The Deposit Insurance Reserve Fund Adequacy Assessment Framework Report

The Financial Services Regulatory Authority of Ontario (“FSRA”) is an independent regulatory agency created to improve consumer and pension plan beneficiary protections in Ontario. FSRA’s objects when it comes to credit unions (as they are articulated in the Financial Services Regulatory Authority Act, 2016 (“FSRA Act”) are to:

1. provide insurance against the loss of part or all of deposits with credit unions;
2. promote and otherwise contribute to the stability of the credit union sector in Ontario with due regard to the need to allow credit unions to compete effectively while taking reasonable risks; and
3. pursue the objects set out in clauses (a) and (b) for the benefit of persons having deposits with credit unions and in such manner as will minimize the exposure of the Deposit Insurance Reserve Fund (“DIRF”) to loss.

FSRA’s Annual Business Plan 2021-2024 (“ABP”) sets out FSRA’s priorities, strategic direction, financial overview and supporting activities. One of FSRA’s key priorities for the sector in this fiscal year is to enhance the Deposit Insurance Reserve Fund Adequacy Framework (“DIRF Framework”). As part of this important priority, FSRA is looking to engage the sector. The ongoing work on the DIRF Framework will lead to a report to the Minister of Finance on DIRF adequacy later in the year.

This consultation paper supports FSRA’s priority 6.2 in its ABP and looks to enhance the DIRF Framework through a transparent approach, alignment with international standards and enhancement of the model used to assess DIRF adequacy. In this paper, FSRA is sharing its approach and thinking about the DIRF Framework and model, providing a summary of the report from last year’s DIRF review and model and is seeking sector feedback on the proposed enhancements (including high level view of scenarios). 

To further this important priority, FSRA established and met with a Technical Advisory Committee on Data Strategy and Digital Transformation (“TAC”). At the inaugural meeting on May 7, 2021, committee members provided valuable feedback on the proposed (a) approach to the DIRF Framework; (b) parameters around this year’s stress testing model; (c) data needs and data collection strategy. FSRA thanks all the TAC committee members for their contributions (which have been considered in the FSRA approach articulated in this paper). FSRA is now looking to collect broader sector feedback on the approach and model features.  

Consultation questions are included on page 11 and stakeholders are asked to submit their feedback no later than September 9, 2021.

Back to top

DIRF adequacy is essential to sector stability. However, FSRA recognizes that there is a cost to the sector since it is a use of capital that cannot otherwise be deployed to service a credit union’s members and/or grow the institution. Accordingly, FSRA is seeking to develop better tools and information sources (in consultation with the sector) to allow FSRA to better balance the need for a DIRF that is sufficient to ensure deposit protection and stability while not putting an undue economic burden on the sector.

FSRA’s proposed approach is based on transparency, principles of proportionality and engagement with the sector. Once FSRA develops a better model for assessing DIRF adequacy, the process will be ongoing to refine it, improve the data and identify scenarios and other evolving factors that may affect the analysis and conclusions of adequacy.

The purpose of this consultation paper is to provide the sector with the results of the 2020 DIRF adequacy assessment review and seek input on the future approach and model features.

1. Statutory framework

The Credit Unions and Caisses Populaires Act, 1994 (“CUCPA”) requires FSRA to maintain a Deposit Insurance Reserve Fund (“DIRF”). In addition, FSRA has a statutory mandate related to the DIRF to provide insurance against the loss of part or all of deposits with credit unions and to pursue the objects set out in the Financial Services Regulatory Authority Act, 2016 (“FSRA Act”) with the benefit of persons having deposits with credit unions and in such manner as will minimize the exposure of the DIRF to losses.

Further, as outlined in the CUCPA, the DIRF may be used to pay for the following:

1. deposit insurance claims;
2. the cost associated with the orderly winding up of credit unions in financial difficulty; and
3. financial assistance to vulnerable credit unions.^[1]

2. Premiums and funding

The DIRF is funded through deposit insurance premiums that are calculated in accordance with the formula set out in the CUCPA Regulations^[2] and the Differential Premium Score Determination document^[3]. It is important to note that DIRF premiums are calculated and invoiced separately from FSRA annual assessments. FSRA’s annual assessments are calculated under the FSRA Fee Rule^[4] and used to cover FSRA’s regulatory costs of the credit union sector.

