Chapter 1 – Overview |
---|
Stakeholders requested clarification regarding which OSFI guideline version should be applied until the new credit valuation adjustment (CVA) risk framework is in effect. Stakeholders also inquired if banks with internal models method (IMM) approval can use Advanced CVA risk capital charge for their capital floor requirement between 2023 and 2024? | The revised CAR chapter relating to CVA risk will come into effect in fiscal Q1-2024. Therefore between 2023 and 2024, Chapter 8 of the 2023 CAR Guideline should be used for guidance on CVAs and banks with IMM approval can continue to use the Advanced CVA risk capital charge for purposes of the capital floor. |
Stakeholders noted that the Basel III minimum capital conservation provisions apply to both CET1 and leverage requirements and inquired if OSFI would consider adding a cross reference to leverage ratio minimums in the CAR Guideline. | A new column was added to Table 7 in Chapter 1 with the corresponding leverage ratio requirements. |
Chapter 2 – Definition of Capital |
---|
Stakeholders requested clarification on OSFI’s revised calculation to split general allowances between the Standardized Approach and the Internal Ratings Based (IRB) Approach and whether this should be done in a manner consistent with the institution’s internal and external allowance reporting. They also inquired if they should follow the approach applied when updating their Pillar 3 CR1 template. | Institutions are required to follow their internal allocation of allowances in accordance with IFRS 9 rather than the current RWA-based proxy. CAR Chapter 2 specifies that institutions that have partially implemented an IRB approach should split “general allowances between the Standardized Approach and the IRB Approach in a manner consistent with the institution’s internal and external allowance reporting”. Table CR1 from Pillar 3 is an example of such external reporting based on accounting values. |
Stakeholders questioned the proposed new deduction from CET1 capital for institutions’ prepaid portfolio insurance assets. They argued that prepaid portfolio insurance has value in resolution or distress and applying a deduction to these assets would create a disincentive for institutions to purchase prepaid portfolio insurance as a risk mitigation tool. | OSFI has amended the treatment for prepaid portfolio insurance assets to include a 100% risk weight and prescribed amortization expectations as outlined in CAR Chapters 2 and 4. |
Chapter 3 – Operational Risk |
---|
Since banks that use the Simplified Standardized Approach (SSA) have a Business Indicator (BI) less than $1.5 billion, stakeholders recommended that OSFI use a coefficient of 12% for the SSA, which would align with the marginal coefficient for BI under $1.5 billion used in the Standardized Approach (SA). | The Business Indicator (used in the SA) is generally greater than or equal to Adjusted Gross Income (used in the SSA), and the amount of the difference between these measures varies greatly between institutions. It is, therefore, appropriate to have a higher coefficient under the SSA than the SA. OSFI believes that 15% is an appropriate level as it is consistent with both the coefficient used in the current Basic Indicator Approach used by most SMSBs and the same as the marginal coefficient under the SA for BI over $1.5 billion. |
Stakeholders noted the challenges in providing loss data information for subsidiaries, since it is not available on a legal entity basis. | Subsidiaries are only required to provide loss data if they are in Category 1 and have annual Adjusted Gross Income greater than $1.5 billion. |
Stakeholders recommended that OSFI align with the BCBS Basel III Framework and allow the initial use of 5 years of loss data for the SA on a transitional basis. | The Basel framework does allow national discretion for a shorter observation period of 5-10 years in exceptional cases if 10 years of good quality loss data is not available when the institution first moves to the SA. OSFI has decided to retain a requirement of 10 years of loss data as this is a more representative sample size of operational loss history and creates more consistency by ensuring that all SA institutions are using data from the same duration. Category 1 SMSBs may apply to use the SA (with a minimum Internal Loss Multiplier of 1) with 5-9 years of loss data. |
Under the proposed requirements, if more than 5% of an institution (measured using Business Indicator) does not have 10 years of high-quality loss data, the institution would need to use an Internal Loss Multiplier (ILM) equal to or greater than 1. Stakeholders requested that OSFI consider a higher threshold than the proposed 5%. It was noted that under the AMA requirement, OSFI had defined a “significant” part as 75%, while a "material" part was defined as 90%, of an institution’s operations. Stakeholders also questioned the proposed requirement to include loss data for acquired businesses as it may pose implementation challenges to mergers & acquisitions. For example, it would be challenging to validate the quality of loss data prior to an acquisition or if data is beyond the record retention period. | The final CAR Guideline Chapter 3 has increased the threshold to 10%, which is more consistent with the materiality threshold that was used for AMA. The threshold is also now measured as the percentage of total loss data that is estimated, which means that even if an acquired entity does not have 10 years of loss data, the data that is available is taken into account. As well, institutions will have the ability to be above the 10% threshold on a temporary basis without an automatic ILM adjustment. Institutions must come below the threshold in a timely manner. |
OSFI’s guidance for a timing loss event, is to only include the gross loss amount without including the previous revenue overstatement as an offsetting entry. Stakeholders felt that such treatment of timing losses could imply a disproportionate amount of capital is attributed to a type of loss event that carries no financial impact over time. The draft CAR Guideline Chapter 3 provides examples of timing losses, including revenue overstatement, accounting errors and mark-to-market errors. Stakeholders asserted that these events do not result in a true financial impact on an institution (i.e. zero impact over time), and that not all timing loss events will give rise to legal risk. In situations where the legal implications arising from a single or multiple timing loss events are deemed relevant, the additional risk and the related loss is captured through appropriate legal provisions. Stakeholders felt that the proposed requirement to track timing losses using only the “offsetting entry” in the loss database for capital calculation, in addition to the loss data already captured under legal provisions, will result in an overstatement of an institution’s risk profile and capital volatility, particularly if the accounting adjustments are sizable. | Institutions should include the notional amount of timing losses in their loss data sets, and the previous revenue over-statement or expense under-statement cannot be used as a recovery against a timing loss. For clarification, the wording “and give rise to legal risk” was removed from the definition of timing losses in the CAR. After considering feedback and recognizing that accounting errors are a unique type of timing loss, OSFI decided that the materiality threshold for inclusion of timing losses that are accounting errors may be set at a level higher than $30,000. However, banks must set the threshold for timing losses that are accounting errors at a level below that used by external auditors in their annual financial audit when determining the summary of material misstatements. Accounting errors do not include errors in the mark-to-market valuation of financial assets or timing errors that involve third parties (e.g. customer over-billing or underpayment to third parties), which must be included in the loss data set when the amount of the timing loss exceeds $30,000. |
Stakeholders felt that the proposed CAR Guideline requirements related to timing losses deviate from the guidance previously issued under AMA and may create a gap in the historical loss database that could be impracticable to close prior to implementation. | OSFI recognizes that the guidance around treatment of timing losses under AMA may not have been as clear as the definition and examples provided in the 2023 CAR Guideline. Therefore, institutions will not be penalized through an ILM adjustment if their historical loss data sets before 2023 do not include all timing losses meeting the definition in the 2023 CAR Guideline. |
There is no Income Statement entry to capture losses from events where revenue was not collected due to an operational error and the date of recognition for these losses cannot be determined. Stakeholders recommended that these events be excluded from the capital calculation. Instead, the banks can record them for risk management purpose as opportunity costs. | Revisions to CAR Guideline Chapter 3 further clarify the treatment of these types of losses by specifying that loss data must include costs from uncollected revenue that can be quantified based on the contractual obligations of the institution’s client or customer. The CAR Guideline further clarifies that for losses from uncollected revenue, institutions may use either the date in which the revenue should have been collected, or the date in which the decision was made not to collect the revenue. |
Stakeholders commented that the proposed methodology to estimate loss data for acquired businesses in section 3.4.7 may be challenging to implement in practice. Specific challenges related to the methodology include: - Calculating the BI for acquired businesses
- Calculating an institution’s BI excluding an acquired business
- Calculating BI for an asset purchase
To simplify the inclusion of Merger & Acquisition activities, stakeholders requested that OSFI: - Allow exclusion of pre-acquisition losses and BI for asset purchases (loan/credit portfolios).
