Quantitative Impact Study #5 – Commonly Asked Questions - Version 4

We have received a number of questions regarding Quantitative Impact Study #5 (QIS#5).  We are providing answers to a number of the most commonly asked questions below. The questions are organized by risk.

Corrections or clarifications will be made if necessary [in brackets] to original questions and answers from earlier versions of this document. The dates of the changes or additions are indicated in the brackets. New questions and answers will be appended below under the version posted, if needed.

General Instructions

Version 2

G.1) QIS#5 introduces the condition that products must meet the criteria in MCCSR section 1.2.6 to qualify for the par credit. As MCCSR section 1.2.6 applies on a risk by risk basis, clarification is needed for the application to QIS#5.

Insurers are requested to apply the criteria in MCCSR section 1.2.6 except the criteria that requires the policies pay meaningful dividends has been eliminated for QIS#5 purposes. In the instructions, insurers should treat all comments related to par as qualifying par according to MCCSR section 1.2.6. Non-qualifying par surplus can be excluded from the calculation of the par credit floor. This issue will be reviewed further for QIS#6.

Insurance Risk

Version 1

I.1) In the insurance risk cash flow file, there is a “morbidity termination – volatility” tab but no instructions how to fill it. Are there instructions missing or should the tab not have existed?

The “morbidity termination – volatility” tab was meant to calculate a solvency buffer for disability termination volatility risk to be included in the exposure for the disability level risk SFF calculation. A shock for disability termination volatility risk was not specified in the instructions and the exposure will therefore be removed from the calculation of the disability level risk SFF. The tab will be removed and a new form will be posted that uses the disability SFF based on the other active lives volatility buffer for the disability termination level risk SFF. The methodology for disability level risk SFF will be reviewed further for the next QIS.

Version 2

I.2) In QIS#5, the treatment of tax timing differences was clarified in the instructions. This is still unclear and why is this adjustment necessary?

If it is not practical or possible to calculate the tax timing differences on a QIS basis, use the CALM cash flows and include a comment in the “Questions and Comments” worksheet.

I.3) For the longevity trend risk test for business outside of Canada, we intend to use CIA base improvement rate. Please confirm that was the intention.

Yes, that was the intention for the test.

Version 3

I.4) In the French version of the insurance risk instructions in the last paragraph of page 13, is there an error in the description of the mortality volatility risk?

Yes, in this paragraph, the wording should be "(fondés sur la survie et les décès)" rather than "(fondés sur la survie)".

I.5) Please confirm the volatility cash flows for the lapse risk on page 22 of the Insurance Risk Instructions - I'm not sure whether I should be applying a shock of +/-30% or +/-50% in the first year in determining the cash flows to input to the lapse volatility cash flow worksheets. Based on our interpretation last year for QIS 4 we only applied +/-30% for the volatility cash flows but it looks like you want us to apply +/50% in determining the cash flows for volatility.

Please confirm (or advise otherwise) that:

• the cash flows in the worksheet "Lapse-Sensitive-Level & Trend" are to be the cash flows calculated using lapse rates of 120% of best estimate in all years (if we are not using crossover logic), and
• the cash flows in worksheet "Lapse-Sensitive-Volatility" are to be the cash flows calculated using lapse rates of 150% of best estimate in year 1, 100% of best estimate thereafter.

Yes, for level and trend risks.

No, the volatility risk buffer itself is based only on 130%. The additional 20% is for the level risk that can be directly calculated in the level risk calculation

I.6) The Instructions on mortality catastrophe risk says this: “The shock for catastrophe risk varies by location of the policyholders at issue. The shock is an absolute increase in the number of deaths per thousand insured over the following year for:”. Is the “following year” meaning the first projection year or the year after the first projection year?

This means the first projection year as all the volatility and catastrophe shocks are applied to the first year cash flows.

I.7) Industry requests clarification on flooring otherwise negative solvency buffers and how these zero floors should be reflected in the template. It is unclear whether floors should be at the product level or the total geography level.  From OSFI's documents:

• Insurance risk instructions, page 6: "Any negative solvency buffers should be floored to zero at the geography level and disclosed in the "Questions and Comments" worksheet."
• Lapse volatility risk instructions, page 22: "The solvency buffer should be floored at zero at the portfolio level if necessary."
• Lapse catastrophe risk instructions, page 23: "The solvency buffer should be floored at zero at the portfolio level if necessary."
• Template: The catastrophe risk formulas include logic to floor the buffer at zero (for example, tab "Lapse Risk", cells AQ22, AR22, AQ24, AR24, AY24, BE24, BK24, BQ24, BW24, CC24). The formulas do not floor the volatility component at zero.

