Session 7 Defaults And Spreads Final

21
Copyright © 2007 by the American Academy of Actuaries The Year in Review, November 2007 1 NAIC VM-20 Impact Study Training September 22, 2010 1 NAIC Training Seminar for VM-20 Impact Study November 18-19, 2010 Session 7: Default Assumptions and Reinvestment Spreads David E. Neve, FSA, CERA, MAAA Vice President - Capital Management, Aviva USA Gary Falde, FSA, MAAA Vice President & Appointed Actuary, Pacific Life Insurance Co. Copyright © 2007 by the American Academy of Actuaries The Year in Review, November 2007 2 NAIC VM-20 Impact Study Training November 18-19, 2010 2 Overview of Default Cost Methodology for Existing Assets Default Cost Example Overview of Reinvestment Spread Requirements Agenda for This Session Agenda for This Session

Transcript of Session 7 Defaults And Spreads Final

Page 1: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 1NAIC VM-20 Impact Study TrainingSeptember 22, 2010

1

NAIC Training Seminar for VM-20 Impact StudyNovember 18-19, 2010

Session 7: Default Assumptions and Reinvestment Spreads

David E. Neve, FSA, CERA, MAAAVice President - Capital Management, Aviva USA

Gary Falde, FSA, MAAAVice President & Appointed Actuary, Pacific Life Insurance Co.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 2NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

2

Overview of Default Cost Methodology for Existing Assets

Default Cost Example

Overview of Reinvestment Spread Requirements

Agenda for This SessionAgenda for This Session

Page 2: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 3NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

3

Prescribed Default Cost Methodology: Existing Prescribed Default Cost Methodology: Existing AssetsAssets

Overarching objectives provided by LHATF:

Default costs for the same or similar assets should be the same across companies

Companies should not be able to lower reserve by investing in riskier assets beyond some threshold.

In the short term, default costs should reflect current economicconditions and grade into historic conditions over the longer term.

The method should be relatively simple.

Note: these objectives are not completely consistent with each other

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 4NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

4

Assets Included in the Prescribed Methodology

The prescribed methodology only applies to fixed income assets with an NAIC designation. Current examples: corporate bonds, preferred stocks, RMBS and CMBS.

A different prescribed methodology provided by LHATF applies to fixed income assets that lack an NAIC designation. This includes but is not limited to commercial mortgage loans and

residential mortgages (whole loans).

For these assets, the default assumption shall be established such that the net yield shall be capped at 104% of the applicable historical U.S. Treasury yield rate most closely coinciding withthe purchase date and maturity structure, plus 25 basis points.

Page 3: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 5NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

5

A Word on Commercial Mortgage LoansA Word on Commercial Mortgage Loans

The Life Risk Based Capital Working Group has been discussing an ACLI proposal to revamp C-1 RBC for this asset class

Would be a long-term solution to replace the Mortgage Experience Adjustment Factor (MEAF) construct

ACLI presented an updated version of the proposal to Life RBCWG at the August NAIC meeting

Would evaluate each individual performing loan in a company’s portfolio based on a combination of Debt Service Coverage ratio and Loan to Value

Academy Life Capital Adequacy Subcommittee plans to evaluate andcomment

Such an approach may hold promise for future incorporation into the VM-20 asset default structure

The LRWG recommended deferring any further work to develop an alternate approach to commercial mortgage defaults for VM-20 until the Life RBCWG concludes its work on this proposal

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 6NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

6

Asset Default Assumption

A prescribed methodology with parameters set by regulators

Projected default costs are the sum of three components:

Baseline annual default cost factor. Based on historical corporate default and recovery experience. Includes a margin for conservatism. Factor is a table “look-up” based on a “PBR credit rating” from 1-21, and weighted average life of the asset.

Spread related factor. Based on the corporate bond spread environment as of the valuation date. Adjustment can be positive or negative, and grades off over three years (subject to a floor and a cap).