- Allow exclusion of pre-acquisition losses and BI for minor acquisitions.
- For material acquisitions, allow use of proxy data (internal or external) to estimate losses for the missing years.
- Define a materiality threshold based on generally available information such as total assets acquired, or revenues for the previous year at the time of acquisition.
| To address concerns that the proposed approach would be challenging to implement, the CAR Guideline will allow institutions to use, as an alternative methodology, 125% of Adjusted Gross Income (detailed in section 3.3 of CAR Guideline Chapter 3) for the year prior to the merger or acquisition as a proxy for BI of the merged or acquired entity. The CAR Guideline has also been amended to clarify that the ILM for the quarter prior to the merger or acquisition is to be used when determining how to estimate losses in section 3.4.7. As a result, Institutions will not need to calculate the ILM excluding the acquired business or entity. For mergers and acquisitions in the ten years prior to the implementation of the revised CAR Guideline (i.e., from 2013-2022), institutions may estimate any year of missing loss data at 1% of the acquired business’ BI (since institutions will only have an ILM once the new SA is implemented). Note that institutions must include actual or estimated loss data for
all merged or acquired businesses within the past ten years, including for the period before the acquisition. The only exception is when an institution purchases assets (as opposed to the acquisition of, or merger with, a legal entity), in which case no adjustment to BI or loss data is required for the period before the purchase. |
Chapter 4 – Standardized Approach to Credit Risk |
---|
Differences in the definition of asset classes between Chapters 4 and 5 can lead to operational inefficiencies for an IRB institution subject to the capital floor based on the Standardized Approach. These differences will require institutions to report two sets of asset classes – one for the Standardized Approach and another for the IRB Approach. Stakeholders requested that OSFI eliminate the differences in definitions, as appropriate, to improve overall efficiency. Examples of such differences include the definitions of retail exposures, exposures secured by residential real estate, regulatory retail exposures, and qualifying revolving retail exposures. | OSFI has incorporated revisions to better align the definitions across the Standardized Approach and IRB Approach, including clarifications to the definitions of transactors, qualifying revolving retail exposures and exposures secured by residential real estate. |
Stakeholders requested further clarification on the expectation for the new due diligence procedures. Smaller institutions have limited resources and expertise to effectively perform adequate due diligence procedures and stakeholders requested these institutions continue to be able to rely on external ratings without the requirement for additional validation. Stakeholders also requested that OSFI clarify whether due diligence is also required for IRB banks that need to calculate the output capital floor. | Chapter 4 provides details on the types of information that could be used in meeting the due diligence requirements and includes details on the required timing of due diligence analyses. Simplified treatments (which involve a flat risk weight regardless of external credit rating, and do not require that the due diligence requirements be met) have been developed for those asset classes requiring due diligence analyses to validate external credit ratings. These simplified treatments were introduced to reduce operational burden for institutions with immaterial exposures to these asset classes. In addition, due diligence is required for IRB banks to calculate the capital floor. |
For A-IRB banks that must use the Standardised Approach for the purposes of the capital floor, stakeholders asked whether OSFI would allow them to continue with the existing CAR Guideline Standardised Approach requirement of using one notch below the sovereign rating for their bank exposures. | Institutions will no longer be permitted to apply ratings one notch below the sovereign rating to their bank exposures, including for purposes of the output floor. Unrated bank exposures can be treated using the Standardized Credit Risk Assessment Approach. Alternatively, institutions may choose to apply a 100% risk weight to all of their unrated bank exposures. |
Stakeholders indicated that OSFI has proposed “base” risk weights for rated and unrated covered bond exposures that are more punitive than the Basel Committee on Banking Supervision (BCBS) risk weights under Basel III. In particular, the proposed risk weights for rated covered bond exposures are identical to unsecured bank exposures. Stakeholders expressed concerns that this approach puts Canadian covered bond issuers at a disadvantage as compared to international peers. The proposed risk weights would likely reduce future participation from certain investors, which would in turn limit Canadian banks' access to the global debt capital markets. The more punitive risk weights proposed by OSFI do not fully consider the structuring, underlying security, and operational features of the Canadian registered covered bond programs. As such, stakeholders requested that the BCBS “base” risk weights be used. | OSFI has considered the structure and features of the Canadian registered covered bond program and will continue to use the “base” risk weights for institutions’ holdings of covered bonds given the collateral pledged is not eligible under the Standardized Approach. In addition, OSFI allows issue specific ratings (instead of issuer ratings) to be used for covered bond holdings and, as such, covered bonds will often have a lower risk weight compared to unsecured debt issued by the same bank. |
Stakeholders requested that OSFI maintain the current materiality exemption allowing 100% risk weighting for immaterial equity holdings. In addition, they requested that OSFI adopt a phased-in approach for the treatment of equity exposures consistent with the BCBS treatment and similar to the approach recently adopted for the capital floor. | OSFI has considered the merits of the materiality exemption and the phase-in approach for the treatment of equity exposures. However, equity exposures are inherently risky (first to absorb losses) and, as such, the risk weights prescribed under the Basel III framework are appropriate. In addition, a phase-in is not required given the simplicity of the rule. |
Under the proposed requirements, paragraph 93 limits a residential real estate exposure to a first mortgage. This is not viewed as consistent with the ‘Claims over the property’ section in paragraph 88, applicable to all real estate exposures, and Chapter 5, Paragraph 25, where residential mortgage loans include junior liens. Stakeholders recommended that OSFI remove the "first mortgage" requirement in the residential real estate definition to align with the BCBS framework and Chapter 5, Paragraph 25 (2), which includes subsequent liens. | Consistent with Chapter 5, paragraph 25, the reference to a "first mortgage" requirement has been removed from the definition of residential real estate in Chapter 4, paragraph 93. |
Regarding income-producing real estate, stakeholders noted that the 50% income/cash flow test is an onerous requirement where institutions may not have enough data to perform the test. They expressed appreciation for OSFI’s previous confirmation that continuous tracking of the source of income is not required throughout the life of the mortgage. For both SA and IRB mortgage loans, stakeholders also understood that determination of the source of income is only required at the time of credit adjudication (e.g., loan origination and/or loan re-finance) and that OSFI will permit institutions to use a flag in their systems to identify the property purpose (i.e., owner occupied or rental property) provided the flag results in as conservative an identification of income producing properties as the 50% income/cash flow test. | OSFI has provided flexibility on this requirement to allow institutions to use their own internal definitions of “income producing” provided such definitions are more conservative than the definition in the CAR Guideline. |
Stakeholders asked OSFI to clarify the requirements to use the risk weights in Table 10 titled “Risk weights for general residential real estate exposures”. | A residential real estate exposure is an exposure secured by a residential property (e.g., individual condominium residences and one-to four-unit residences) made to a person(s) or guaranteed by a person(s). Investments in hotel properties and time-shares are excluded. Table 10 applies as long as the exposure is secured by a residential property and made to a person(s) or guaranteed by a person(s); less than 50% of the income from the borrower used in the institution's assessment of its ability to service the loan is from cash flows generated by the residential property; and it is not a land acquisition, development and construction (ADC) exposure. If the exposure relates to an immovable property but is not made to a person(s) or guaranteed by a person(s), then the commercial real estate risk weights would apply. |