 Industry has identified situation(s) of business units having negative volatility and catastrophe risk buffers for lapse supported business. When cash flows are combined with those for other business units in the geography, the total geography solvency buffers are positive. Should we allow the cash flows for that business unit to flow through to the total geography level (so that the business unit would contribute a negative impact to the total geography solvency buffer)? Or should we set the shocked cash flows for that business unit equal to the best estimate cash flows, so that it does not contribute to the total geography solvency buffer?

In addition, should the volatility risk solvency buffer formulas be modified so that they apply a zero floor (similar to the catastrophe risk solvency buffer formulas?)

[Updated December 20, 2013:

Lapse risk should be tested at the portfolio level to determine if lapse supported or lapse sensitive. Each of level, trend, volatility and catastrophe should be calculated at the portfolio level. Insurers are requested to change the formulas to apply a zero floor for lapse volatility risk similar to catastrophe risk since this is needed before volatility and catastrophe are aggregated. The changes are as follows:

In the lapse risk tab of the insurance risk form, change the following cells for Canada:

Cell	Formula
AQ17	=MAX(AQ16-AQ15,0)
AR17	=MAX(AR16-AR15,0)
CE17	=MAX(CE16-CE15,0)
CF17	=MAX(CF16-CF15,0)
DS17	=MAX(DS16-DS15,0)
DT17	=MAX(DT16-DT15,0)
FG17	=MAX(FG16-FG15,0)
FH17	=MAX(FH16-FH15,0)
GU17	=MAX(GU16-GU15,0)
GV17	=MAX(GV16-GV15,0)
II17	=MAX(II16-II15,0)
IJ17	=MAX(IJ16-IJ15,0)
AQ24	=SQRT(MAX(SUM(AU17:AY17)+SUM(BS17:BW17)+SUM(BG17:BK17),0)^2+MAX(SUM(AU22:AY22)+SUM(BS22:BW22)+SUM(BG22:BK22),0)^2)
AR24	=SQRT(MAX(SUM(BA17:BE17)+SUM(BY17:CC17)+SUM(BM17:BQ17),0)^2+MAX(SUM(BA22:BE22)+SUM(BY22:CC22)+SUM(BM22:BQ22),0)^2)
AY24	=SQRT(MAX(SUM(AU17:AY17),0)^2+MAX(SUM(AU22:AY22),0)^2)
BE24	=SQRT(MAX(SUM(BA17:BE17),0)^2+MAX(SUM(BA22:BE22),0)^2)
BK24	=SQRT(MAX(SUM(BG17:BK17),0)^2+MAX(SUM(BG22:BK22),0)^2)
BQ24	=SQRT(MAX(SUM(BM17:BQ17),0)^2+MAX(SUM(BM22:BQ22),0)^2))
BW24	=SQRT(MAX(SUM(BS17:BW17),0)^2+MAX(SUM(BS22:BW22),0)^2)
CC24	=SQRT(MAX(SUM(BY17:CC17),0)^2+MAX(SUM(BY22:CC22),0)^2)

Proceed with similar changes for other jurisdictions.]

I.8) If a lapse-supported policy has a cross-over in the 3rd year after the valuation date and the test confirms that the buffer is higher with lower lapses. But, the catastrophe and/or volatility buffer is negative with lower first year lapse. Which of the following approaches is correct?

1. Use the cross-over logic and use a higher lapse – it gives a positive buffer for all policies in the portfolio.
2. Apply the zero floor to the policy – it gives a positive or zero buffer for all policies in the portfolio.
3. Leave the negative buffer for the policy – if it results in a negative buffer for the portfolio, apply zero floor to the portfolio

The approach intended in the instructions was #3, but insurers could use #1 as an approximation. Negative numbers for sub-risks (i.e. volatility, catastrophe or level and trend) of individual policies can be used as long as the solvency buffer for each portfolio is positive.

Market Risk

Version 2

M.1) QIS#5 clarified the treatment of universal life cash flows to be consistent with CALM but using QIS forward rates as investment return assumptions. This may only be appropriate in circumstances where the credited rate is directly linked to an external bond yield. It would be more appropriate to keep existing CALM cash flows where credited rates are not as sensitive to the current yield curve.

Insurers may vary the credited rate appropriately with the scenarios. If the credited rate and product cash flows depend on the bond market rates, use the QIS forward rates as investment return assumptions. If they do not, use CALM credited rates. This also applies to the universal life cash flows in the insurance risk instructions.