Maximum net spread adjustment factor. Portfolio-wide upward adjustments, graded off over three years, if the net spread of the portfolio exceeds the net spread of a “regulatory threshold” index bond.

Page 4: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 7NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

7

Company-determined Inputs for each asset

The company will need to determine several items for each asset:

1. Investment expense assumption for each asset type

2. Option adjusted spread (OAS) for each asset OAS = average spread over zero coupon Treasury bonds that equates a bond’s

market price as of the valuation date with its modeled cash flows across an arbitrage free set of stochastic interest rate scenarios.

For floating rate bonds, the OAS equals the equivalent spread over Treasuries if the bonds were swapped to a fixed rate.

Market conventions and other approximations are acceptable.

3. Weighted Average Life (WAL) Equals the weighted average number of years until 100% of the outstanding

principal is expected to be repaid (rounding to the nearest whole number, but not less than 1).

For bonds or preferred stocks that are perpetual or mature after 30 years, the WAL shall be 30.

Market conventions and other approximations are acceptable.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 8NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

8

Step One: Determine Baseline Annual Default Cost Factor

The baseline annual default cost factor in all projection years shall be taken from the most current available baseline default cost table published by the NAIC using the PBR credit rating and weighted average life (WAL) of the asset on the valuation date.

The methodology for creating this table can be found in Appendix 2 of VM-20 (will not go into details of how the table was created in this presentation).

Table A of Appendix 2 shall be the initial NAIC table for this purpose.

Page 5: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 9NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

9

Table A. Prescribed Baseline Annual Default Costs (in bps) (using Moody’s Data as of Feb 2008)

PBR credit rating

Moody's\WAL 1 2 3 4 5 6 7 8 9 10

1 Aaa 0.0 0.0 0.0 0.0 0.1 0.1 0.1 0.1 0.1 0.1

2 Aa1 0.0 0.1 0.3 0.5 0.5 0.6 0.7 0.8 0.8 0.9

3 Aa2 0.1 0.4 0.8 1.0 1.2 1.3 1.4 1.5 1.7 1.8

4 Aa3 0.2 0.9 1.7 2.2 2.4 2.7 2.9 3.1 3.3 3.7

5 A1 0.4 1.7 3.4 4.1 4.5 4.9 5.2 5.5 5.9 6.4

6 A2 0.8 3.3 6.5 7.5 8.1 8.6 9.2 9.5 10.1 11.1

7 A3 2.8 7.0 10.6 11.8 12.6 13.5 14.4 14.9 15.6 16.7

8 Baa1 6.4 13.0 16.5 18.1 19.1 20.4 21.7 22.7 23.5 24.3

9 Baa2 16.3 26.3 32.5 36.9 39.8 40.3 42.4 44.0 44.7 45.2

10 Baa3 42.0 61.4 70.0 76.8 81.0 80.0 80.6 81.4 81.9 81.8

11 Ba1 90.5 123.4 134.7 143.1 148.8 143.9 140.4 138.4 137.2 135.7

12 Ba2 173.5 226.2 243.5 257.9 267.6 253.8 241.0 232.5 228.0 224.1

13 Ba3 262.0 295.0 311.3 328.6 349.6 334.4 321.0 313.1 308.2 305.9

14 B1 436.4 453.8 468.5 480.1 495.0 464.0 441.5 425.5 415.2 409.4

15 B2 621.8 573.8 565.2 560.8 567.4 525.7 492.9 467.1 449.6 436.4

16 B3 1,009.1 832.5 789.8 779.3 788.6 726.3 689.6 663.7 641.2 626.1

17 Caa1 1,440.9 1,095.2 1,004.3 983.8 999.3 922.7 879.6 855.0 840.7 839.5

18 Caa2 2,026.5 1,427.1 1,253.0 1,191.4 1,191.9 1,089.4 1,023.7 982.5 960.8 952.3

19 Caa3 3,974.3 2,806.9 2,385.2 2,269.9 2,316.1 2,090.5 1,942.9 1,850.2 1,809.0 1,815.6

20 Ca 7,090.1 7,090.1 7,090.1 7,090.1 7,090.1 7,090.1 7,090.1 7,090.1 7,090.1 7,090.1

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 10NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