As required in the FSRA Act, FSRA must file an annual report to the Minister of Finance on the adequacy of the DIRF. The report includes results of assessment work to determine the adequacy of the DIRF and makes recommendations on the target size of the DIRF and any changes to the deposit insurance premium calculation methodology.

As part of the legislative amendments brought in place during the merger of the Deposit Insurance Corporation of Ontario (DICO) and FSRA, the governance of the DIRF now includes the DIRF Advisory Committee of the FSRA Board of Directors. DIRF investments are managed by the Ontario Financing Authority (OFA) under a formal investment agreement and in accordance with a board approved investment policy.

The current target size for the DIRF of 100 bps of insured deposits was introduced in 2014. As of March 31, 2021, the DIRF was $365 million (or 80 bps of insured deposits).  As shown in Table 1, the DIRF is anticipated to reach 100 bps by approximately 2025.  This table is predicated upon the assumption that there will be no losses to the DIRF to meet the 2025 target.

Table 1: Growth of the Deposit Insurance Reserve Fund

Image

Appendix 1 provides a comparison of the DIRF to current and target sizes of the reserve funds maintained by other provinces.

Back to top

One of FSRA’s key priorities for the sector this fiscal year as outlined in the FSRA’s Annual Business Plan (“ABP”) is to enhance the Deposit Insurance Reserve Fund Adequacy Framework (“DIRF Framework”). This priority will extend over several years and has three primary parts:

1. DIRF Stress Testing Model: Continue to develop, refine and mature the model used to assess DIRF adequacy based on sector data.
2. Sector Engagement: Creation of a standing Technical Advisory Committee on Data Strategy and Digital Transformation (“TAC”) and sector engagement on structure and data requirements.
3. Report to the Minister: Report on the adequacy of the DIRF as required by legislation based on results of the model and sector engagement.

FSRA’s inaugural TAC meeting provided valuable feedback which FSRA has considered in its approach.

Future early intervention tools (and the implementation of the Risk-Based Supervisory Framework), combined with strong liquidity in the sector and more risk data should reduce risks of future failures and inform DIRF stress testing and outcomes.

Back to top

1. DIRF adequacy assessment prior to 2020

Prior to 2020, the assessment of the DIRF was completed by an external consultant using an actuarial model. This specific model is one of various options available to assess the adequacy of the DIRF.^[5] The consultant was provided with sector information to periodically update the parameters in the model and would complete several iterations under various economic scenarios to determine the adequacy of the DIRF.  The DIRF was deemed to be adequate if the value of the DIRF did not become negative at any point over a 20-year assessment horizon. 

However, the model used actual historic losses to the DIRF as representing the potential for future DIRF losses. The credit union sector (e.g., the size of credit unions, competitive markets, business loan growth and other new and potentially riskier businesses) of today and tomorrow, and the way it is being regulated, is not reflected in historical losses. Better tools exist to understand the risk profiles of different credit unions today, such as probability of default (PD) and expected losses (EL) should they default, so merely looking at historical realized losses will not well inform us on DIRF adequacy.

2. 2020 DIRF adequacy assessment

As part of FSRA’s required review of the adequacy of the DIRF, a decision was made to open the process and allow consulting firms with the requisite expertise to propose leading methodology or innovative approaches to complete the assessment of the DIRF. 

In early 2020, FSRA issued a Request for Submission (RFS) for external consultants to propose an enhanced risk-based stress testing framework to reassess the adequacy of the DIRF. Deloitte was the successful candidate in the RFS process based on their experience developing such models in accordance with international standards. The proposed framework was based on a mix of statistical methodology and expert judgment to assess the adequacy of the DIRF under various scenarios. An overview of the framework is detailed in Appendix 3.

