A Rating Agency Perspective On Life Insurer Risk Management Presented by: Joel Levine Vice President & Senior Analyst Life Insurance Group April 26, 2004
Dec 16, 2015
A Rating Agency Perspective OnLife Insurer Risk Management
Presented by:
Joel LevineVice President & Senior AnalystLife Insurance Group
Presented by:
Joel LevineVice President & Senior AnalystLife Insurance Group
April 26, 2004
2
Presentation Outline Usefulness of available risk management
information - rating agency focus versus management information and regulatory reporting
What rating agencies worry about
– Lessons from past failures
– Growing complexity of business – GMDB and UL no
lapse guarantees
– Secondary market for insurance contracts
Hedging VA guaranteed living benefits
Conclusions
3
A Lot Of Data,
Not Enough
Information
4
Rating Agency Focus Credit analysis focuses on operating entity and
enterprise level (holding co.) financial condition
Analyzes the organization’s ability to withstand stress from multiple sources of risk
– Comprehensive in scope: liquidity, asset default,
concentration, interest rate, market, inflation,
actuarial, competitive, reputation, regulatory,
litigation, merger/business integration, etc.
– Ongoing concern point of view, except for liquidity
– Compound interaction among risk factors, not just
one at a time
– Medium term horizon
5
Issuer Management Reporting Focus
Many issuers are managed as decentralized business segments with fragmented reporting
– Separately managed investment portfolios with
tailored ALM requirements
– Management reports (e.g., scenario projections)
are produced on an independent, business segment
basis
– Surplus assets are generally ignored
– Corporate (holding co) balance sheet and income
components are ignored
– Affiliated entities are excluded
– Scenario analysis on a run-off basis
6
Regulatory Reporting Focus Cash flow testing provides useful information,
but has its limitations
– Reserve adequacy focus as opposed to company
performance (e.g., capitalization level over time)
– Not all businesses/products are included
– Performed on a run-off basis
– Surplus assets are ignored
– Corporate (holding co) balance sheet and income
components are ignored
– Affiliated entities are excluded
– Primarily focused on interest rate risk (except for
NAIC C-3 Phase II)
– SAP only, no GAAP information
7
Desirable Supplementary Risk Mgt Reporting
Goal is to achieve a clearer understanding of the drivers of performance and the potential for significant losses – tail risk
– Break-even scenario analysis - what it would take
to impair the organization over a given time
horizon
– Stress economic scenario analysis
– VaR/EaR analysis – a more comprehensive analysis
of risk that would reflect compound interaction of
various risk factors
– Fair value accounting
– Embedded value – sensitivity analysis
8
What Do Rating
Agency Analysts
Worry About?
9
Lessons From Past Failures
General American (1999) – liquidity risk of 7-day putable funding agreements, reinsurance with a weak company, loss of confidence by institutional investors
Confederation Life (1994) – heavy investment portfolio losses in real estate and commercial mortgages, unsuccessful acquisition strategy
First Capital Holding (1991) and First Executive (1991) – losses on high yield bonds, increase in policy surrenders impairing company liquidity
10
The Business Is Getting More Complex and Risky Increasingly, new products are being created with a
focus on guarantee elements
– Equity market uncertainty has made investors more
receptive to floors on returns
– Guarantees have become huge drivers of annuity sales
(GMWB, GMIB)
– Guarantees have crept into life insurance products -
e.g., UL no lapse guarantees (AXXX)
New embedded risks are extremely complex and require sophisticated modeling in order to understand and hedge them
Some companies are developing me-too designs without having a full appreciation of the risk
11
What Keeps A Rating Agency Analyst Up At Night?
Issuers that engage in products/activities that require undemonstrated competencies and/or that lack a cohesive process to manage them
– Does the issuer recognize the risks it has assumed
– e.g., dollar for dollar partial withdrawals?
– VA secondary guarantees – when did life insurers
become expert managers/traders of market risk?
– Does the issuer currently have or can it acquire
sufficient resources to manage the risks?
– Is senior management committed?
– Who’s accountable for the process –
nobody/everybody, multiple committees, CRO?