M.2) Why is the solvency buffer cash flow test still included? There are significant practical difficulties to performing this test. Also, there will be lack of consistency among companies since there is no guidance in the instructions on how to reflect the credit for par/adjustable and diversification.

This test is voluntary.

M.3) The market risk QIS instructions for index linked risk pass through (RPT) product are not clear about the application of the interest rate risk component for these products. The instructions near the top of page 2 say that “Index linked RPT products and supporting investments should be included in the interest rate risk solvency buffer regardless of the treatment of the index linked RPT product risk component.” However, those at the bottom of page 1 imply that one would not have to do any other component of the market risk QIS (other than the RPT product risk component), including the interest rate risk component. Could you please clarify?

The instructions to include the index linked RPT products and supporting investments in the interest rate risk solvency buffer aim to have a consistent present value of best estimate cash flows between insurance risk and market risk components. However, the following instructions on page 6 under “Pooled Funds – Index Linked RPT Products” apply to all Index Linked RPT products:

“If the index linked RPT product risk component is used, the liability cash flows should be equal to the asset cash flows in each scenario, except amounts for minimum interest guarantees should be used if higher. If the index linked RPT product risk component is not used, the liability cash flows should be related to the asset cash flows in each scenario according to the same rules as used for the CALM valuation. If minimum interest guarantees do not apply, time zero cash flows may be used”.

M.4) Please clarify the meaning of the following clause from the same section of the market risk QIS: “If the index linked RPT product risk components are not used …”.  This clause is also used in the Credit Risk QIS.  Does this mean that the product is reported on ref  rows 010, 011, or 012 of Page 35.010 of the current MCCSR Form?

This refers to the following instructions near the top of page 17:

“For products with a 30% factor including products with new asset funds, insurers could alternatively use for those products (and their matched assets) the requirements for credit risk and other categories of market risk, including the interest rate risk and equity risk”.

Version 3

M.5) The Market risk instructions are inconsistent on page 2 (use same as MCCSR) and page 15 (use same as MCCSR, but with +/-30% instead of +/-15%). Please clarify which is the correct basis.

Please use the same approach and methodology as the current MCCSR but with +/-30% instead of +/-15%.

Version 4

M.6) We would like some clarification on the new wording added in QIS5 under Equity Risk – substantial equity investments.

Page 15 of the instructions says “Substantial equity investments, including in a joint venture but where the company does not have control, should be calculated using a time zero deterministic shock of 40%.”

Please confirm that this shock applies only to Joint Venture corporations which were previously deducted from available capital under MCCSR Guideline 2.10 and that it does not apply to limited partnerships under Guideline 3.1.12 which are not deducted from Available capital but might still be joint ventures and are handled under a ‘look through’ basis in 3.1.12.

Insurers are to apply the 40% shock to substantial equity investments without control, currently deducted from available capital per MCCSR Section 2.10. The look through method continues to apply for investments subject to Section 3.1.12.

M.7) We have issued non-cumulative perpetual preferred shares and subordinated debt. For the interest rate risk buffer, should we assume a redemption date for the preferred shares and if so how, and should we project subordinated debt renewals?

As the preferred share dividends are not guaranteed, insurers are requested to use their balance sheet value at time zero. Renewals should not be projected for subordinated debt. Insurers should assume reimbursement at the end of the interest rate guarantee period.

Credit Risk

Version 1

C.1) The instructions and forms are inconsistent regarding the treatment of unrated ABS. Pages 1 and 4 of the instructions state that the risk factor for unrated ABS was removed and is being reviewed in available capital. Pages 8-9 of the instructions and the credit risk template still show the 100% factor for ABS rated B and lower. Which approach is correct?

Please follow the forms. There was a last minute change that was not updated in the instructions.

C.2) Could you please clarify the treatment of negative reinsurance assets in the calculation of reinsurance counterparty risk?

Negative reinsurance assets should be floored at zero before the factor is applied to the reinsurance asset.

Version 2

C.3)Why was the methodology for the bond credit risk factors changed from QIS#4 with risk factors reflecting a mid-point interpolation of the risk factors? The complexity of the interpolation does not justify the perceived precision.

The QIS#4 methodology did not assume midpoint interpolation, the factor for the duration range should have used that factor for all bonds in that range. For QIS#5, insurers may assume midpoint interpolation as an approximation, however, once the final standard is issued, insurers will be expected to use calculations as described in the QIS#5 approach.

Version 3

C.4) Regarding the reporting of mortgage-backed securities within the Credit Risk file, should we treat each pay-down as a maturity, thus requiring us to split out the maturity over the life of the asset?  Or, should we report the entire current value according to the stated maturity date (e.g. if matures in 2035, then report entire amount under the maturity column 10 years or more)?