Paragraph 109 defines an ADC exposure as “one for which the source of repayment is either the future uncertain sale of the property or cash flows which are substantially uncertain”. Stakeholders noted that repayment uncertainties can be mitigated by both pre-sales and pre-lease contracts sufficient to repay bank loans. With respect to paragraphs 110 and 111, stakeholders noted that Basel CRE 20.91(2) states that, in addition to pre-sale, “pre-lease contracts amount to a significant portion of total contracts” and can be a criterion for risk-weighting ADC exposures to residential real estate at 100% instead of 150%. Under the draft CAR Guideline issued for public consultation, the pre-leasing option was not available. Stakeholders recommended that OSFI consider the inclusion of the pre-leasing option as it can satisfy the uncertain cash flows described in the above ADC definition. OSFI has added "construction projects" as part of the requirement to qualify for a 100% risk weight which differs from the BCBS framework. Stakeholders inquired about the intent to exclude certain ADC exposures/residential real estate such as development projects and land acquisition, which can often serve as feeder transactions for construction projects. Stakeholders noted that high-rise, purpose-built rental projects would be entirely excluded from the preferential 100% risk weight if qualification is only based on meeting the 50% pre-sale criterion. Stakeholders also noted that 35% equity at risk (based on appraised as-completed value) contributions are rarely seen in construction projects and suggested that OSFI reduce this value. Stakeholders recommended allowing the use of "total project costs" as an alternative to "appraised as-completed value” in measuring equity at risk. They submitted that the “appraised as-completed value” includes a profit component, not just equity at risk. They further noted that in institutions’ risk assessments, they typically compare the amount of borrower contributed equity to the total estimated project cost. | Pre-lease contracts are not included because pre-leasing contracts are not prevalent in Canada. Application of the preferential 100% risk weight is being made available for land acquisition up to a maximum LTV of 60%. The higher threshold for land acquisition reflects the higher risk of those exposures. High-rise purpose-built rental construction projects will be eligible for the 100% risk weight if the equity threshold is met. All other residential construction projects will only be eligible for the preferential risk weight if the 50% pre-sale requirement is met. To be more reflective of market practices in Canada, OSFI has lowered the equity at risk threshold to qualify for the preferential 100% risk weight to 25% (based on appraised as-completed value) from 35%. “Appraised as-completed value” is more consistent with the approach used in the CAR Guideline more broadly. The general relationship of “appraised as-completed values’ to ”total project costs” were taken into account when calibrating equity at risk and LTV thresholds for the preferential 100% risk weight. |
Under section 4.1.16, the proposed requirement to apply a 1.5 risk-weight multiplier for certain exposures with currency mismatch would involve considerable resources to change front-end systems. The challenges are further amplified by setting a threshold of where more than 10% of the borrower’s income, without a natural or financial hedge, used to qualify for the loan is denominated in a foreign currency. Institutions’ systems do not capture the currency of the borrower’s source of income. To do so would require changes to adjudication systems and this data would only be captured for new and, potentially, renewed loans. As these exposures are immaterial for many institutions, stakeholders recommended that OSFI provide an exemption to this requirement for domestic branch local currency retail loans and assume no currency mismatch exists. | OSFI has assessed this risk to be relatively low and as such, to reduce operational burden, the 1.5 risk-weight multiplier for currency mismatch will only apply to residential real estate exposures. The 1.5 multiplier will need to be applied to all residential real estate exposures with currency mismatch starting upon implementation of the new rules in institutions’ fiscal Q2 2023. |
Stakeholders requested clarification on why undrawn balances of credit and charge card exposures are subject to a 25% credit conversion factor (CCF). They assert that the 10% CCF for unconditionally cancellable commitments used in the Basel framework is more appropriate. | The 25% CCF will be retained and was calibrated based on data collected from Canadian banks on Canadian credit card and charge card portfolios. |
Paragraph 162 reads as follows: “Deferred placement fees receivable, non-credit-enhancing interest-only strips, and any other assets that represent the present value of future spread income subject to prepayment risk.” Our comprehension of this text is that non-credit-enhancing interest-only strips must be weighted at 250% when exposed to a prepayment risk. This strong weighting reflects the fact that these I/O strips are more or less as volatile as shares. Since I/O strips are created from pools 966 and 990, which prohibit prepayment and have a very limited history of defaults, and consequently a relatively low probability of prepayment, are we correct to conclude that these I/O strips are free from prepayment risk? | New text has been added to this paragraph to emphasize that I/O strip transactions not subject to prepayment will be risk-weighted at 100% to conform to the risk weighting assigned to similar assets. This approach accounts for the fact that, in the case of I/O strips like 966 commercial mortgage pools, a borrower who decides to prepay would have to pay all future interest charges at the same time. CMHC would only cover up to 12 months of accrued interest and would not cover any default on payments resulting from physical damage to the buildings. If a mortgage remains outstanding for more than 12 months, or is severely damaged by an uninsured risk, the I/O strip asset would probably lose all its worth. |
Regarding paragraph 181, stakeholders sought clarification on why OSFI did not align with the BCBS CRE 21.15 standard and remove the "unrated" application specification from the domestic currency and foreign currency ratings requirement. | The BCBS and CAR Guideline treatments are equivalent. The fact that the exposure is unrated is assumed in the BCBS text as the exposure is being risk-weighted “based on the rating of an equivalent exposure to that borrower” (i.e., not that of the exposure itself). |
Regarding paragraphs 234, 237 and 238, stakeholders requested that OSFI clarify what is meant by the “currency of the exposure”. In the simple case of a loan denominated in currency 1, and collateral denominated in currency 2, it is easy to determine whether there is a “mismatch”. However, in the case of a derivative or in some cases a securities financing transaction (SFT), there is no obvious answer as to what is the currency of the exposure. Stakeholders also asked OSFI to confirm that there is no conflict between these paragraphs and footnote #19 in the current CAR Guideline, Chapter 4. This footnote states that no foreign exchange haircut is to be applied to cash collateral as long as the currency of that cash is eligible to be posted as per the terms of the credit support annex (CSA) (which, by definition, must be the case). The table in paragraph 237 refers only to “cash in the same currency” as having a 0% haircut. The stakeholders assumed footnote 19 takes precedence where applicable and, assuming that is the case, asked OSFI to confirm if there is intended to be a divergence in treatment of cash collateral for derivative agreements compared to SFT agreements. | For derivative and SFT netting sets, the currency of the exposure for variation margin is any currency which is permitted as eligible variation margin in the credit support annex or equivalent documentation. For initial margin or independent amounts, the currency of the exposure is the termination currency of the netting set. Footnote 19 clarifies what is meant by cash in the same currency and therefore is applicable. The text in Chapter 4 has been updated. There is no intent for there to be divergence for cash collateral between derivative and SFT agreements. We have updated the CAR Guideline to reflect this. |
According to paragraph 268(2), the portion of the exposure that is below a materiality threshold must be deducted from CET1 capital by the institution purchasing the credit protection. Stakeholders asked that OSFI confirm that “the deductible portion” of a privately insured mortgage (PMI) exposure with the prescribed RWA treatment does not fall under the CET1 capital deduction requirement under this subparagraph. | OSFI confirms that this paragraph does not apply to PMI-insured mortgages where the Government of Canada backstop is reflected. These are subject to the rules in Chapter 4, paragraph 271 or Chapter 5, paragraph 146. |