12
Lesson Learned – GMDB Is More Than Actuarial Risk
Historically, life insurers accumulated actuarial risks and managed them using risk selection, diversification, and reinsurance
GMDB reinsurers tried to “diversify” by writing new business at different points in time (over a market cycle) – benefit payment on death only would provide “diversification” by exercise date
Back-testing using historical equity market prices validated the “insignificant” economic risk of GMDB; risk neutral valuation not well understood
Ignored the systematic nature of market risk and the “temporary” impacts on statutory surplus
13
A Case Study – UL No Lapse Guarantee UL no-lapse guarantees expose issuers to
potentially significant and systematic risks
– Simultaneous occurrence of low lapse rates and low
interest rates may produce very large losses;
moderate adverse deviations can create material
losses
– Moody’s is concerned that issuers’ pricing lapse
assumptions may be too high and rely upon
naive/irrational policyholder behavior
– Pattern of statutory earnings under adverse lapse
scenario is for losses to emerge in later policy
durations; masking of the problem in the early
years
14
A Case Study – UL No Lapse Guarantee (Cont’d)
Portfolio Yield Lapse Rate 7% 5% 3%
6% 15,611 -20,327 -39,435 4% -15,232 -48,246 -64,886 2% -64,997 -91,755 -103,530
Present Value Of Profits At 9% Discount Rate For A $1 Million Policy
15
A Case Study – UL No Lapse Guarantee (Cont’d)
-$5,000
-$3,000
-$1,000
$1,000
$3,000
$5,000
$7,000
$9,000
$11,000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37
Normal
Lapse 4%
Lapse 2%
Statutory Profits With Varying Lapse Rates
16
How Will Middle-Tier Issuers Cope? VA secondary guarantee hedging requires a
significant and expensive commitment of resources
– Systems development to integrate policy admin system
with new actuarial modeling systems
– Hardware – massive amount of policy records and many
thousands of stochastic simulations
– Accounting support and internal controls
– Experienced traders
Outsource it? – can I afford it? - actuarial consulting firm, derivatives dealer with a turnkey program, emerging reinsurance programs
Can a middle-tier issuer remain competitive?
17
Secondary Market For Insurance Contracts
Firms are being formed to facilitate a secondary market for annuity contracts and life insurance policies – and arbitrage the “irrational” policyholder behavior pricing assumptions
– IBuyAnnuities.com
– CoventryFirst.com
Wall Street capital has not been deployed in a significant way to-date, but that could change
Potentially dire implications for some issuers
– Lapse assumptions for in-the-money GMDB,
GMWB/GMIB, $ for $ partial withdrawals
– Mortality assumptions for life insurance policies
18
Analyze This –
What’s The “Right”
Objective For A
Hedging Strategy?
19
VA Secondary Guarantee Hedging Objectives Reduce the tail risk – potential for large
economic (pv of net cash flows) losses
But, subject to external constraints: NAIC RBC, rating agency capital requirements, preserve company shareholder dividend capacity, maximum tolerable GAAP income loss, etc.
Competing constraints make it difficult to resolve analytically (e.g., with an optimizer)
For a rating agency, reported GAAP net income is not necessarily controlling; economics are the primary concern
20
Secondary Guarantees Subject To FAS 133
Both assets and liabilities are marked-to-market, so GAAP results are reasonably predictable with one major exception
Popular dynamic hedging practice is to match the liability “greeks” – delta, rho, gamma, vega, cross-sensitivities; trade-off between effectiveness and cost
Arguments made that implied volatility for long-term derivatives fluctuates and tends to be mean-reverting: makes it difficult to match the liabilities’ sensitivity to changes in volatility (vega)
21
Secondary Guarantees Subject To FAS 133 (Cont’d)
Historical Realized Volatility On S&P 500 Options
Notes1. Underlying S&P 500 data from Bloomberg (daily frequency from 1/2/1941 to 3/26/2004)2. Historical volatility numbers are based off daily S&P 500 returns adjusted by an annualization factor 1
5% and 95% Confidence IntervalMedian
0%
5%
10%
15%
20%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
10-Year Rolling Historical VolatilityPer Annum (%)
0%
5%
10%
15%
20%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