The methodology associated with this proposed framework is based on estimating the DIRF’s expected loss as suggested by the International Association of Deposit Insurers (IADI) and widely used by deposit insurers in Canada (including the Canadian Deposit Insurance Corporation [CDIC]) and other international jurisdictions. Under the IADI methodology, the DIRF should be sufficient to cover the expected losses (EL) on insured deposits at the sector level under different macroeconomic stress scenarios.  EL at the sector level is calculated as the product of probability of default (PD), loss given default (LGD) and exposure at default (EAD).

The framework includes:

• Identifying the vulnerabilities of the credit unions such as credit default loss, capital loss and liquidity crisis including risk drivers.
• Constructing a range of forward-looking stress scenarios that adequately capture potential future market conditions including extreme but plausible ones.
• Modeling the stress testing approaches and using these scenarios to measure defaults and capital losses.
• Aggregating the risk from all credit unions and evaluating the adequacy of the DIRF under the designed stress scenarios

The COVID-19 pandemic and assumed economic disruptions were used as a basis for macroeconomic forecasts and stress testing. Three scenarios were developed (see Appendices 4, 5, and 6 for details):

1. “Baseline”: a swift recovery from the COVID recession (V-shaped)
2. “Adverse”: a protracted recovery from COVID recession (U-shaped)
3. “Severe”: a COVID recession followed by a housing crash (W-shaped)

Key macroeconomic drivers were identified and projected over a three-year horizon to 2022 for each of the scenarios. These drivers were then used to forecast each credit union’s risk profile (using measures such as return on assets, capital ratio and delinquency rates) by analyzing its major business segments (residential mortgages, commercial and personal loans).

Deloitte used available credit union data from the last ten years (as FSRA provided a data extract from information received in the monthly and annually filings) and supplemented this information with industry proxies and expert judgment. They then determined the appropriate PD, LGD and EAD metrics to calculate the EL amounts for each credit union under each stress scenario.

PD

PD was determined by creating a risk rating for each credit union under each economic scenario and mapping the risk rating to the Moody’s scale to obtain the equivalent PD value. Risk ratings are based on:

• Capital Adequacy (Regulatory Capital/Risk Weighted Assets)
• Profitability (Net Income/Total Assets)
• Asset Quality (Percentage of Loans Greater Than 90 Days Delinquent)
• Efficiency Ratio (Expenses/Revenue)
• Business Diversification (Residential Mortgages, Commercial and other Loans)

LGD and EAD

LGD values were determined through expert judgment and based on industry proxies and averages for potentially analogous assets rather than on the risk inherent in a credit union’s portfolio based on its actual loan book and what the realizable value is expected to be in a particular scenario. This is due to lack of representative historical data on credit union defaults, failures and recoveries. LGD considers two stages: the shortfall required to pay insured depositors as required immediately after insolvency of a credit union (immediate impact to the DIRF) and the long-term loss after a credit union’s assets are liquidated (“net losses” of the DIRF after all recoveries). EAD is the projected amount of insured deposits at each credit union.

EL

EL was calculated for each credit union and the amounts aggregated at the sector level to determine the immediate shortfalls and long-term losses.  Long-term losses are considered the appropriate metric to assess the adequacy of the DIRF. Immediate shortfalls are expected to be funded by the DIRF and if required by the line of credit provided by the OFA. Each would be replenished over an estimated two-five-year period as assets from an insolvent credit union are liquidated.

Based on this analysis, the DIRF is: 

1. Greater than the expected long-term losses in the Baseline scenario of $223 million or 58 bps of insured deposits;
2. Greater than the expected long-term losses in the Adverse scenario of $277 million or 72 bps of insured deposits; and
3. Less than the expected long-term losses in the Severe scenario of $432 million or 112 bps of insured deposits.