10

Determination of PBR Credit Rating

Table J of Appendix 2 converts the ratings of NAIC Approved Ratings Organizations (AROs) and NAIC designations to a numeric rating system from 1-21 that is to be used in the steps below. A rating of 21 applies for any ratings of lower quality than those shown in the table.

For an asset with an NAIC designation that is derived solely by reference to underlying ARO ratings without adjustment, the company shall determine the PBR credit rating as the average of the numeric ratings corresponding to each available ARO rating, rounded to the nearest whole number. Example: public corporate bonds

For an asset with an NAIC designation that is not derived solely by reference to underlying ARO ratings without adjustment, the company shall determine the PBR credit rating as the second least favorable numeric rating associated with that NAIC designation.

Example: non-agency RMBS; traditional private placements

Page 6: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 11NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

11

Table J. Conversion from NAIC ARO Ratings and NAIC Designations to PBR Numeric Rating

Moody's Rating A aa A a1 A a2 A a3 A 1 A 2 A 3 Baa1 Baa2 Baa3

S&P Rating AA A A A + A A A A - A + A A - BBB+ BBB BBB-

Fitch Rating AA A A A + A A A A - A + A A - BBB+ BBB BBB-

DBRS Rating AA A A A high A A A A low A high A A low BBB high BBB BBB low

RealPoint Rating AA A A A + A A A A - A + A A - BBB+ BBB BBB-

A M Best Rating aaa aa+ aa aa- a+ a a- bbb+ bbb bbb-

NA IC Des ignation 1 1 1 1 1 1 1 2 2 2

Numeric Rating 1 2 3 4 5 6 7 8 9 10

Moody 's Rating Ba1 Ba2 Ba3 B1 B2 B3 Caa1 Caa2 Caa3 Ca

S&P Rating BB+ BB BB- B+ B B- CCC+ CCC CCC- CC

Fitch Rating BB+ BB BB- B+ B B- CCC+ CCC CCC- CC

DBRS Rating BB high BB BB low B high B B low CCC high CCC CCC low CC

RealPoint Rating BB+ BB BB- B+ B B- CCC+ CCC CCC- D

A M Bes t Rating bb+ bb bb- b+ b b- ccc+ ccc ccc- cc

NA IC Des ignation 3 3 3 4 4 4 5 5 5 6

Numeric Rating 11 12 13 14 15 16 17 18 19 20

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 12NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

12

Comments on PBR Credit Rating System

The 1-21 PBR credit rating system attempts to provide a more granular assessment of credit risk than has been used for establishing NAIC designations for risk based capital and asset valuation reserve purposes.

The reason is that unlike for RBC and AVR, the VM-20 reserve cash flow models start with the gross yield of each asset and make deductions for asset default costs. The portion of the yield represented by the purchase spread over Treasuries is often commensurate with the more granular rating assigned, such as A+ or A-.

Thus, use of the PBR credit rating system may provide a better match of risk and return for an overall portfolio in the calculation of VM-20 reserves.

However, for assets that have an NAIC designation that does not rely directly on ARO ratings, a more granular assessment consistent with the designation approach is not currently available.

Page 7: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 13NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

13

Step Two:Determine Spread Related Factor

The spread related factor is based on the difference between current and historical mean spreads of index bonds, and it produces the same result for all assets with the same PBR credit rating and WAL.

A floor and cap have been provided to assure that this component cannot reduce the total default cost factor in year one by more than the baseline default cost factor and cannot increase the total default cost factor in year one by more than two times the baseline default cost factor.

