exploration – the Dec 2018 FSR had a thematic exploration of the performance of Prompt Corrective Action (PCA) and non-PCA banks while the current FSR explores the developments in the consumer credit and NBFC space. While both themes were topical, they seem disjointed when viewed together. It is the prerogative of RBI to determine which topics are relevant and explore them accordingly. However, the RBI could ensure that there is a connecting thread among the topics it chooses and that these analyses are complete. For instance, while there has been progress with some banks coming out of the PCA framework, a substantial number of PSBs (Public Sector Banks) are still under PCA. Thus, it seems that a follow up on the PCA analysis would still be relevant.
- Inclusion of Housing Finance Companies (HFC) for contagion analysis (Chart 2.33, June 2019 FSR) – analysing the contagion impact of the default of HFCs is important given the extensive and substantial linkages they have with the banking sector. However, these linkages were not formed overnight, and RBI was aware of these linkages even before the liquidity stress hit the NBFC and HFC sectors. Indeed, the contagion analysis of FSRs recognises HFCs as an important contributor in the past (see Chart 2.32 of Dec 2018 FSR, Chart 2.29 of June 2018). The RBI can therefore instead be proactive in identifying the types of institutions which are contributing to the development of risks in the financial system and include them in the contagion analysis and other stress tests in a continuous basis rather than as a response to recent market events, such as that concerning the housing finance sector that prompted this analysis (See pages 42-45, June 2019 FSR).
- Change in methodology for stress tests – In the bank level sensitivity analysis of credit risk, the shocks given to the GNPA ratios, Shock 1 and Shock 2, are 1 SD and 2 SD respectively (Chart 2.10, FSR June 2019). This is different from the shocks used in editions before the December 2018 FSR. For example, in the June 2017 FSR (Chart 2.10), five shocks of i)1 SD on NPAs, ii) 2 SD on NPAs, iii) 3 SD on NPAs, iv) 30% of restructured advances turn to NPAs (sub-standard category, and v) 30% of restructured advances turn to NPAs (Loss category – written off) are applied. The rationale for such changes in methodology, when they occur, must be provided as it affects the interpretation of results and aids better comparative monitoring by the consumers of the FSR reports.
- Some explanation is required in the usage of methodological models like the econometric bank group level models used to estimate the slippage ratios (see Bank Group Level Models, Annex 2, June 2019 FSR). The models differ at the bank group level, i.e., across PSBs, PVBs (Private Sector Banks) and FBs (Foreign Banks). The rationale behind the specific inclusion or exclusion of variables in the model must be provided. For example, the current account balance to GDP ratio and the gross fiscal deficit to GDP ratio are included in the PSB model but not for PVBs and FBs. The exports to GDP variable has been lagged by 1 year and 5 years respectively for PVBs and FBs and has been entirely excluded for PSBs. Further, these models have been changing over the years without adequate explanation for either the choice of the variables or their change.
The other theme that is missing, as we mentioned earlier, is narrative building. The FSR contains a plethora of information in the form of tables and graphs. However, the narrative following the graphs and tables describes the information presented in the figures without providing a sense of how the RBI perceives these insights. While it might not be possible to determine the cause of all trends observed precisely, a commentary on the same by the RBI would give an insight into the thinking of the central bank and how it views various developments. Some examples to illustrate the point are –
- The macro stress test results of credit risk (Chart 2.7) for PSBs show that the GNPA under the baseline, medium and severe shock are better than the Actual. However, for PVBs and FBs, the GNPA levels for the medium and severe stress scenarios are higher than Actual. It looks anomalous that PSBs, who have been doing poorly, have their GNPA ratios reduced under stress scenarios while for other banks, the GNPAs increase. This is a point that needed further analysis and comment by the RBI.
- In the section on performance of SCBs, Chart 2.2 (e) shows the changes in the Provision Coverage Ratio (PCR) of the various bank groups. While earlier charts make it clear that PSBs have NPA ratios much higher than PVBs or FBs, we find that the PCR of FBs is higher than both PSBs and PVBs. The RBI can provide more in-depth analysis to understand what drives this difference, for instance, by estimating what percentage of these provisions were due to regulatory requirements and how much was due to proactive provisioning.
- The macro stress test results on credit risk show that there is an improvement in system-level conditions compared to the results from the December 2018 FSR, with the GNPA ratios of the SCBs expected to come down and the system level CRAR values expected to go up (Charts 2.7, 2.8). However, the bank level sensitivity analysis of credit risk shows that under a severe shock, the system level effects are greater this time as compared to six months back (December 2018 FSR). Additionally, the reverse stress test results indicate that it required a shock of only 2.9 SD to bring down the system-level CRAR to 9% while in the previous FSR, this value was 4.15 SD (Section 2.22, June 2019 FSR). This appears counterintuitive and RBI could have made a comment on the same.
- The Chapter on ‘Financial Sector: Regulation and Developments’ has in the past two FSRs prominently dealt with the topic of risk, in two Boxes, namely ‘Box 3.4: Risk Culture’ (Dec 2018 FSR), and ‘Box 3.1: Risk Society – The paradigm of risk?’ (June 2019 FSR). The former deals with how to assess the culture of risk-taking in financial institutions while the latter discusses risk as a construct within a globalised society. These, however, do not connect back to if and whether RBI is considering moving towards a more sophisticated capture of operational risks and misconduct risks which are an integral part of the risk culture of firms, and which have ramifications for the broader economy. The available analysis in the FSR is restricted to ex post dissection of banking frauds across bank groups and the quantum of such frauds. A much more comprehensive analysis of operating risks including supervisory observations on risk culture and a complaints analysis, from the Ombudsman, across bank-group types and penalties levied is direly needed.
 Standard deviation