Office of the Superintendent of Financial Institutions
The Office of the Superintendent of Financial Institutions (OSFI) issued a final version of the Instruction Guide for the Preparation of Actuarial Reports for Defined Benefit Pension Plans (the Guide) in November 2020 following consultation that was initiated in December 2019. The purpose of the Guide is to set out the reporting requirements of actuarial reports filed with OSFI for defined benefit pension plans, including those with a defined contribution component.
Almost all respondents provided comments on section 2.7.4 – Alternative Settlement Methods of the Draft Guide, which includes OSFI expectations for pension plans using a replicating portfolio approach. Given the extent of the changes outlined in the Draft Guide and comments received, the consultation led OSFI to postpone significant changes to a later date. Only changes required to clarify the previous text and OSFI expectations related to disclosure requirements were made to section 2.7.4 of the Guide.
Further review has since been conducted and enhancements have been formulated. The attached table in Annex 1 presents a summary of the comments that were received and how OSFI has addressed them. We thank those who participated in the consultation process.
Some of the comments that were received (e.g. solvency reserve account) relate to changes to the current legislation for federally registered private pension plans, which does not fall under OSFI’s responsibility. These comments have been forwarded to the Department of Finance.
The revisions to the Guide provide updated expectations for plans using a replicating portfolio approach. The revised section 2.7.4 of the Guide is being issued in draft form so that pension plan stakeholders have the opportunity to provide input on these changes. It is expected that these modifications will apply to actuarial reports with a valuation date on and after December 31, 2021.
Questions and comments concerning these changes may be sent to Marc Sauvé, Senior Manager, Actuarial in the Private Pension Plans Division, by email at email@example.com. Comments should be provided no later than September 10, 2021.
A non-attributed summary of comments received regarding the revised section 2.7.4 of the Guide, along with OSFI’s response, will be posted on OSFI’s website when the final version of the Guide is released.
Managing Director Private Pension Plans Division
For solvency valuation purposes, the CIA Guidance on alternative settlement methods describes four alternative settlement approaches that may be considered, as follows:
OSFI allows options #1 and #2 only, but should also consider permitting options #3 and #4. In accordance with CIA Guidance, the actuary may contemplate an exercise of regulatory discretion or change in legislation to permit the settlement of benefits under one of these alternative method options.
Options #3 and #4 do not comply with the current legislation (see footnote 93 of the Guide).
Assuming a change in the legislation, for which the Department of Finance is responsible, would not be consistent with OSFI’s mandate of administering the legislation as it is currently. Also, the objective of performing a solvency valuation based on the plan provisions as at the valuation date would not be met.
Some respondents have signaled that consistency between plans will not be achieved by shifting to a life insurance model with extremely high thresholds due to differences within the models themselves used by pension plans.
One respondent suggested that OSFI commit to signing off on alternative approaches to the one described in the Guide, ahead of the first actuarial valuation subject to the revised Guide.
OSFI agrees that model risk could lead to significant differences in results from one plan to another if a common approach is not followed to validate and calibrate the models.
OSFI recognizes that reviewing, testing and providing guidance on how tail risks are addressed and calculated should be consistent between plans. Leveraging existing methods used by pension plans and relying on CIA Guidance for the use of models and accepted actuarial practices for stochastic modeling is deemed to be sufficient to provide this consistency.
OSFI was asked to explain its concerns regarding the approaches currently being used and why the use of Conditional Tail Expectation (CTE) and a one-year period model was instead proposed in the Draft Guide.
A few respondents argued that OSFI should permit other models that may be developed in the future, if they can be shown to provide the targeted level of benefit security. Other respondents expressed the view that the use of life insurance models is inappropriate for pension plans, notably because excess funding in a pension plan is released as trapped surplus unlike for insurers, and also because these models are too complex for administrators and beneficiaries to understand.
The main objective for updating OSFI expectations with respect to the replicating portfolio approach is to ensure similar benefit security across defined benefit pension plans subject to the federal private pension plans legislation, by addressing issues such as:
In light of the feedback received, OSFI concluded that imposing the use of a CTE measure and a one-year period model taking into account remaining risks thereafter may not be practical to achieve its stated objectives. The Guide was adjusted to align with the approach predominantly used by pension plans when stochastic modeling is performed and featuring a Value at Risk (VaR) measure over the runoff period of liabilities.
Further, the revised section 2.7.4 moves away from trying to reproduce how a plan using a replicating portfolio would be funded after the termination date (e.g. intervals or times for filing the actuarial report, payment of deficit, utilization of surplus …). The requirement for adjusting the available financial support from the employer based on the termination scenario was removed. The focus is on the security of benefits at the valuation date, not at a future valuation date.
Many respondents to the consultation on the Draft Guide agree that the level of benefit security under the replicating portfolio approach should be high, though not as high as the implied 99.5% probability.