The cap is intended to help limit reserve volatility, which remains a concern for both the spread related factor and the maximum net spread adjustment factor.

The spread related factor shall grade linearly from the prescribed amount in year one to zero in years four and after.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 14NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

14

Spread Related FactorSpread Related Factor

The prescribed amount in year one may be positive or negative and shall be calculated as follows:

Multiply 25% by the result of (a) minus (b), where:

(a) equals the current market benchmark spread published by the NAIC consistent with the PBR credit rating and WAL of the asset on the valuation date (see tables F and G in appendix 2 of VM-20).

(b) equals the most current available historical mean benchmarkspread published by the NAIC consistent with the PBR credit rating and WAL of the asset on the valuation date (see tables H and I in appendix 2 of VM-20)

The resulting amount shall not be less than the negative of the baseline annual default cost in year one and shall not be greater than two times the baseline annual default cost in year one.

Page 8: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 15NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

15

Step Three:Determine Maximum Net Spread Adjustment Factor

The maximum net spread adjustment factor shall be the same amount for each starting fixed income asset, and shall grade linearly from the prescribed amount in year one to zero in years four and after.

For each model segment, a comparison is to be made of two spread amounts, both being net of the default costs calculated thus far and net of investment expenses. In each case, the gross option adjusted spread is based on current market prices at the valuation date. The first result represents the weighted average net spread for all the assets in the

model segment as if all the assets were purchased at their current market spreads. The second result represents the net spread for a portfolio of index Baa bonds

(NAIC 2, PBR credit rating of 9, which is the “regulatory threshold asset”) as if the index Baa portfolio were purchased at the current average market spread.

If the first result is higher than the second, additional default costs must be added to each asset until the two results are equal for the first projection year. This additional amount of default cost on each asset then grades off linearly in the model until it reaches zero in year four and after. This process is repeated on each valuation date.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 16NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

16

Maximum Net Spread Adjustment FactorMaximum Net Spread Adjustment Factor

A company that invests in an asset mix earning an average gross spread greater than Baa bonds initially, or an asset mix whose average market spread could widen significantly relative to market spreads for Baa bonds are examples of situations likely to trigger additional assumed default costs either initially or in the future.

The prescribed amount in year one shall be the excess, if any, of (a) over (b): (a) Weighted average net spread for the asset portfolio, calculated as follows:

1. For each asset, calculate a preliminary year one net spread equal to the option adjusted spread of the asset on the valuation date less the sum of the results from steps 1 and 2 (baseline and spread related factor) and less the investment expense for the asset.

2. Calculate a weighted average preliminary year one net spread for the total asset portfolio (i.e., assets subject to this section) using weights equal to each asset’s statement value on the valuation date multiplied by the lesser of 3 years and the asset’s WAL on the valuation date.

(b) The net spread for a “hypothetical asset” (see next slide) which is called the “regulatory threshold asset.”

Page 9: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 17NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

17

Hypothetical Asset Net Spread(based on regulatory threshold asset)

Calculate the preliminary year one net spread for a hypotheticalasset with the following assumed characteristics (the regulatorythreshold asset): A PBR credit rating of 9 (equivalent to Baa2/BBB). A WAL equal to the average WAL on the valuation date for the assets in

the portfolio. An option adjusted spread equal to the current market benchmark spread

published by the NAIC for the assumed PBR credit rating and WAL (see tables F and G in appendix 2 of VM-20).

Investment expense of 0.10%.

