Joint Regional Seminar 2016 Risk Analysis of Equity-linked Products - - PowerPoint PPT Presentation

1

Joint Regional Seminar 2016

Risk Analysis of Equity-linked Products

2

Equity-linked products

3

Structured products

• Market-linked investment (Investment-linked products)

– Credit-linked notes (CLN) – Interest rate-linked notes – Equity-linked note (ELN) – FX and commodity-linked notes – Equity-indexed annuities (EIA) – Variable annuities (VA)

• Guaranteed benefits
• Embedded options

– Exotic option features

4

EIA vs VA

• Equity-indexed annuities (EIA)

– Like a call option

• Variable annuities (VA)

– Like a put option

Fixed Annuity 0 1 2 3 4 5 6

$100,000

(fixed at 4%)

$4,000 $4,000 $4,000 $4,000 $4,000 $4,000

Indexed Annuity 0 1 2 3 4 5 6

$100,000

(capped at 8%) (floor rate: 0%)

$4,000 $5,000 $8,000 $2,000

 Periodic interest credit between $0 and $8,000  PTP indexing

Index: 100 104 105 108 109 102 99

$8,000

EIA guarantee

index payoff

VA guarantee

• Guaranteed minimum

maturity benefit (GMMB)

• Guaranteed minimum death

benefit (GMDB)

• Guaranteed minimum

accumulation benefit (GMAB) VA guarantee

index payoff

5

Historical development of investment-linked insurance

• UK

– Unit-linked policy (1957)

• US

– Variable life (mid-1970s) – Variable universal life (1986)

• HK

– Investment-linked long-term insurance products (late 1980s) – [OCI] As of 2013, office premiums amounted to

• HKD 69,895.8 million in-force business (29.0%)
• HKD 19,115.7 million new business (21.5%)

(of total office premiums of individual life business)

6

Special Features

• Airbag
• Daily accrual
• Early call

Acknowledgment: Investor Education Centre

• Airbag
• Daily accrual
• Early call

7

Examples

• Target redemption note
• Autocall structured deposit
• Accumulators
• Variable life insurance
• Equity indexed annuity

Target Redemption Note – Example (2004):

• 7.5% USD Target Redemption Index Linked Deposit
• Selling point

 Enjoy potentially higher returns with Index Linked Deposit

• 100% principal protection plus 7.5% guaranteed coupon return over a

maximum of 5-year investment period

• 1st year annual coupon is guaranteed at 3.5% (), payable semi-

annually.

• The remaining coupon rate of 1% will be based on the LIBOR
• movement. The inverse floater formula is

Max{7% – 2 × (6-month LIBOR), 0}

• However, the total coupon received cannot shoot beyond the target

accumulation rate of 7.5%. If the coupon payment accrued during the deposit period is less than the target rate, then the remaining amount will be paid at maturity. Otherwise, it will be terminated earlier. Autocall Structured Deposit – Example (2009):

• Selling point

 Profit from China’s equity market recovery

• 7-year contract
• Principal received upon maturity only.
• If it is redeemed or sold prior to maturity, the

investor may face fees.

• hard to sell in the secondary markets.
• Coupon payment: 5% guaranteed () at the end of the first year
• If the worst off stock in the basket > 10%, it is “auto-called” and will pay.
• Otherwise, the contract continue to the next year until the 7th in which no

coupon would be paid except for the guaranteed 5%.

8

Risk involved in equity-linked investments

• Not principal protected
• Limited potential gain
• Credit risk of the issuer
• No collateral

Acknowledgment: Investor Education Centre

• Limited market making
• Different from investing

in the reference asset(s)

• Conflicts of interest

Practical issues in modeling

• Model financial market events

Deterministic or random

• Model policyholder events

Based on various contingencies

9

Data issues

• Data source:

Actuarial organizations such as Society of Actuaries Canadian Institute of Actuaries Bloomberg Yahoo! Finance

• Data should be readily available and clean
• Good data enhances credibility of modeling

results

10

11

Level of data

• Country or region index:

Standard & Poor 500, Hang Seng Index

• Sub-index:

Hang Seng Property Index

• Individual stocks
• Which one should be used?
• What is the composition of your portfolio?

Length of historical data

• Usually starts from initial public offering
• Maybe too short for long term estimations
• Proxies can be prior company; parent, sister or

competitor in same industry

• Maybe too long and include extreme events
• Beware of change in name or type of shares

12

Frequency of data

• Annual
• Monthly
• Daily

13

Issues with daily equity data

• Missing data
• Typos in data
• Data for holidays (calendar days vs trading days)
• Typhoons (exchange closure)
• Trading suspension

14

Timing of daily equity data

• Last or mid-day
• Time difference for data from different exchanges
• Closing: median of last 5 trades
• Adjusted closing

15

Adjustment of equity data

• Merger or split
• Currency
• Dividend: cash or stock
• Price vs return

16

Size of data files

• Can be very large
• Difficulty in storing and transferring data

17

Modeling approaches

• Deterministic vs stochastic
• Choice of model

18

Deterministic approach

• Factors:

– Interest rate – Dividend rate – Economic growth rate

• Relationship of equity returns with these factors
• Scenarios to be tested

19

Stochastic approach

• Number of simulation paths
• Does it involve nested stochastic?
• Is the range of results reasonable?
• Is it necessary to truncate the extreme values?

20

Choice of model

• Requirements for a model to work
• Will the model produce odd patterns of results?
• Is it appropriate for the product under study?
• Compare modeling results with theoretical values
• Monitor differences between actual and modeled

values

21

Types of models

• Cascade model: Wilkie model
• Log-normal model
• Draw down model
• Regime switching model
• Stochastic volatility model

22

Wilkie model

23

Log-normal model

yt = μ + σεt with εt ~ N(0, 1)

• r

yit = μi + σiεit with εit ~ N(0, 1) and E(εitεjt) = ρij where i ≠ j

24

Draw down model

yt = κ + βdt + σεt with εt ~ N(0, 1) and dt = Min(0, dt-1 + yt)

25

Regime switching model

In regime 1, yt = μ1 + σ1εt with εt ~ N(0, 1) In regime 2, yt = μ2 + σ2εt with εt ~ N(0, 1) Pr(Going to regime 2 | Regime 1) = p12 Pr(Going to regime 1 | Regime 2) = p21

26

Stochastic volatility model

27

Goodness of fit measures

• Log-likelihood
• Akaike information criteria

AIC = - 2*log-likelihood + 2*npar

• Schwartz Bayes criteria

SBC = - 2*log-likelihood + npar*log(nobs)

• Quantile-Quantile (QQ) plot of residuals

28

Calibration of equity returns

• Criteria on percentiles for different time horizons
• How stringent are the calibration criteria?
• Is correlation among returns reflected?
• Can some parameters be negative?
• When will the scenarios have to be re-calibrated?

29

Simplified model

• For quick turnaround
• Must provide reasonably good fit to full model
• Can be used to predict outcome of full model run

30

Risk measures

• Standard deviation
• Value at risk
• Conditional tail expectation
• Opportunity cost compared to another investment

31

Q & A

32