The instructions require that asset backed securities be grouped according to the maturity and rating of the issue, on a similar basis as previously set out for public and private bonds. In the public bonds section, maturity is defined as effective maturity which is itself defined in Appendix II. The effective maturity would be lower than 22 years. In fact, depending on the principal distributions/pay/downs, the effective maturity could be less than 10 years. So, the total amount should be reported in the appropriate column.

C.5) What is the difference between the requirements for the first section titled “Asset Backed Securities by Rating and Maturity” and the section below it titled “Average maturity”?  Should the stated maturity date be used for the top portion and the effective maturity date be used for the second section?  I also noticed that there’s no row for “Guaranteed by CMHC or another entity eligible for a 0% factor” in the “Average maturity” section, why is that?

In the first section, you input the dollar value of the securities corresponding to the rating (row) and maturity (column); in the second section, you input the aggregate average maturity of all the securities included in the corresponding cell from the first section. Use effective maturity for both sections (average of effective maturities in the second section) unless a 0% factor is applicable then average maturity is not necessary to calculate the solvency buffer

Forms

General Form

Version 1

GF.1)  The formulas in the insurance risk template and the summary template refer to incorrect file names. As an example, in the summary template, on tab "Summary", the formulas refer to workbook "QIS#5_mr_e.xlsm" but the market risk template is actually named "qis5_mr_e.xlsm".

Links between workbooks will have to be refreshed to link to the files as saved on your company’s network. This can be done in the data – edit links menu in Excel by using the change source function for each source workbook. The summary and insurance risk workbooks have external links that need to be refreshed. It is suggested to have the source workbook open as well when refreshing the links as it will significantly shorten the processing time.

GF.2)  We've noticed an inconsistency between the QIS#5 instructions for operational risk and the summary template. The instructions state that the 5% risk charge should be applied to "the solvency buffer net of credit for participating and adjustable products and before credit for diversification between risks." However, the summary template applies the risk charge to the solvency buffer before diversification credit and before credit for participating and adjustable products in tab "Ops Risk Data". Which is correct?

The instructions are correct and a new form will be posted with the correction to link the charge on the solvency buffer to the solvency buffer after credit for participating and adjustable products but before credit for diversification between risks (line 36 – (line 43 + line 44)). Cells that were corrected are F46 to K46.

GF.3)  There is an issue in the Op Risk tab of the main Summary template. There is a row missing for “Annuity liabilities – accumulation annuities” under the Business Volume Risk Charge and the Large Increase in Business Volume Risk Charge sections.

[Updated December 20, 2013:

Accumulation annuities should be included in the existing line “Mutual Funds, GICs and Other Deposit Products and Annuities”.]

Version 4

GF.4) In the TAR comparison tab, does line 33 for seg fund CALM liabilities include PfADs? Also, what if there are seg fund PfADs not in the categories listed in lines 34-43?

Yes, line 33 does include PfADs. Insurers are requested to include any other PfAD not included in lines 34-43 in the most appropriate category listed. The by risk PfADs are used for the analysis by risk component.

GF.5) For unregistered reinsurance, should I calculate all solvency buffers on a net basis, even though there isn’t sufficient collateral in the trust to cover the ceded solvency buffers (i.e. “Excess assets in trust under QIS” is less than 0)?

Yes, we assume that the company would request more assets in trust if necessary under the new framework.

GF.6) On the “Unregistered reins” tab of the summary file, can you please confirm that the “Assets in trust according to QIS solvency buffer (at 100%) before diversification credit” should be equal to the difference between the gross/net (of unregistered reinsurance) solvency buffers for market risk, credit risk and insurance risk only?

Yes, that is correct.

GF.7) For operational risk, should we enter deposit amounts (e.g. dividend on deposit, future premiums (premiums received in advance)) in the line:

Mutual Funds, GICs and Other Deposit Products and Annuities

Or should we not enter these amounts in operational risk?

No, operational risk for these deposits is covered by the risk of the underlying products. For example, the risk for amounts on deposit is covered by the operational risk for the insurance contracts to which these deposits are linked.

Insurance Risk Form

Version 1

IF.1) In the “Summary” tab, cells O39, S39, Q39, AA39, AE39, O40, S40, Q40, AA40, and AE40 are all linked to cell T41 instead of X41 in the respective Morbidity Risk tabs. All geographies other than Canada are linked to the wrong cell.

Summary tab morbidity termination buffer links (other than Canada) have been fixed and a new form will be posted with the correction.