The preliminary year one net spread is equal to the option adjusted spread of the asset on the valuation date less the sum of the baseline and spread related factor for the hypothetical asset, and less the investment expense for the hypothetical asset.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 18NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

18

Calculation of Annual Default Cost Factors for Two Sample Portfolios

Page 10: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 19NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

19

Two companies: A and B

Each has $1B in fixed income assets on valuation date

Each company’s portfolio consists of five bonds with different WALs

Distribution of WAL is equivalent for the two companies

Company A maintains higher-rated portfolio than Company B

Note: The current market spread Tables (F and G) may appear high, since they are based on 9/30/2009 bond market conditions. The illustrated OAS for each asset in this example will also seem high, since they are also based on 9/30/2009 bond market conditions.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 20NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

20

Ratings (Moody’s / S&P) for each company’s fixed income asset portfolio by WAL

Bond WAL% of fixed income

assetsCompany A Company B

1 1 10% Aaa / AAA Aa2 / AA

2 5 30% Aa2 / AA Baa2 / BBB

3 10 30% Baa2 / BBB B2 / B

4 20 10% Baa1 / BBB+ Caa1 / CCC-

5 30 20% Aa2 / AA Baa2 / BBB

Page 11: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 21NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

21

Determination of PBR Credit Rating(Section 9.F.3)

Bond WAL NAIC Moody’s S&P PBR Rating

NAIC Moody’s S&P PBR Rating

1 1 1 Aaa AAA 1 1 Aa2 AA 3

2 5 1 Aa2 AA 3 2 Baa2 BBB 9

3 10 2 Baa2 BBB 9 4 B2 B 15

4 20 2 Baa1 BBB+ 8 5 Caa1 CCC- 18

5 30 1 Aa2 AA 3 2 Baa2 BBB 9

Company A Company B

Per 9.F.3.b, must assign PBR rating equivalency for each available ARO rating, using Table J in Appendix 2.

Caa1 → PBR rating of 17CCC- → PBR rating of 19

Whole # average of PBR equivalent ratings for each ARO is 18 for this bond.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 22NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

22

Calculating Baseline Annual Default Costs (Section 9.F.1.a)

Bond WALPBR Credit

RatingAnnual Default

Cost (bps)PBR Credit

RatingAnnual Default

Cost (bps)

1 1 1 0 3 0.1

2 5 3 1.2 9 39.8

3 10 9 45.2 15 436.4

4 20 8 24.3 18 952.3

5 30 3 1.8 9 45.2

Company A Company B

From Appendix 2, Table A

Wtd Avg Ann Default Cost 16.7 247.1

Page 12: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 23NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

23

Calculating Spread-Related Factor(Section 9.F.1.b)

Bond WAL

(A)PBR

Credit Rating

(B)Current Spread

(Tables F&G)

(C)Historical

Spread(Tables H&I)

25% x(B – C)

Minimum:- (baseline

annual default cost)

Maximum: 2x (baseline

annual default cost)

Spread Related Factor

1 1 1 108.9 60.3 12.2 0.0 0.0 0.0

2 5 3 150.2 99.1 12.8 -1.2 2.4 2.4

3 10 9 264.2 202.0 15.6 -45.2 90.4 15.6

4 20 8 247.8 187.0 15.2 -24.3 48.6 15.2

5 30 3 190.8 130.2 15.2 -1.8 3.6 3.6

Company A

“Spread-related factor” cannot be less than negative of baseline annual default cost (9.F.1.a) and cannot exceed 2 x baseline annual default cost.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 24NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

24

Spread-Related Factor by Projection Year (Section 9.F.1.b)

Bond Year 1 Year 2 Year 3 Year 4

1 0.0 0.0 0.0 0.0

2 2.4 1.6 0.8 0.0

3 15.6 10.4 5.2 0.0

4 15.2 10.1 5.1 0.0

5 3.6 2.4 1.2 0.0

Company A

Page 13: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 25NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

25

Calculating Spread-Related Factor (Section 9.F.1.b)

Bond WAL

(A)PBR

Credit Rating

(B)Current Spread

(Tables F&G)

(C)Historical

Spread(Tables H &I)

25% x (B – C)

Minimum: - (baseline

annual default cost)

Maximum: 2x (baseline

annual default cost)