Stakeholders inquired whether the choice between the four methods of calculating RWA for privately insured mortgages is decided on a case by case basis, or the election is applicable to all exposures and must be treated that way on a consistent basis. Stakeholders inquired about paragraph 271 and the applicability of the 2.2 factor to approaches i) and ii) or just to approach ii). | Institutions are permitted to choose between any of the four approaches on a case by case basis. The approach may change for a particular mortgage over the life of the mortgage. The 2.2 factor applies regardless of whether approach 1) i) or ii) is chosen. The 2.2 multiplier is used to account for the deductible (first loss position) nature of the losses incurred by lenders prior to the Government of Canada backstop being available. |
Chapter 5 – Internal Ratings Based Approach to Credit Risk |
---|
Stakeholders asked OSFI to reconfirm its expectations are not to change the granularity or calibration of existing models to account for (i) the new categorization of risk weight function (e.g., residential mortgage where repayment is materially dependent on cash flows generated by the property) as well as (ii) the requirement to use the foundational approach for specific populations (e.g., exposures to the bank asset class, financial institutions, corporates with annual revenue greater than CAD $750 million). | The expectations around current granularity or calibration of existing models are not changing. Existing models will continue to be subject to back-testing requirements to ensure models continue to be appropriate for an institution’s current portfolio. |
Regarding the definition of income-producing real estate (IPRE), stakeholders expressed appreciation for OSFI permitting institutions to use a flag in their systems to identify the property purpose (i.e., owner occupied or rental property) provided the flag results in as conservative an identification of income producing properties as the 50% income/cash flow test. They requested that OSFI also align the definition of IPRE with income-producing commercial real estate (IPCRE) in Chapter 4. In paragraph 20, stakeholders noted that OSFI has not designated any Canadian property types as being high-volatility commercial real estate (HVCRE) but still provides specific examples of the application of HVCRE risk weights in relation to certain Canadian loans financing ADC of properties and Canadian institutions’ foreign operations loans on properties. They requested further clarification on these examples and any implications with respect to the capital floor calculation. When OSFI makes references to ADC exposures, stakeholders asked OSFI to align with the definitions in Chapter 4. | OSFI does not believe a similar change is required to Chapter 4 for IPCRE since CRE projects often rely on income generated by the property for repayment. The HVCRE risk weights will apply to Canadian loans financing ADC of properties where the source of repayment is uncertain without substantial equity at risk as well as Canadian institutions’ foreign operations’ loans on properties where the national supervisor has designated these types of property as HVCRE. OSFI has clarified the language in this section. Regarding references to ADC exposures, the definitions have been aligned between Chapter 4 and Chapter 5. |
Stakeholders expressed appreciation regarding the creation of an ‘Other Retail’ (non-regulatory retail) category for exposures failing to meet the six criteria related to the nature of the borrowers and the size of the pool of exposures in paragraph 25. As noted, these exposures would be subject to the Corporate SME risk weight function. They sought the following clarifications: - Considering there is already an “Other Retail” asset class, can OSFI define this population as “non-regulatory retail” instead to improve transparency? This would also apply for Chapter 4.
- In order to ensure consistency in the categorization of exposures in this category (i.e., regulatory vs non-regulatory retail) between Chapters 4 and 5, can OSFI use a consistent parameter for aggregation in both chapters (e.g., authorization amount, standardized EAD)?
- What parameter floors, particularly for LGD, would such exposures be subject to per section 5.4.1 or section 5.4.2? The LGD floor for an unsecured corporate and PSE exposure is 25% per paragraph 97 while it is 30% for All other retail (the sub-asset class under Regulatory Retail) per paragraph 143.
- For exposures to individuals, there is not a concept of “annual sales amount” for use in the Corporate SME RWA formula under paragraphs 68 and 69. Stakeholders asked OSFI to provide guidance on what should be used as “annual sales” for this population to ensure consistent treatment.
| - Yes. This change has been made.
- The consistent parameter will be exposure amounts.
- The unsecured LGD floor of 25% would apply to unsecured exposures to small business loans treated under the Corporate SME risk weight function.
- For these exposures, annual sales should be set to CAD$7.5 million.
|
Stakeholders recommended removing the added requirement in paragraph 25(4) that stipulates that the maximum aggregated retail exposure should not exceed CAD$1.5 million. This is viewed in contradiction with paragraphs 25 (1) and (2) where exposures to individuals and residential mortgages are eligible for retail treatment regardless of the exposure size. This is not deemed consistent with the BCBS definition of retail exposures under the Advanced IRB Approach in the current CAR Guideline. Stakeholders’ view is that this requirement introduces unnecessary complexity and a lack of transparency in the Advanced IRB RWA reporting and disclosure, as the small amount of retail exposures exceeding the threshold would have to be reported using the corporate risk weights. | The requirement in part (4) of paragraph 25 will be maintained. However, OSFI has provided additional flexibility to allow institutions to confirm immateriality of the amount of exposures that breach this threshold on at least an annual basis. Immateriality will be defined as 2% of total retail exposures. When material breaches occur or an institution cannot comply with these rules, they should contact OSFI to discuss the treatment of these exposures. |
Regarding paragraph 67 and 68, the "consolidated group" definition is not aligned with paragraph 39. Stakeholders recommended aligning the two definitions and using the definition in paragraph 39 which relies on the accounting standard applicable to the ultimate parent of the consolidated group. They also requested alignment of the definition of “sales” under paragraph 68 and “revenue” under paragraph 38 to simplify data gathering. | OSFI has aligned the definition of consolidated group. Footnote 10 now refers to the same definition as in paragraph 39. Revenue for large corporate exposures may include income beyond that from sales. If an institution wishes to use revenue instead of sales for purposes of paragraph 68, they may do so provided it is always as conservative (i.e., there is no negative revenue that would result in revenue being less than sales). |
Stakeholders indicated that OSFI’s proposed capital requirements for mortgages insured by private mortgage insurers where a backstop guarantee is provided by the Government of Canada do not appropriately recognize all of the risk mitigation. They noted that the assumption of the 100% LGD on the deductible exposure is overly punitive and not reflective of the risk of a PMI-insured mortgage. If OSFI believes there is a need to modify the existing rules for the "deduction portion" under the backstop guarantee option, stakeholders requested that OSFI prescribe an LGD of 45% instead of 100% to align with private mortgage insurers’ unsecured LGD under the Foundation IRB Approach. As such, they also suggested that OSFI remove the multiplier of 2.2 in paragraph 271 under the Standardized Approach. | The revised rules for PMI-insured mortgages only require the use of the 100% LGD on the deductible exposure if an institution wishes to recognize the benefits of the Government of Canada backstop. This backstop is structured such that it will only be triggered if an institution suffers losses equal to 10% of the original principal, which aligns with the structure of the partial government guarantee of privately insured mortgages provided pursuant to the
Protection of Residential Mortgage or Hypothecary Insurance Act. As such, if the backstop is recognized, it comes with the implicit assumption of the institution having suffered losses equal to the deductible amount, which justifies the 100% LGD on this exposure. In addition, the revised rules for PMI exposures reflect the risk mitigation options available to institutions. That is, institutions can choose: i) whether or not to reflect the Government of Canada guarantee, and ii) whether or not to reflect the benefits of the insurance provided by the PMI. Institutions may choose the approach that minimizes capital requirements at any point throughout the life of the mortgage. While promoting flexibility for institutions, the revised treatment prudently reflects the potential for related risks to compound, while aligning to foundational elements of the capital framework (i.e., that there is no recognition of double default or double recovery). |