However, this analysis assumes many credit unions contribute to draws on the DIRF based on combined ELs. Because draws on the DIRF should only happen when a credit union fails to recover and needs to be resolved, two additional discretionary analyses were completed:

1. Instead of assuming a probability of default for all credit unions, the focus of the analysis was on those credit unions that are more likely to be in an insolvency situation (i.e., insolvency is defined as a credit union’s leverage ratio falls below 4% or the BIS capital ratio falls below 8%). The adequacy of the DIRF is determined based on long-term situations.
2. Determining the credit union’s specific capital ratio forecasts from the proposed framework and assuming the necessity of capital injections whenever the credit union’s leverage ratio falls below 4% or the BIS capital ratio falls below 8%. The size of the DIRF is assessed as the required injection (i.e. capital gap) to increase the capital and leverage ratios to pre-determined levels (above the regulatory minimums) for the credit unions at risk.

Based on this analysis and the same scenarios, the DIRF is: 

1. Greater than the expected long-term losses in the Baseline scenario of $218 million or 57 bps of insured deposits;
2. Greater than the expected long-term losses in the Adverse scenario of $215 million or 56 bps of insured deposits; and
3. Less than the expected long-term losses in the Severe scenario of $708 million or 184 bps of insured deposits.

As part of the scope of the 2020 DIRF assessment mandate, Deloitte was asked to provide recommendations on how to enhance the framework and the assessment results. FSRA’s view is that until we have a more objective and robust DIRF adequacy assessment framework and tools, scenarios are of lesser importance as it is assumptions and expert judgment that will drive results.

The key recommendation was the collection and incorporation of additional data elements to replace industry proxies and reduce dependency on expert judgment. This facilitates a framework that will use actual data reflecting the risk profiles in credit unions and changes over time in those profiles. The framework will be useful in more advanced scenario testing including running multiple stochastic model scenarios to get a better understanding of the range of potential DIRF draws on a probabilistic basis. Enhanced data collection also supports access by FSRA to adequate funding beyond the DIRF so that losses will not be increased due to the need to fire-sale assets at a loss.

3. 2021 DIRF adequacy assessment

As conveyed at the March 4, 2021 Credit Union Sector Town Hall, one of FSRA’s key priorities in the proposed 2021-2022 Annual Business Plan for the credit union sector is to enhance the DIRF Framework.

To achieve this priority, FSRA has thus far:

• Re-engaged Deloitte to assess DIRF adequacy based on updated model parameters and scenarios;
• Begun to build out internal capabilities and knowledge transfer to develop in-house capacity;
• Set up and met with a Technical Advisory Committee (TAC) comprised of credit union representatives on May 7, 2021 to get their views on the approach (feedback on the model was incorporated into this consultation paper);^[6] and
• Begun work internally to review collected data from the sector to identify data needs.

Further, the development of stress testing scenarios is currently underway with the determination of effects of the scenarios on the key macroeconomic variables to follow.

The scenarios proposed are:

1. Base Case: The economic recovery is in full swing. The economic growth is projected to gain momentum in the second half of 2021. Household spending and government support will remain key to growth, but high debt levels will meet their long-term contribution. Areas of the economy that continue to struggle are business investment, trade and hospitality.
2. Adverse Case: Examines the impact of a sustained, 30% decline in Canadian resale home prices beginning in the first quarter of 2022.
3. Severe Case: Examines the impact of renewed public health measures. These last through 2022. The initial impact is focused on industries subject to the lockdown. However, the large increase in unemployment also takes a toll on other industries. Bankruptcies begin to rise as governments are forced to cut support. The steep decline in GDP places significant stress on some provincial governments already in a precarious fiscal situation. Substantial credit downgrades occur while the federal government is required to provide support to some governments.

Once scenarios are completed, they will be shared with the TAC together with any feedback received through this process. As this is a multi-year dialogue, any comments and inputs will be taken into account to be incorporated as the model evolves.

Back to top

To help inform our work, FSRA is seeking sector input into the approach. FSRA would appreciate your feedback by September 9, 2021.

1. Please provide feedback on the overall approach to the DIRF Adequacy Review Framework as articulated in this consultation paper.
2. Do you have specific comments about the methodology used?
3. Please provide feedback on the recommendation to obtain and incorporate additional “risk data” (refer Appendices 7 to 9 for details) to enhance the model and refine the DIRF.