Spread Related Factor

1 1 3 120.3 76.3 11.0 -0.1 0.2 0.2

2 5 9 253.3 192.3 15.3 -39.8 79.6 15.3

3 10 15 730.9 650.5 20.1 -436.4 872.8 20.1

4 20 18 1,168.7 1,311.1 -35.6 -952.3 1,904.6 -35.6

5 30 9 272.0 209.1 15.7 -45.2 90.4 15.7

Company B

“Spread-related factor” cannot be less than negative of baseline annual default cost and cannot exceed 2 x baseline annual default cost.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 26NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

26

Spread-Related Factor by Projection Year

Bond Year 1 Year 2 Year 3 Year 4

1 0.2 0.1 0.1 0.0

2 15.3 10.2 5.1 0.0

3 20.1 13.4 6.7 0.0

4 -35.6 -23.7 -11.9 0.0

5 15.7 10.5 5.2 0.0

Company B

Page 14: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 27NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

27

Maximum Net Spread Adjustment Factor (Section 9.F.1.c.i)

Bond WALPBR

Credit Rating

(A)Baseline Default

Cost

(B)Spread RelatedFactor

(C)Investment Expenses

(bps)

(D)

OAS (bps)

Prelim Year 1

Net SpreadD-A-B-C

1 1 1 0 0 10 100 90

2 5 3 1.2 2.4 10 150 136.4

3 10 9 45.2 15.6 10 275 204.3

4 20 8 24.3 15.2 10 350 300.5

5 30 3 1.8 3.6 10 140 124.6

Company A

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 28NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

28

Maximum Net Spread Adjustment Factor (Section 9.F.1.c.i)

Bond WALPBR

Credit Rating

(A)Baseline Default

Cost

(B)Spread Related Factor

(C)Investment Expenses

(bps)

(D)

OAS (bps)

PrelimYear 1

Net SpreadD-A-B-C

1 1 3 .1 .2 10 170 159.7

2 5 9 39.8 15.3 10 305 240.0

3 10 15 436.4 20.1 10 780 313.5

4 20 18 952.3 -35.6 10 1220 293.3

5 30 9 45.2 15.7 10 325 254.1

Company B

Page 15: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 29NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

29

Weighted Average Preliminary Year 1 Net Spread(Section 9.F.1.c.ii)

Bond WALPrelim Yr 1 Net Spread

Weighting

1 1 90.0 3.57%

2 5 136.4 32.14%

3 10 204.3 32.14%

4 20 300.5 10.71%

5 30 124.6 21.43%

Weighted Prelim Year 1 Net Spread 171.6

Company A

Weightings for Prelim Net Spread:

Bond WALPrelim Yr 1Net Spread

Weighting

1 1 159.7 3.57%

2 5 240.0 32.14%

3 10 313.5 32.14%

4 20 293.3 10.71%

5 30 254.1 21.43%Weighted Prelim Year 1 Net Spread 269.6

Company B

Bond WAL(A)

Statement Value (000,000s)

(B)

Min(WAL,3)A x B %

1 1 100 1 100 3.57%

2 5 300 3 900 32.14%

3 10 300 3 900 32.14%

4 20 100 3 300 10.71%

5 30 200 3 600 21.43%

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 30NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

30

Hypothetical Asset(Section 9.F.1.c.iii)

PBR credit rating = 9

WAL = wtd average of actual portfolio = 12.6. Round to 13

OAS from Table F = 265.4

Investment expenses = 10 bps

Baseline annual default cost factor from Table A = 45.2

Spread-related factor from Tables F and H = 15.6

Preliminary year 1 net spread = 265.4 – 45.2 – 15.6 – 10 = 194.6

Page 16: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 31NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

31

Prescribed Maximum Net Spread Adjustment Factor (Section 9.F.1.c.iv)