Stakeholders requested clarification on the inclusion of LGD in the footnote about seasoning for paragraph 207. Does this requirement apply only to the LGD rates that institutions use for accounts that are already in default? Would this requirement be satisfied if their calculations of Best Estimate of Expected Loss, which is used to calculate EL and RWA for defaulted accounts, takes time-since-default into account? | This requirement for seasoning related to LGD is simply stating that if there is evidence of seasoning effects (e.g., evidence that LGDs peak several years after default), estimates should be adjusted to account for the identified seasoning effects. |
Footnote 26 of paragraph 280 noted that “post-default advances and corresponding accrued interest can be captured in LGD or EAD estimates, provided it is done consistently across the institution”. Institutions typically have a large variety of asset classes/portfolios of distinct credit characteristics. It is conceivable that, for some asset classes/portfolios, it would make more sense to capture post-default advances and corresponding accrued interest in LGD; however, for some other asset classes/portfolios, it would make more sense to do so in EAD. If this is the case, would OSFI permit institutions to make such choices with appropriate justification? | Yes, institutions may choose to reflect post-default advances and accrued interest in LGD or EAD depending on the asset class. However, institutions are not permitted to switch back and forth between approaches for a given asset class. If an institution wants to switch approaches for a given asset class, they must notify OSFI. |
Regarding paragraph 286, given the dynamic nature of cities publicly available for the Teranet – National Bank House Price Index, stakeholders requested that OSFI be more prescriptive as to which cities should be used. In a previous version of the CAR Guideline, OSFI mentioned that 25 cities should be used. In the most recent public consultation version, it is 26 cities; however, as of the end of March 2021, the Teranet website has indices for 32 publicly available cities and it is expected that more cities could be added by the time the updated CAR Guideline comes into effect. Stakeholders asked OSFI to define the cities (similar to the current CAR Guideline with 11 cities) as well as the factors required for the Supplementary Capital Requirement indicators (SCRI) in Appendix 5-3 of the guideline. For loans outside the 26 cities, stakeholders asked OSFI to confirm that the composite-11 information should be utilized. | Paragraph 286 has been updated to reflect all 32 publicly available cities, and a list of the 32 cities has been included in Appendix 5-3. For loans outside that list of 32 cities, please use the composite-11 index. |
Regarding paragraphs 300, 302 and 304, the benefits of credit risk mitigation from both borrowers and guarantors can be recognized for capital purposes if an institution can establish that it can simultaneously and independently realize on both the benefits (e.g., collateral provided by the borrower and a third party guarantee), and in doing so not recognize double recovery. However, OSFI still imposed a risk weight floor to ensure that the risk weight resulting from PD or LGD adjustments must not be lower than that of a direct comparable exposure using the PD of the guarantor and the LGD for an unsecured exposure to the guarantor. This could be overly punitive and prevent the recognition of both the benefits even when the aforementioned conditions are met. Especially when the guarantee is received from a third-party bank, the unsecured LGD of 45% prescribed under the FIRB approach should be used. The collateral provided by the borrower could have brought down the LGD significantly (i.e., if the secured LGD of the obligor/guarantor can be used for setting the risk weight floor). The stakeholder asked if this would be an unintended consequence. They noted that the BCBS does not prescribe that the unsecured LGD of the guarantor must be used for the same purpose and sought clarification from OSFI that institutions cannot realize on both the guarantor and collateral provided by the borrower and therefore institutions can reflect one or the other but not both. | In a situation where collateral and a guarantee are both received, institutions may choose to reflect whichever is most beneficial from a capital perspective. In the case of collateral provided, there is no risk weight floor. OSFI does not believe there are any circumstances where lenders can simultaneously and independently realize on both a guarantee and collateral. That is, recover each dollar lost from both the collateral and the guarantor ($2 recovered for every $1 lost). In the case of a guarantee, the risk weight floor described is applicable. OSFI does not see any unintended consequences from this treatment and it is not an oversight despite the Basel III rules. |
Regarding paragraph 348, under the FIRB there is no capital incentive for a lender to establish security as a first lien lender if the collateral does not meet the narrow requirements currently outlined. The stakeholder’s default data and industry data both support that priority of debt is a better predictor of recovery rates than the value of collateral (excluding Real Estate, ABL, etc.). By not incentivizing lenders to have a secured position in the capital structure (and no further incentive for structures with meaningful junior debt), there could be unintended consequences of reduced recovery rates in the next downturn. The stakeholder further noted that consistency between FIRB and AIRB collateral is essential given they are both IRB approaches. As such, they requested that OSFI allow the use of a debt cushion methodology for large corporates based on the existing parameter estimation methodology. They further noted that the adoption of the FIRB approach fails to give recognition of the substantial collateral available under a general security agreement (GSA) if a counterparty goes into default. | GSAs are deemed eligible collateral under the FIRB approach provided the assets of the counterparty include eligible collateral. Only the assets included as eligible collateral may be recognized as credit risk mitigation when reflecting the benefits of a GSA. If institutions are having difficulties with particular aspects of the requirements related to GSAs, they may reach out bilaterally to OSFI to discuss them. It should be noted, however, that a debt cushion approach is not permitted under the FIRB approach. |
Chapter 6 - Securitization |
---|
Paragraph 4 states that transactions of real estate cash flows may be considered specialized lending rather than securitization. The stakeholders inquired about what would warrant treating a transaction as a specialized lending exposure rather than a securitization. | Securitizations are limited to the tranching of credit. Securitizations structured through SPVs (specialized lending or ABS) would not fit into the securitization framework if there is no credit enhancement. |
Stakeholders asked whether an institution that participates in a non-AAA rated securitization exposure (e.g., an “A” rated exposure from a specific securitization transaction) is allowed to re-tranche this exposure (e.g., into “AAA”, “AA” and “A” tranches) to optimize capital without its exposures being treated as re-securitization exposures. | The exposure will not be considered a re-securitization exposure so long as all cash flows may be replicated by a single tranching of the transaction. This would be the case if the “A” tranche is re-tranched. If, however, securitization is used to pool tranches from separate securitizations and then re-tranche the pool of tranches, the resulting tranches would be considered re-securitization exposures. |
Stakeholders suggested that if a prepayment speed is not used or factored into the pricing of the transaction, then one of the alternative fall-back approaches (i.e., if the prepayment rate in #1 of paragraph 26(a) is not applicable to be used, then the prepayment rate specified in #2 or #3 of paragraph 26(a)) should be used): Under approach 1, both transaction pricing and hedging take into account the economic life of the transaction and would be a market-based proxy for the prepayment rate. Under approach 2, both portfolio-based prepayment analysis and origination curve analysis should be allowable approaches. Under approach 2, a suitability qualifier should be added because data from the originator’s other transactions is more informative than a country-level average. Under approach 3, a materiality qualifier should be added. | OSFI has added language clarifying that where an approach is not applicable, the lower of the remaining approaches would be used. OSFI has also modified paragraph 26 to address concerns with respect to allowing hedging to be used as a proxy for the prepayment rate, allowing both portfolio-based and origination curve analysis to be applied, and adding a materiality qualifier to the transaction-level data. Rather than implement a suitability qualifier however, OSFI has indicated explicitly that data from the same originator should be applied if available and, if not available, then data from the same country should be applied. |