All comments received and responses provided will be posted to the consultation web page. If you have any additional questions or need clarification please contact Alena Thouin and Bradley Hodgins at [email protected] and [email protected]

Back to top

Appendix 1 - Comparison of provincial reserve funds

The 100-bps target in Ontario is at the low end of the range when compared with other Provincial insurers:  

Back to top

Appendix 2: Options for the DIRF adequacy assessment model

The adequacy of the DIRF can be assessed using different frameworks. The following table provides an overview of the characteristics, advantages and limitations of three types of frameworks.

Back to top

Appendix 3: Overview of the stress testing framework

Image

Back to top

Appendix 4: Stress testing scenarios – 2020 base case

Back to top

Appendix 5: Stress testing scenarios – 2020 adverse stress scenario

Back to top

Appendix 6: Stress testing scenarios – 2020 severe stress scenario

Back to top

Appendix 7: 2021 Proposed approach – Option 1: all required data is available

Assumptions

• Loan Portfolio Data is available at account, or at least segment level for at least one economic full cycle. Credit data should be available at monthly frequency, and the loan attributes should include loan type, origination date and loan age, loan interest rate, borrower risk rating or credit scores, utilization, balance, loan-to-value etc. All historic loan defaults, default exposures, default losses, credit provisions should also be available.
• Deposit Portfolio Data is available at account, or at least segment level for at least one economic full cycle.  Deposit data should be available at monthly frequency. Deposit data attributes should include deposit type, origination date, maturity, balance, interest rate, origination and run-off etc. All historic deposit data should include run-offs.
• Investment and Liquidity Management Portfolio Data is available at instrument, or at least product level for at least one economic full cycle. The investment portfolio data should be available at monthly frequency. The data should include instrument type, maturity, notional, market data driving the value change, fair-value, book value.
• All the above data is available for all credit unions regulated by FSRA.
• Macroeconomic economic factors affecting each of credit unions credit, capital or liquidity risk, can be obtained from FSRA or from public sources.

Proposed Approach

• Collect data, conduct data assessment and process the data for the purpose of DIRF adequacy stress testing by leveraging our well tested tools and accelerators.
• By leveraging our previous experience, industry insight and expertise in PPNR, CCAR, ICAAP, credit, capital and liquidity stress testing modeling practices, as well as our well tested tools and accelerators, to review loan portfolio data, deposit data and investment portfolio data from each credit union to understand the risk profile and identify the key risk drivers affecting the loan credit loss, deposit run-off and investment loss, and identify the leading macroeconomic risk drivers.
• Based on the outcomes of the above, leverage the tools and accelerators to identify the choice of the most fitted forecasting and stress testing models, and calibrate them based on historic data, and apply designed stress scenarios to estimate the loan portfolio losses, deposits run-offs, and investment and liquidity management portfolio losses for each credit union in scope. Various candidate and component models can be developed including capital and liquidity forecast and stress testing models from cost/income valuation approaches, PD, LGD and EAD parameter stress testing models, credit loss forecast and stress testing models, PPNR forecast and stress testing models.
• Under each of the designed stress scenarios, estimate the potential demand for the DIRF for each of the credit unions, and aggregate them to estimate the total demand of DIRF. This will inform the analysis and conclusion of the DIRF adequacy under each stress scenario.

Back to top

Appendix 8: 2021 Proposed approach – Option 2: partial data is available

Assumptions

1. Loan Portfolio Data is available at segment or portfolio level for at least one economic full cycle. Credit loss data should be available at least quarterly, and the loan portfolio attributes should include loan type, balances and weighted interest rates, credit provisions and segment or portfolio level default rates etc.
2. Deposit Portfolio Data is available at segment or portfolio level for at least one economic full cycle.  Deposit data should be available at least quarterly. Deposit portfolio data attributes should include deposit type, weighted interest rates, run-offs etc.
3. Investment and Liquidity Management Portfolio Data is available at product or at least portfolio level for at least one economic full cycle. The investment portfolio data should be available at least quarterly. The portfolio level data should include instrument type, market data driving the value change, fair-value, book value.
4. All the above data is available for all credit unions regulated by FSRA.
5. Macroeconomic economic factors affecting each of credit unions credit, capital or liquidity risk, can be obtained from FSRA or from public sources.