% of fixed income assets Company A Company B

Wtd Avg Prelim Year 1 Net Spread 171.6 269.6

Hypothetical Asset Prelim Year 1 Net Spread 194.6 194.6

Difference-23.0 75.0

Max Net Spread Adjust Factor Year 1 0.0 75.0

Year 2 0 50.0

Year 3 0 25.0

Year 4 0 0

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 32NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

32

Total Annual Default Cost

Component Year 1 Year 2 Year 3 Year 4+

Baseline 45.2 45.2 45.2 45.2

Spread-related Factor 15.6 10.4 5.2 0.0

Max Net Spread Adjustment 0.0 0.0 0.0 0.0

Total Default Cost 60.8 55.6 50.4 45.2

Company A: Bond 3

Page 17: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 33NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

33

Total Annual Default Cost

Component Year 1 Year 2 Year 3 Year 4+

Baseline 436.4 436.4 436.4 436.4

Spread-related Factor 20.1 13.4 6.7 0.0

Max Net Spread Adjustment 75.0 50.0 25.0 0.0

Total Default Cost 531.5 499.8 468.1 436.4

Company B: Bond 3

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 34NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

34

Total Annual Default Cost for Year 1

Company A

16.7 + 7.6 + 0 = 24.3 bps

Company B:

247.1 + 10.2 + 75.0 = 332.3 bps

Page 18: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 35NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

35

Reinvestment Spread Assumptions

Gross spreads on new assets purchased in the model subsequent to the valuation date are generally prescribed in VM-20.

VM-20 currently has two prescribed approaches on reinvestment spreads.

Alternative 1: A simple net spread formula favored by the NY State InsuranceDept:

the net yield on reinvestment assets equals the then-current U.S. Treasury interest rate curve times 104% plus 25 basis points.

Since the net spread is prescribed, no assumption is needed for default costs.

Does not currently address what to use for the gross spread assumption to compute market values when modeling asset sales.

Alternative 2: A gross spread methodology developed by the LRWG that reflects historical spread levels and is consistent with the prescribed approach to determine default costs on existing assets (see next slide).

Both approaches to determine reinvestments spreads will be included in the field testing exercise.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 36NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

36

Reinvestment Spread Methodology that is Consistent with Default Costs on Existing Assets (Alternative 2)

For public non-callable bonds:

Gross spreads are determined based on actual current and historical market data, using the same tables used in the calculation of default costs.

The prescribed gross spreads start at current average market spreads in effect at the valuation date (published by the NAIC from a market source) and grade by the start of projection year four to long-term benchmark spreads (derived and published by the NAIC based on actual historical data from the same market source).

Prescribed default costs are then deducted explicitly for purchased assets, using the same approach to calculate default costs as for existing assets (but ignoring step 3: maximum net spread adjustment factor).

For investments other than public, non-callable corporate bonds, gross spreads are not prescribed, but are to be consistent with and in reasonable relationship to the prescribed spreads for public, non-callable corporate bonds.

Page 19: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 37NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

37

Reinvestment Spread Methodology that is Consistent with Default Costs on Existing Assets (Alternative 2)

Gross spread assumptions are needed to compute market values when modeling the sale of starting assets as well as the purchase and sale of reinvestment assets. Spreads used in calculating the market value of assets sold are to be consistent with but not necessarily the same as the spreads used for purchases, recognizing that specific starting assets may have different characteristics than the modeled reinvestment assets.

The methodology incorporates a minimum floor. The company’s model investment strategy together with the prescribed and non-

prescribed spreads must not produce a lower minimum reserve than would result using an alternative investment strategy made up solely of a stated blend of “A2/A” and “Baa2/BBB” public, non-callable corporate bonds along with their associated prescribed spreads.

The proposed blend of 50% A and 50% BBB is intended to represent an approximate equivalent of the industry average asset allocation. This is based in part on data incorporated in a NAIC Rating Agency Work Group report.

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 38NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

38

Reinvestment Spread Methodology that is Consistent with Default Costs on Existing Assets (Alternative 2)

The PBR actuarial report shall include documentation demonstrating compliance with the minimum floor requirement.