Regarding paragraph 26, a stakeholder noted that the approach for revolving transactions appears to be so onerous and time consuming that it may not be usable. One would need to update this calculation monthly and, as updated pool cuts may not be received monthly, make further manual adjustments each month for differences between actual paydowns at a program level vs expected paydowns. They further noted that not all of these revolving pools will be fully drawn during the commitment period. | Additions to this paragraph have been incorporated to clarify that the calculation must be performed with refreshed transaction cash flow data at least quarterly and to clarify that the undrawn portions may remain undrawn under this calculation. |
For commitments, stakeholders proposed an alternative approach be available to calculate the effective maturity should the approach proposed by OSFI in paragraph 26 for revolving transactions not be used. The other alternative would allow institutions to calculate the maturity as the sum of the remaining commitment term and the expected weighted average maturity for a new draw, assuming that draw happens at the end of the commitment term. | This additional conservative approach has been added as an available approach under paragraph 27 under appropriate conditions. |
Regarding paragraph 30, stakeholders requested clarification on the new requirement for 29(a) through (c), instead of only (a). | This change from “(a)”to “(a) through (c)” confirms that the maximum 40% risk retention test must be met for synthetic securitizations. |
A stakeholder noted that even if there were draws on the backstop liquidity facilities, they would never be 100% drawn. As a result, they recommended that OSFI consider applying a CCF somewhere between 50% and 100% for these ABCP backstop liquidity facilities, provided that they are for bank-sponsored ABCP conduits funding traditional asset securitizations. | OSFI has not received sufficient evidence that liquidity facilities are not expected to behave as a direct credit substitute under stress to modify the current treatment. |
Stakeholders requested that OSFI modify the SEC-SA and SEC-IRBA approaches to allow institutions to recognize the risk-mitigating effects of excess spread. | OSFI did not receive any feasible suggestions on how excess spread might have been incorporated into the SEC-SA and SEC-IRBA approaches. |
Regarding paragraph 79, stakeholders requested that OSFI clarify that if a transaction were to include credit insurance, then institutions could use the risk weight attributable to the credit insurer’s rating and that this credit insurance does not need to meet the criteria of a Basel guarantee. As an example, EDC credit insurance could be included in trade receivables transactions. | Guarantees may be reflected as a second step after the application of the top-down approach. |
Stakeholders asked OSFI to clarify that the pool for the reference securitization exposure does not have to be shared with the asset pool backing the tranche with an inferred rating. Specifically, if the asset pool backing both exposures (reference and inferred rating) is from the same seller selected based on substantially the same eligibility criteria and concentration limits on a no-adverse selection basis, and does not have any meaningful difference in the credit risk profile, then a separate asset pool from the pool backing the reference securitization exposure can be used. | While OSFI recognizes risk profiles of separate asset pools from the same seller based on the same criteria are likely to be similar, differences may occur due to vintage or other effects. |
Chapter 7 - Settlement and Counterparty Risk |
---|
Stakeholders sought clarification on the application of the clause in paragraph 37 to the Effective Maturity (M) calculation under IMM and confirmation that institutions do not need to change their models. They further requested clarification in the text with respect to Specific Wrong Way Risk (SWWR), Jump-To-Default, and resetting, including on the treatment of reset products for the effective maturity calculation under SA-CCR. | This clause was mistakenly inserted in a generic manner when it was indeed meant only for resettable trades. Updates have been incorporated to reflect this. |
Stakeholders expressed appreciation for OSFI's inclusion of Specific Right-Way Risk (SRWR) transactions in the SA-CCR framework under paragraphs 66-69. Given that institutions were required to comply with the SWWR guidelines by Q1 2020, they requested that OSFI consider an earlier adoption of the SRWR treatment provided that institutions have robust due diligence processes in place. This would allow for consistency in the treatment of specific risks in the capital framework and ensure that the pricing of the trades reflects the economics of the trade. Stakeholders further requested that OSFI adopt the new SRWR rules on a current basis via an FAQ process. This is supported by the fact there is no corresponding BCBS requirement. | The rules for SRWR will be implemented as planned in Q2 2023. OSFI does not believe there is appropriate justification to early adopt these rules and without doing so for other rules. |
Paragraph 134 singles out interest rates in respective currency as the only asset class where lambda may be applied. A stakeholder requested an extension of this option to all relevant asset classes to address the possibility of negative price and strike values. For example, there are commodity spread option trades where the price and strike are both negative, rendering the delta calculation ineffective. In the case of interest rate products, they requested additional granularity of lambda to be curve specific. | OSFI has added the potential for the commodity asset class to also have this same issue. |
Footnote 37 of the draft Chapter 7 requires that inflation trades be captured under the interest rate asset class but form a separate hedging set. Stakeholders requested that interest rate and inflation swaps (and correlation between the two) sets be given cross-hedging set correlations consistent with the SA-CVA framework. | Inflation trades must be placed in their own hedging sets under the SA-CCR and as such no cross-hedging benefits will be recognized in the SA-CCR. |
OSFI is requiring a regulatory deduction for default fund contributions to non-QCCPs under paragraph 72 of Chapter 2. However, paragraph 185 in the current Chapter 4 and paragraph 209 in the draft Chapter 7 state that a risk weighting of 1250% is only required. Can OSFI please confirm that the proposed regulatory deduction treatment in Chapter 2 is correct and that OSFI will update the other related sections of the CAR Guideline accordingly? | This provision has been removed from the draft Chapter 7 given its replacement with regulatory deduction described in Chapter 2. |
Chapter 8 – Credit Valuation Adjustment (CVA) Risk |
---|
Stakeholders requested that Advanced Internal Ratings-Based approach (AIRB) or Foundation Internal Ratings Based approach (FIRB) institutions that have already received OSFI approval for internal rating be able to utilize the internal rating mapping to an external rating without additional approval from OSFI. Stakeholders believe existing procedures already approved for AIRB purposes should also be allowed for the purposes of CVA capital calculations. | Similar to the current rules, institutions with approved IRB models may map their internal ratings to external ratings for purposes of assigning a risk weight to a counterparty under the BA-CVA. |
Stakeholders noted that restricting eligible BA-CVA hedges to only single name Credit Default Swap (CDS), single name contingent CDS and index CDS per paragraph 18 could create an incentive to hedge Canadian derivatives counterparties in a less than optimal way. For example, Canadian provincial governments have liquid bonds but no liquid CDS. Hedging CVA with derivatives referencing the underlying bonds may have less basis risk than hedging with CDS. Stakeholders requested that OSFI add more hedging products, specifically bonds, total return swaps on bonds and bond indices, and risk participation agreements to the list of eligible CVA credit hedges under the BA-CVA approach. This approach would better align the recognition of hedges in BA-CVA capital with the management of accounting CVA risk. | The BA-CVA is a simple approach to capitalizing CVA risk. Institutions wishing to make use of a broader set of instruments to hedge CVA risk must use the SA-CVA approach. |
Stakeholders noted that the Canadian energy sector leverages oil reserves to secure derivatives and loan facilities. To enable competitive derivative pricing in the Canadian market, this security has to be considered in the LGD used for CVA pricing. Stakeholders asked that OSFI clarify the items below to better align SA-CVA capital with accounting of CVA and LGD under FIRB / AIRB. - The derecognition of collateral for LGD for collateral already included in exposures similar to current paragraph 31, Chapter 4.