OSFI considers that the level of benefit security should be similar from one plan to another. The replicating portfolio approach should not be interpreted as providing a form of partial solvency funding relief to some pension plans.
For parity with plans subject to the annuity proxy, the starting point remains at the 99.5% level, which is the underlying benefit security level of pension benefits guaranteed by life insurance companies.
A number of suggestions were made to adjust and clarify the approach in setting the appropriate margins for economic risks.
In consideration of the feedback received, OSFI has adjusted the Guide to require that the provision for adverse deviations be aligned with a confidence level that the pension plan will be fully funded over the runoff period of liabilities ranging from 85% to 99.5%, based on the credit rating of the employer. Those levels are currently deemed reasonable in consideration of OSFI’s stated objectives.
Stochastic modeling requires the generation of a high number of scenarios, commensurate to the confidence level to be achieved. However, the earlier reference to 10,000 scenarios was deleted to avoid confusion, as it was not an explicit requirement but an indication only that some confidence levels may require so many scenarios.
Some respondents wanted to better understand how the stress test to the best estimate assumptions were determined.
A few respondents asked why the provision for non-economic risks was apportioned based on the provision for economic risks. It was also suggested to remove the provision for non-economic risks.
About mortality more specifically, it was noted that most plans using the replicating portfolio are sufficiently large to have credible data (or partially credible data) to adjust the mortality rates so that the imposition of an arbitrary margin is unnecessary. With respect to longevity, it was suggested that the margin should be decreased when a base table resulting in longer life expectancies than that of the CPM2014 table is used.
With respect to plan expenses, it was argued that economies of scale could be generated in the context of unfavorable scenarios and that an additional provision for expenses may not be required. There was also a comment on the small size of the provision if a margin for additional expenses as a result of the stress testing of the mortality level and trend was introduced.
Also, one respondent noted that data risk is a component of non-economic operational risk for which a provision may be considered.
The levels of stress test for non-economic risks aim to align with a VaR at the 99.5% confidence level, based on expert judgment. The Guide was revised to allow for a reduction in the provision based on the credit rating of the employer. A simple method for the determination of the provision for non-economic risks was previously the only permitted method. The Guide was clarified to indicate that other methods may be used.
The Guide already confirms that mortality tables and improvement scales should follow CIA Standards and Guidance, which suggests in particular that large plans perform regular mortality studies, which provides some comfort in the level of the best estimate assumption. Nevertheless, OSFI considers that some uncertainty still remains, which explains why a margin is included for the mortality level. The margin for longevity trend also remains unchanged for all plans since future mortality improvement is subject to a high level of uncertainty and discussion.
While margins for some non-economic risks (e.g. expenses) may be small, the related risks are not insignificant and OSFI considers that a provision should still be calculated. In consideration of the feedback received, OSFI has adjusted the Guide:
Other non-economic risks are generally considered not significant compared to those identified in the Guide. As such, no specific margins with respect to these other risks are required.
Many comments were received on the determination of the available financial support from the employer. Above all, respondents consider that OSFI’s criteria should:
Several submissions argued that OSFI should permit lower margins to be used where it is reasonable to expect that there would be available financial support from the employer after plan termination.
Respondents suggested that the Guide make clear that all private sector employers, and not only crown corporations, would be able to reduce solvency liabilities of their pension plan with demonstrated financial strength at the time of the valuation.
Some respondents suggested that the amount of available support be based on the credit rating of the employer determined by one of the recognized rating agencies operating in Canada
Other respondents proposed that the actuary should be able to rely on an attestation of financial health of the sponsor from the plan administrator.
OSFI agrees that the amount of available support be based on the risk profile of the employer. Such amount will not require an actuarial opinion, will be readily available and be consistent for all plans.
The adjustment to the provision for financial support from the employer was revised. Given the priority level of pension plan obligations when an employer is bankrupt or insolvent, the Guide was adjusted to determine the level of financial support of the employer based on the credit rating of subordinated debt.
Some respondents asked OSFI to expand the possible universe of high-quality fixed income investments included in the replicating portfolio and to explicitly permit the use of private debt or similar private fixed income products that would meet an equivalent minimum credit rating.
OSFI has adjusted the Guide to clarify its expectations about the content and quality of the assets included in a replicating portfolio. The definition of “high-quality” remains unchanged.
Some submissions mentioned that disclosure of information and assumptions related to economic models used to develop the replicating portfolio could contain proprietary information.
The Guide was not adjusted. Transparency and appropriate disclosure are part of good governance. Consistent with paragraph 3260.09 of the CIA Standards, OSFI expects actuaries to provide sufficient details in their actuarial reports to enable another actuary to assess the reasonableness of the data, assumptions and methods used.
The Draft Guide required an extensive list of disclosures which would be onerous for ongoing pension plans. Several respondents recommended that the actuary should have some latitude to adapt this list of disclosures to reflect differing model features.
These comments were addressed in the revised Guide issued in November 2020.