Page 1

GOALS FOR THIS SESSION Status of American Academy of Actuaries’ Economic Scenario Work Group (ESWG)

1. Review the mandate of the Economic Scenario Work Group 2. Summarize recommendations

NAIC LHATF Winter Meeting December 4, 2008

3. Highlight ongoing challenges and discuss next steps

ESWG Chair Max Rudolph, FSA CFA CERA MAAA

Copyright © 2008 by the

2008 by the Copyright © 2008 by Copyright the American© Academy of Actuaries American Academy Economic Scenario Work Group Report of Actuaries Life & Health Actuarial Task Force Winter Meeting NAIC LHATF December 4, 2008 December 4, 2008

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 1 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 1

ESWG GOALS (SET BY STANDARDS FOR STOCHASTIC METHODS WORK GROUP (SSMWG) FALL 2006) AND STATUS

UPDATES FROM C-3 PHASE I ADOPTED IN 1999

1. The ESWG will recommend a prescribed generator containing updated parameters – Minor changes from Fall 2007 report – Full scenario set and subset have been provided to Academy work groups and SOA Research team for testing

„

ESWG chose to continue with Stochastic Log Volatility model used for C-3 Phase I adopted in 1999

„

Refreshed some parameters using Treasury data from 1953 – 2008

„

Prepared a Microsoft Excel generator for broad distribution to the industry

„

Documented the model, data sources, key decisions and parameters

2. The ESWG will recommend calibration criteria so companies can use their own generator – Provide to NAIC at winter meeting 3. Generators will not use pre-selected criteria to approximate specific blocks of business – Different methodology from C-3 Phase 1

Copyright © 2008 by the

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 2 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 2

1

Copyright © 2008 by the

2

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 3 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 3

3


ASSUMPTION UPDATES FROM C-3 PHASE I ADOPTED IN 1999 „

Soft cap of 18% limits the maximum long rate – reduces maximum rates with minimal impact on overall results

„

Yield curve interpolation uses historical curves

„

Established processes (formulas) for automatically updating Mean Reversion Parameter (MRP) for target long interest rate – Long L rate t is i th the 20 20-year US T Treasury rate t † Current MRP, 5.50% † C-3 Phase I MRP, 6.55%

CALIBRATION GOALS

Copyright © 2008 by the

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 4 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 4

„

Subject to calibration requirements, companies should be able to use their preferred interest rate generator

„

Objective – allow models with results similar to the Academy model with Academy parameterization to “pass”

„

Criteria should be dynamic and not require frequent revision by the Academy

„

Standards will include qualitative and quantitative requirements – Qualitative means documentation supporting choices made Copyright © 2008 by the

4

CALIBRATION CRITERIA

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 5 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 5

5

Football goal posts analogy for tolerances

„

By definition, the 10,000 scenarios of the SLV model are “calibrated” – Subsets of this scenario set must pass the calibration criteria

„

Key idea is to develop “acceptable tolerances” around the SLV statistics

„

ESWG recommends the following statistics: – Distribution results at the 5% and 95% point-in-time percentiles for long rate and short rate distributions at 1, 5, 10, and 30 year horizons Relaxed tolerances at 1 year time horizon – Distribution results at the 5% and 95% cumulative percentiles for the spread distribution at the 30-year time horizon †

Copyright © 2008 by the

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 6 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 6

Copyright © 2008 by the

6

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 7 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 7

7


STOCHASTIC COMPANY GENERATOR WITH CALIBRATION CRITERIA VS. REGULATORY PRESCRIBED DETERMINISTIC SCENARIOS

MEAN REVERSION PARAMETER (MRP) „

Mean reversion parameter will adjust over time to better reflect current conditions. The reversion rate is a simple average of the – Long rate (median over most recent 50 years adjusted down by 25 bp) – Long L rate t (mean ( over th the mostt recentt 36 months) th ) – Rounded to near 25 basis points

„

Stochastic company generator – Captures uniqueness of Asset/Liability portfolio risk – Evolution to best practice – Consistent with internal risk assessment – Peer review allows companies p to better understand their situation in order to defend it – Allows modeling of other factors (e.g., hedging, currency, equity, credit spreads) – Some interest rate scenarios move between high and low

„

Prescribed deterministic – Standard scenario set provides consistency between companies – Fewer scenarios require less effort – Regulator needs less expertise to peer review

Copyright © 2008 by the

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 8 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 8

Copyright © 2008 by the

Copyright © 2008 by the American Academy of Actuaries American Academy Actuaries Copyright © 2007 by the American Academy Actuaries Economic Scenario Work Group Reportof of NAIC LHATF Life in & Health Task Force Winter Meeting 9 The Year Review,Actuarial November 2007 December 4, 2008 December 4, 2008 9

8

9

Description of the Stochastic Log Volatility (SLV) Model

NEXT STEPS

The SLV model simulates the following three (3) correlated stochastic processes in discrete monthly time:

„

MRP update frequency Ad hoc with NAIC approval Annually using formula – Recommendation is to refresh early in calendar year and use until following calendar year Quarterly using formula

i

1.

†

2.

†

3.

1 t

The natural logarithm of the long maturity interest rate

αt

The nominal spread between the long and short maturity rates

νt

The natural logarithm of the volatility of the long maturity rate process

The SLV discrete time equations are:

i = Max 1 λL , Min ⎡⎣ 1 λU , (1 − β1 ) ⋅ 1 it −1 + β1 ⋅ ln τ 1 +ψ ⋅ ( 2τ t − α t −1 ) ⎤⎦ + 1σ t ⋅ 1 Z t

1 t

α t = (1 − β 2 ) ⋅ α t −1 + β 2 ⋅ 2τ t + φ ⋅ ( 1it −1 − ln l τ 1 ) + σ 2 ⋅ 2 Z t ⋅ ( 1 rt −1 )

θ

ν t = (1 − β3 ) ⋅ν t −1 + β3 ⋅ ln τ 3 + σ 3 ⋅ 3 Z t

†

where

„

Company generators must calibrate each time scenarios are built

i = ln ( 1 rt )

1 t 1

λ = ln ( 1 rMax )

1

„

r = exp ( 1 it ) − α t

Release current versions of models, scenarios, and scenario picking tool

2 t

If 2 rt < 2 rMin , then 2 rt = κ ⋅ 1 rt

σ t = exp (ν t )

1 1

Copyright © 2008 by the

Copyright © 2008 by the American Academy of Actuaries Academy of Actuaries Copyright © 2007Scenario by theAmerican American Academy Economic Work Group Reportof Actuaries NAIC LHATF Lifein&Review, Health Actuarial Task Force Winter Meeting 10 The Year November 2007 December 4, 2008 December 4, 2008 10

U

λL = ln ( 1 rMin )

Z t , 2 Z t , 3 Z t ~ N ( 0,1) with constant correlation matrix ρ Copyright © 2008 by the

10

Copyright © 2008 by the American Academy of Actuaries Academy of Actuaries Copyright © 2007Scenario by theAmerican American Academy Economic Work Group Reportof Actuaries NAIC LHATF Lifein&Review, Health Actuarial Task Force Winter Meeting 11 The Year November 2007 December 4, 2008 December 4, 200811

11

lhatf_pres_dec08  
Read more
Read more
Similar to
Popular now
Just for you