- The recognition in LGD for non-CSA security or other eligible physical collateral which impacts the seniority in the recovery process.
| CVA is a market price of risk and the LGDs used in its determination must be market based. The rules clearly state that collateral pledged by the counterparty does not alter the seniority of an exposure and, as such, cannot be used to modify the market implied LGD. |
Regarding paragraph 47, stakeholders requested OSFI clarify that institutions, in addition to small values of risk factor shifts according to the text, may use a larger risk factor shift where necessary provided proper validation has been conducted to confirm appropriateness. The concerns with small values of risk factor shifts is that they may materially restrict some banks from producing higher quality sensitivities based on their models such as Monte Carlo simulations for the calculation of CVA. Specifically, small shifts can often lead to noisy or unstable values of the sensitivities for the risk factors involved. | OSFI agrees and recognizes that there are times when larger shock sizes are needed in order to improve stability of measurement with Monte Carlo simulation. Such an approach would be subject to internal governance and control arrangements to identify when such efficiencies are warranted with the use of larger shocks. |
Stakeholders noted that paragraph 62 specifies that FX vega sensitivities are to be measured by shifting volatilities for exchange rates between an institution’s reporting currency and other currencies. However, it is common for FX volatility inputs to be against USD (the base currency), as such volatilities are most commonly observed in the market and therefore also used for hedging purposes as well as model calibration. Stakeholders proposed that OSFI permit institutions to measure FX vega to volatilities between the base currency in an institution’s CVA model and other currencies (including the reporting currency), rather than between the reporting currency and other currencies. | OSFI is reverting to the language in the 2017 revisions to CVA capital requirements that provides better detail on the use of two FX vega volatility sensitivities with the base currency as opposed to utilization of sensitivities with respect to the reporting currency. The substance of the policy did not change so it represents a valid representation of the determination of FX vega risk factor sensitivities. |
Stakeholders noted that, in the revised CVA framework, all financial entities are included in the same sector bucket (counterparty credit spread risk class). This sector bucket is very broad, capturing a diverse set of counterparties including highly regulated institutions with multiple financial business lines (commercial and investment banks, insurance companies), unregulated and highly leveraged institutions (hedge funds, private equity), and institutions with narrowly defined missions based on earning appropriate returns for investor stakeholders (pension funds, mutual funds, asset management accounts). Stakeholders requested that OSFI consider increasing the granularity of risk weights for financial counterparties to align and improve the representation of underlying CVA risk (e.g. differentiating risk weights between regulated and unregulated financials). | No additional granularity has been provided in the risk weight buckets for financial counterparties. If there is evidence of changes in key international rules making changes to the granularity and risk weights for financial counterparties, OSFI will review and determine the correct course of action for Canadian institutions. OSFI is, however, exploring the possibility of collecting data with the additional granularity for financial counterparties in future regulatory returns. |
Stakeholders welcomed the exemption provided on the capitalization of market risk hedges of collateral valuation adjustment (ColVa) and the exposure component of funding valuation adjustment (FVA) to enable the management of valuation adjustments that are well accepted within current accounting standards. Stakeholders suggested that OSFI consider a further exemption in the future should other valuation adjustments (e.g., capital valuation adjustment (KVA)) be well integrated into accounting standards. The proposed annex clarifies treatment of FVA hedges and only considers an exemption for market risk hedges for FVA, but explicitly excludes hedges for the credit / funding component of FVA. Institutions are actively managing the general market risk component of the funding spread (e.g. through proxy basket or index hedges) so excluding these hedges from the exemption and including them in market risk as open exposures penalizes institutions’ management of this key P&L component of FVA. Stakeholders further noted that backtesting of hedges including the xVA component would be very challenging as VAs are not typically measured under VaR and backtesting of total P&L would be challenging. Stakeholders proposed to extend the exemption that is to be applied to the market risk component of FVA to the prudent P&L management of all FVA components. They requested that OSFI exempt general market risk hedges for funding spreads from market risk capital as this is critical to enable management of the credit/funding risk P&L component of FVA. In addition, they requested removing the backtesting requirement from the required conditions, and that OSFI provide flexibility to leverage current internal control frameworks subject to internal audit and OSFI’s satisfaction. | The rules text allows for the exemption from market risk capital requirement for hedges of xVA risk whereby a market standard is well established and more or less consistently used. OSFI feels that there has not been sufficient convergence in market practice for hedging other areas of xVA risk to permit these hedges to be exempted from market risk capital requirements. OSFI agrees with the approach that institutions take with respect to validation of valuation and hedge effectiveness. Conducting a daily VaR backtest of risk measure for xVA risk can be alternately achieved through the combination of P&L attribution of relevant risk factors and assessing the correlation relationship between xVA and market risk hedges. |
Chapter 9 – Market Risk |
---|
Stakeholders asked OSFI to clarify its interpretation of paragraph 61 in draft Chapter 9 given their understanding that, while accounting classification and market risk capital treatment should align where possible, they will not necessarily be the same, and neither treatment should dictate the other. | OSFI has not incorporated any change to this paragraph. The chapter includes other criteria for instruments to be classified in the trading book – i.e., short term resale, profiting from price movement, locking in arbitrage profit, etc. Daily fair value accounting through the recognition of any valuation change in the profit and loss account is the treatment required for trading book instruments. |
Stakeholders noted the potential divergences between how different institutions classify “liquidity management” and proposed the following: - Amend footnote 9 in draft Chapter 9 to clarify either the liquidity management purpose or the accrual accounting methodology as a necessary and sufficient condition for exempting repo-style transactions from the presumptive list; or
- Clarify the definition of “liquidity management” either directly in the CAR Guideline or through an FAQ to avoid ambiguity.
| OSFI has amended the text. Repo-style transactions that are (i) entered for liquidity management or (ii) valued at accrual for accounting purposes are not part of the presumptive list. Institutions must have documentation for the definition of liquidity management and internal control processes to monitor these transactions, which should be made available to OSFI upon request. |
Stakeholders suggested revising the exemption for arm’s length transactions between the trading book and the banking book. They suggested amending the “CAD-denominated” requirement to “CAD-denominated or Canadian-based issuers”. In addition, stakeholders sought OSFI’s opinion on permitting the movement of all eligible securities that are Level 1 and 2A high quality liquid assets (HQLA) per Chapter 2 of OSFI's Liquidity Adequacy Requirements Guideline. | OSFI has amended part of the related guideline language. Institutions are permitted the following exclusions from the restrictions on moving instruments between regulatory books: - CAD-denominated Level 1 and Level 2A HQLA; and
- non CAD-currency denominated Level 1 and Level 2A HQLA issued by Canadian entities.