IF.2) In the “Summary” tab, cells E47 to I47 are all linked to the incorrect cells in the Lapse Risk tab (i.e. cell E47 should be linked to 'Lapse Risk '!CF12 instead of 'Lapse Risk '!AS12).

Summary tab lapse sensitive level & trend buffer links (other than Canada) have been fixed and a new form will be posted with the correction.

IF.3) In the Summary tab, row 50, the expense solvency buffers for adjustable products under each geography are only picking up the solvency buffers for adjustable products - life from the Expense Risk tab. The Summary tab should be revised to pick up the sum of solvency buffers for adjustable life, annuity and A&S products for all geographies

Summary tab expense solvency buffer links have been fixed and a new form will be posted with the correction.

Version 4

IF.4) On the mortality risk tab of the insurance risk template, there is a place to enter “Less: solvency buffer related to shocked best estimate mortality rates for 1st year only - life supported” (e.g. in cell U20 for non-par individual life products). I didn’t see any mention of this adjustment in the insurance risk instructions. Can you please confirm that this cell should be filled in with the solvency buffer related to the shocked best estimate mortality rates for the first year only – where the shock is based on the minimum of the (a) and (b) level risk mortality shocks?

Yes, that is correct.

Insurance Risk Cash Flows Form

Version 1

ICF.1) In the “Mortality - Volatility Shock” tab, shouldn’t all “Total” cells (i.e. D10, G10, H10) be taking the square root of the sum of squares rather than simple additions?

Sums of mortality volatility risk calculated amounts have been corrected to be a sum of square. This was not being used in the summary but a new form will be posted with the correction.

Market Risk Form

Version 1

MF.1)  There is an error in the equity tab in market risk template. Row 10 is incorrectly summing all the "Closely correlated equity positions" instead of the "Net long substantial positions…", and Row 11 is incorrectly summing the "Scenario table charge for options" instead of the "Net short substantial positions…"

Equity tab calculations have been fixed and a new form will be posted with the correction.

Credit Risk Form

Version 1

CF.1) There is an error in the mortgage tabs in every jurisdiction in the credit risk template. Cells H63 to H73 and P63 to P73 should multiply both the mortgage amount and the redistribution of exposure for collateral guarantees instead of only the redistribution of exposure for collateral guarantees by the factor to calculate the buffer.

Mortgage tabs calculations have been fixed and a new form will be posted with the correction.

Discount Rate Test

Version 1

DRT.1) For the 2011 interest rate test, should we be replacing both the 2012 risk free rates and the 2012 spreads? (Both are supplied in the "Interest Rates" worksheet but p10 of the general instructions just mentions the risk free rates.)

You should use both the 2011 risk free rates and the spreads.

DRT.2) If we should be replacing both the 2012 risk free rates and the 2012 spreads with 2011 rates then I think there is a problem in the interpolated spreads supplied for the 2011 interest rate test - they are referencing the given spreads in 2012 (for 1 year, 5 year, etc.) rather than the given 2011 spreads (for 1 year, 5 year, etc.). e.g. for Canada see spread for 19 years = 0.36% (row 53) vs. 20 years =2.15% (row 54.) Using these spreads that are a mix of 2011 & 2012 spreads produces a rather strange pattern of forward interest rates.

The formulas were done that way so you could cut and paste the 2011 rates and spreads over the 2012 rates and spreads without having to manually change any formulas. The formulas may look wrong because the formulas reference the cells occupied by the 2012 spreads, but when you copy and paste the 2011 rates and spreads over the 2012 ones, the formulas adjust.

DRT.3) It looks like the same approach has been taken in the Market Risk, Summary & Insurance Risk spreadsheets.

Yes, you need to update each form that uses the “Interest Rates” worksheet. You need to update in each of market, summary and insurance risk forms.

DRT.4) Instructions on page 10 of the general directions just mention the "Interest Rate" worksheet. Should we also be changing the 2012 ultimate discount rate to the 2011 rate? And if so, where all do we need to make changes?

Yes, you need to use the 2011 UDR. The UDR is on the “Discount Rate” tab, at the bottom of the worksheet.

Version 3

DRT.5) For purposes of consistency among companies, we recommend OSFI clarify that as part of the tactics to re-run QIS5 using QIS4 discount rates, companies are to override the QIS5 L1/L2 assumption of 85 basis points (40 basis points for Japan) with the QIS4 comparable figure of 75 basis points (35 basis points for Japan).

Insurers are requested to use the 2012 UDR shocks (85 basis points and 40 basis points for Japan). To clarify, the discount rate test uses 2011 risk free rates, 2011 spreads, 2011 UDR and 2012 shocks to the UDR.