10-Year Rolling Historical VolatilityPer Annum (%)
0%
20%
40%
60%
80%
100%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
22 Day Rolling Historical VolatilityPer Annum (%)
0%
20%
40%
60%
80%
100%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
22 Day Rolling Historical VolatilityPer Annum (%)
Mean:12.72%, Median: 11.24%, 5% Conf: 5.82%, 95% Conf: 24.21%, Current Vol: 14.52%
0%5%
10%15%20%
25%30%
35%40%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
1-Year Rolling Historical VolatilityPer Annum (%)
0%5%
10%15%20%
25%30%
35%40%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
1-Year Rolling Historical VolatilityPer Annum (%)
0200
400600
8001000
12001400
1600
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
S&P 500 Price IndexLevel (Points)
0200
400600
8001000
12001400
1600
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
S&P 500 Price IndexLevel (Points)
0%
5%
10%
15%
20%
25%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
3-Year Rolling Historical VolatilityPer Annum (%)
0%
5%
10%
15%
20%
25%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
3-Year Rolling Historical VolatilityPer Annum (%)
0%
5%
10%
15%
20%
25%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
5-Year Rolling Historical VolatilityPer Annum (%)
0%
5%
10%
15%
20%
25%
Feb
-41
May
-46
Aug
-51
Nov
-56
Feb
-62
May
-67
Aug
-72
Nov
-77
Feb
-83
May
-88
Aug
-93
Dec
-98
Mar
-04
5-Year Rolling Historical VolatilityPer Annum (%)
Mean:13.68%, Median: 12.90%, 5% Conf: 9.66%,95% Conf: 20.29%, Current Vol: 20.91%
Mean:13.51%, Median: 13.56%, 5% Conf: 9.90%,95% Conf: 17.47%, Current Vol: 18.18%
Mean:13.47%, Median: 12.52%, 5% Conf: 7.96%,95% Conf: 22.40%, Current Vol: 13.75%
Mean:13.65%, Median: 12.53%, 5% Conf: 8.92%,95% Conf: 22.80%, Current Vol: 21.06%
22
Secondary Guarantees Subject To FAS 133 (Cont’d)
If long-term implied volatility is “noise”, can it be safely ignored – and could one hedge with futures only?
– Maybe, but one shouldn’t ignore the risk that delta
may change significantly with large market moves
(gamma) – therefore, may need some option
exposure
Under NAIC C3 Phase II, stochastic simulations will be used to determine SAP reserve and RBC levels
23
Secondary Guarantees Subject To FAS 133 (Cont’d)
NAIC C3 Phase II methodology ignores spot market parameters such as implied volatility
– Hedging vega will not necessarily produce a more
stable SAP financial result
– Responses of the hedge assets to changes in
implied volatility will not be offset by
corresponding changes in the values of the SAP
liabilities
24
Emerging Direction Of Hedging VA Secondary Guarantees Index hedge – long-term options covering well
defined risks, while the insurer retains the uncertain, unhedgeable risks
– Hedge payoff based upon an assumed basket of indices,
with provision for basket changes over time, and pricing
that varies accordingly
– Insurer retains the basis risk – basket vs actual fund
performance
– Long-tenor hedge whose payoff approximates the
payoff pattern of the particular GLB design; issuer
would assume the basis risk between the actual VA GLB
payoff and the approximation
– Contractholder exercise efficiency would be another
source of error
25
Future Direction Of Hedging VA Secondary Guarantees Index hedge (cont’d)
– Rating implications would depend upon the
effectiveness of the hedge
• Insurer would need to demonstrate effectiveness
through stochastic modeling that the basis risk
would be manageable – i.e., the tail risk (economic
perspective) would not be excessive
– SAP impact would need to be considered – how
would statutory surplus be impacted using such a
hedge under various scenarios?
• SAP surplus impact might be mitigated by using a
combination reinsurance/derivatives structure
26
Conclusions
27
Conclusions Life insurers are rapidly expanding into new types of less
familiar risks – e.g., trading market volatility
Management and regulatory reporting needs to catch up
with the complexity of these new risks – rapid strides are
being made
Risk management processes have been developed and are
evolving - but are untested under extreme market conditions
Basis risk and modeling risk inherent in hedging are difficult
to quantify
Insurance business has become more risky - assessment of
risk management skills has become more challenging and
critical to the credit ratings process