In many cases, particularly if the model investment strategy does not involve callable assets, it is expected that the demonstration of compliance will not require running the reserve calculation twice.

For example, an analysis of the weighted average net reinvestment spread on new purchases by projection year (gross spread minus prescribed default costs minus investment expenses) of the model investment strategy compared to the weighted average net reinvestment spreads by projection year of the alternative strategy may suffice.

The assumed mix of asset types, asset credit quality, or the levels of non-prescribed spreads for other fixed income investments may need to be adjusted to achieve compliance.

Page 20: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 39NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

39

Reinvestment Spread Methodology that is Consistent with Default Costs on Existing Assets (Alternative 2)

The model investment strategy:

Must be a representation of the company’s actual investment policy

Is permitted to be complex and incorporate assets for which spreads are not prescribed, or it is permitted to be simple and be expressed as a function only of assets for which spreads are prescribed (or zero for Treasuries) for ease of demonstrating compliance with the minimum floor requirement.

The model strategy and/or non-prescribed spreads must be adjusted if the combination would result in a lower reserve than would be produced by the alternative investment strategy used to determine the minimum floor.

Swap curve spreads are also prescribed for use throughout the cash flow model (not just purchases) to help standardize the treatment of LIBOR-based floating rate assets, swaps, and hedging strategies (but the swap spread tables are not yet included in VM-20).

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 40NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

40

Sample Gross & Net Spreads (Alternative 2)Sample Gross & Net Spreads (Alternative 2)1010--Year Bonds in Projection Years 4+Year Bonds in Projection Years 4+

50% A /AAA AA A BBB 50% BBB BB B

Gross Spreads (Proj Year 4+)

Mean (7-10 Yr Bucket) 112 130 157 216 187 390 65185% Conditional Mean (7-10 Yr Bucket) 99 116 139 195 167 361 608Prescribed Spread (10-Yr Bond) 101 117 143 202 173 361 608

Deductions (Proj Year 4+)

Prescribed Default Cost (10-Yr Bond) 0 2 11 45 28 224 436Anticipated Investment Expense 10 10 10 10 10 25 25Total Deductions 10 12 21 55 38 249 461

Net Spreads (Proj Year 4+)

Net Spread (10-Yr Bond) 91 105 122 147 135 112 147

Note: Gross spread observation period 7/1/2000 - 9/30/2009. Based on Morgan Markets JULI Index.

Below IG (10-yr)Investment Grade (10-yr)

Page 21: Session 7 Defaults And Spreads Final

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 41NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

41

Sample Gross & Net Spreads (Alternative 2)Sample Gross & Net Spreads (Alternative 2)55--Year Bonds in Projection Years 4+Year Bonds in Projection Years 4+

50% A /AAA AA A BBB 50% BBB BB B

Gross Spreads (Proj Year 4+)

Mean (3-5 Yr Bucket) 98 113 138 215 177 390 65185% Conditional Mean (3-5 Yr Bucket) 78 93 117 186 152 361 608Prescribed Spread (5-Yr Bond) 83 99 123 192 158 361 608

Deductions (Proj Year 4+)

Prescribed Default Cost (5-Yr Bond) 0 1 8 40 24 268 568Anticipated Investment Expense 10 10 10 10 10 25 25Total Deductions 10 11 18 50 34 293 593

Net Spreads (Proj Year 4+)

Net Spread (5-Yr Bond) 73 88 105 142 124 68 15

Note: Gross spread observation period 7/1/2000 - 9/30/2009. Based on Morgan Markets JULI Index.

Below IG (5-yr)Investment Grade (5-yr)

Copyright © 2007 by the American Academy of Actuaries

The Year in Review, November 2007 42NAIC VM-20 Impact Study TrainingNovember 18-19, 2010

42

Questions?Questions?