Cost of guarantees

Yes, future RB consistent with PRE should be allowed for so that the cost of guarantee is [guaranteed benefits including future bonuses] less [projected asset share].

A model that uses a real-world calibration will show scenarios that follow the distribution of what we realistically expect. For example, we might realistically believe that the value of a share in one year's time could be 130, 110 or 70 with equal probability.

If we had a put option with strike price of 100 then we would expect a payoff 100 - 70 = 30 in one-third of the scenarios. So the expected payoff is 30 / 3 = 10.

However, the price of this option would not be 10. Investors are generally risk averse, so they would be prepared to pay a higher price for the option (ignoring discounting for simplicity), say 15.

A risk neutral calibration gives a model that would replicate market prices. For example, the Black-Scholes equation uses a risk neutral valuation. There are lots of ways to calibrate our model to be risk neutral, but one possibility is that the value of this share in one year's time could be 130 (with prob. 0.25), 110 (with prob. 0.25), or 70 (with probability 0.5), so that the expected option payoff is (100-70) x 0.5 = 15 as required.

Best wishes

Mark

 

The FPRL covers:

• costs of guarantees
• cost of financial options
• cost of smoothing
• cost of planned future enhancements to benefits not already allowed for in either the retrospective or prospective methods.

We wouldn't need to allow for terminal bonus in the cost of guarantees as we'd cut the TB to zero if the guarantees were biting.

However, we would need to allow for TB in the cost of smoothing in order to cover the costs of smoothing payouts to above 100% of asset shares.

Best wishes

Mark