Proposed Approach

• Collect data, conduct data assessment and process the data for the purpose of DIRF adequacy stress testing by leveraging our well tested tools and accelerators.
• By leveraging our previous experience, industry insight and expertise in PPNR, CCAR, ICAAP, credit, capital and liquidity stress testing modeling practices, as well as our well tested tools and accelerators, to review loan portfolio data, deposit data and investment portfolio data from each credit union to understand the risk profile and identify the key risk drivers affecting the loan credit loss, deposit run-off and investment loss, and identify the leading macroeconomic risk drivers.
• Based on the outcomes of the above, leverage the tools and accelerators to identify the choice of the most fitted forecasting and stress testing models at segment or portfolio level, and calibrate them based on historic data, and apply designed stress scenarios to estimate the loan portfolio losses, deposits run-offs, and investment and liquidity management portfolio losses for each credit union in scope. The framework and component models to be developed would be limited to segment or portfolio level earning and loss forecasts models and PD, LGD and EAD parameter stress testing models with significant expert judgements.
• Under each of the designed stress scenarios, estimate the potential demand for the DIRF for each of the credit unions, and aggregate them to estimate the total demand of DIRF. This will inform the analysis and conclusion of the DIRF adequacy under each stress scenario.

Back to top

Appendix 9: 2021 Proposed approach – Option 3: very limited data is available

Assumptions

• Loan Portfolio Data: There is no loan portfolio segment level data, and also limited portfolio level data can be available, such as portfolio types, balances, credit provisions, write-offs, default rates etc. The data might not cover one economic full cycle, and not all credit unions may have consistent data.
• Deposit Portfolio Data: There is no segment level deposit portfolio data, and also limited portfolio level data can be available such as portfolio level types, balances, run-offs, weighted interest rate etc. The portfolio level data might not cover one economic full cycle, and not all credit unions may have consistent data.
• Investment and Liquidity Management Portfolio Data: There is portfolio level data, and limited portfolio level data can be available such as portfolio type, balance, fair value of book value, market risk affecting the portfolio value changes. The data might not cover one economic full cycle.
• All the above data is available for all credit unions regulated by FSRA.
• Macroeconomic economic factors affecting each of credit unions credit, capital or liquidity risk, can be obtained from FSRA or from public sources.

Proposed Approach

• Collect data, conduct data assessment and process the data for the purpose of DIRF adequacy stress testing by leveraging our well tested tools and accelerators.
• By leveraging our previous experience, industry insight and expertise in PPNR, CCAR, ICAAP, credit, capital and liquidity stress testing modeling practices, as well as well tested tools and accelerators, to review loan portfolio data, deposit data and investment portfolio data from each credit union to understand the risk profile and identify the key risk drivers affecting the loan credit loss, deposit run-off and investment loss, and identify the leading macroeconomic risk drivers.
• Proxies and expert judgments would mostly be applied when there is missing data or data error across all the credit unions in scope. Some external data might be used.
• Based on the outcomes of the above, leverage the tools and accelerators to identify the choice of the most fitted forecasting and stress testing approaches that could be quantitative or qualitative, and estimate the sensitivities against the identified macroeconomic risk drivers based on limited historic data and other industry data, and apply designed stress scenarios to estimate the loan portfolio losses, deposits run-offs, and investment and liquidity management portfolio losses for each credit union in scope. The models to be developed would be top-down approaches at portfolio level with qualitative assumptions and expert judgments.
• Under each of the designed stress scenarios, estimate the potential demand for the DIRF for each of the credit unions, and aggregate them to estimate the total demand of DIRF. This will inform the analysis and conclusion of the DIRF adequacy under each stress scenario.

Back to top

[1] Section 3(4), Credit Unions and Caisses Populaires Act, 1994, S.O. 1994, c. 11.
[2] Ibid
[3] Differential Premium Score Determination
[4] FSRA Fee Rule
[5] See options analysis in Appendix 2
[6] The TAC has a multi-year mandate to provide advice to FSRA on various topics including the DIRF Adequacy Framework;