OSFI has also included clarifications in the text pertaining to the definition of issuance by Canadian entities. |
Stakeholders asked for an exemption for institutions to purchase new issuances of their own bank-stamped Banker’s Acceptances (BAs) from their dealer. Canadian BAs are a short-term investment vehicle to park excess liquidity and a key source of funding for small and medium size corporate borrowers. | OSFI has incorporated revisions to note that where an institution purchases new issuances of its own stamped Bankers Acceptances from its dealer, such securities do not need to be included within the restrictions on moving instruments noted in the text between the banking book and trading book. |
Stakeholders requested clarity with respect to: - the application of “the Residual risk add-on” (RRAO); and
- the “exact match” definition and the inclusion of multiple transactions within the context of Internal Risk Transfers (IRTs).
| OSFI has removed the original sentence related to the application of the RRAO charge. OSFI added clarifications as to when institutions are expected to capture any RRAO between the internal and external hedge with respect to instruments with exotic underlying and instruments bearing other residual risks, consistent with section of Chapter 9. OSFI has also added language to refer to the permission to include multiple transactions. |
Stakeholders requested that IRT requirements should only apply to IRTs executed on or after the start of the year of FRTB implementation (i.e. November 1, 2023 for institutions with an October 31 fiscal year end). Stakeholders pointed to a need for grandfathering of legacy trades which has become more pronounced as a result of pandemic-related changes in the industry. Existing IRTs extend well beyond the implementation date of the FRTB due to the ALM structure at Canadian institutions. Further, the pandemic has triggered a surge in surge in deposits/loans and consequent hedging requirements which mature well beyond Q1 2024. | OSFI has amended the language for internal risk transfer of general interest rate from the banking book to the trading book. This is to clarify the application of the new requirements for trades executed on or after the beginning of the institution’s fiscal Q1 2024. |
Stakeholders requested that OSFI implement the original BCBS requirements on the frequency and granularity of SA capital and HPL calculation, wherein calculations of SA would be on a monthly basis and on a higher level than desk level, such as legal entity or enterprise level. | OSFI has revised the relevant text to revert back to the BCBS language. Capital requirements under the SA must be calculated and reported to OSFI on a monthly basis. OSFI may request additional reporting requirements or information as part of its supervisory activities. |
Stakeholders requested that OSFI extend the option given for undecomposed Equity Investment Funds (EIFs) under paragraph 147 to undecomposed indices. Namely, they requested that OSFI allow the conservative capital treatment of either placing indices in the “Other sector” equity bucket or applying the mandate / hypothetical approach as long as daily prices and the index description and objective are available. | Guideline text has been amended accordingly to permit a no look-through approach for such index instruments where a look through approach is not possible. This amendment is consistent with the treatment of EIFs. |
Stakeholders asked to extend the treatment and governance granted to an investment grade bond or large cap equity hedging a Total Return Swap (TRS) to a bond forward hedging a Level 1 High Quality Liquid Asset (HQLA). The stakeholders’ view is that the revised market risk rules present an exaggerated charge for derivatives dynamically hedged with their underlying securities. Specifically, bond forwards, where the underlying is a Level 1 HQLA, are widely used in the Canadian markets by a diversified user base consisting of participants in the economy including asset managers and pension plans, small banks and corporations, and provincial treasury departments. | OSFI has extended the treatment and governance previously granted to an investment grade bond or large cap equity hedging a TRS to a bond forward hedging a Level 1 HQLA. The mismatch applied between long and short positions will be capped at 40 days under both the SA and the IMA. |
Stakeholders requested that OSFI lower the coverage requirement significantly to preserve incentives to adopt the IMA and to reduce uncertainty for IMA approval status and initial application. A lower threshold would also align better with international standards. | Institutions applying for the use of internal models will be required to meet an internal model’s coverage threshold of 50% at all times. To simplify the framework, OSFI will not implement a separate higher application threshold but will need to be satisfied through the approval process that the ongoing 50% threshold can be maintained. |
Stakeholders requested that OSFI exclude non-fair market valued (non-FV) positions from APL and HPL requirements at desk and bank wide levels, including the HPL reporting requirement for SA in paragraph 108 of draft Chapter 9. Instead a risk-theoretical P&L (RTPL) back-testing process at the desk level could be established as compensating control for these banking book desks. | OSFI has not incorporated any changes to the text. Under the revised market rules, the risk subject to market risk capital include FX and commodity instruments in the banking book. The text does specify in the terminology that APL includes FX and commodities risks for banking book instruments. OSFI continues to review the applicable and acceptable methodology at institutions related to this requirement. |
Stakeholders requested that subsequent OSFI or BCBS Quantitative Impact Studies (QIS) revisit the proposed statistical test thresholds described in paragraph 356 during parallel reporting or during the grace period to ensure that a modelled approach is not unduly penalized. | OSFI has not incorporated any changes to the text. The BCBS may conduct further QIS exercises and OSFI may collect further information regarding this item as part of parallel reporting prior to implementation. |
Regarding paragraph 371, stakeholders view that capping the Liquidity Horizon (LH) at instrument maturity as at odds with the regulatory LH definition (which is per risk type) and risk management practice, creating an operational burden. Institutions should be allowed to assign all instruments to the regulatory LH. Stakeholders requested capping the instrument maturity optional should an institution chose to do so. | OSFI has confirmed that liquidity horizon should be capped at the maturity of the related instrument when the maturity of the instrument is longer than the prescribed liquidity horizon assigned. If the maturity of the instrument is shorter than the respective liquidity horizon of the risk factor, the next longer liquidity horizon length (10, 20, 40, 60 or 120 days as set out in text) compared with the maturity of the instrument itself must be used. |
Regarding the calculation of capital for NMRFs under paragraph 375, stakeholders requested that institutions be allowed to include modellable risk factors in the SES charge for curves and surfaces given that, as a result of not including them, broken hedges between non-modellable and modellable risk factors (or arbitrage violations) occur. The NMRF charge is determined by only shocking the NMRF in isolation, leaving all other risk factors constant. However, in the case of a NMRF representing a portion of a curve, surface or cube, only the risk factors in one bucket are shocked and the rest of the curve, surface or cube would remain unchanged. This could lead to large fluctuations in P&L from broken hedges or inconsistent results (e.g., arbitrage violations). | OSFI has not incorporated any changes to the text. The construct of the FRTB keeps separate the potential impact of modellable risk factors on the capital charge determined for non-modellable risk factors. |
Regarding paragraph 383, in the event that other national regulators (e.g., EU and US) lower the Probability of Default (PD) floor from 3bps to 1bp for their respective debt securities, Canadian institutions with exposure to sovereigns meeting local criteria should be able to assign a lower PD floor to those sovereigns. This is viewed as critical given the importance of non-domestic sovereign debt to institutions’ prudent management of HQLA requirements in foreign currencies and would ensure a competitive and level playing field across jurisdictions. | OSFI has not incorporated any changes to the text. OSFI is not compelled to meet the policies implemented by other jurisdictions; however, OSFI does consider institutions’ ability to compete effectively and take reasonable risks. |
Paragraph 391 indicates that an institution may use a simplified modelling approach for equity derivative positions with multiple underlyings. Stakeholders proposed that a similar simplified approach be permitted for non-correlation trading portfolio credit derivative positions with multiple underlyings. | OSFI does not believe that a change in the text of Chapter 9 is necessary. OSFI notes that the text in Chapter 9 aims to show one example for equity derivatives products within the category of positions with non linear positions. In this case, a simplified modelling approach may be adopted subject to OSFI approval. The understanding is that a similar approach can be adopted for other similar instruments pending OSFI approval. |
Stakeholders proposed that OSFI remove the formal requirement for a risk factor backtesting program under paragraph 427. Backtesting at the desk level provides sufficient granularity to cover backtesting results by risk factor classes including relevant correlation between risk factors. Furthermore, any required investigation into backtesting breaches typically includes investigation into specific risk factors within the class for relevant desk level breache. | OSFI agrees with the content of the proposal and the removal of such requirement and notes that this sub-paragraph is within the context of “Example of the application of the principles for risk factor modellability”. The text clarifies that OSFI may use discretion regarding the type of evidence required of institutions to provide risk factor modellability. |