The table below contain the expected and actual payouts in respect of the lapse by year.

Expected payouts in respect Actual payouts in respect of

Year
of lapses lapses
2015 143,767,446 121,774,575
2016 181,244,022 161,890,816
2017 178,138,758 181,177,802
2018 152,604,107 191,588,138
2019 170,349,129 189,353,332
2020 186,216,000 195,040,124
2021 191,499,664 246,163,957
2022 187,968,157 250,548,243
 Lapses 7 year 5 year 2 year 1 year 1 std dev
113.4% 120.7% 130.9% 133.3% 12.1%

Expected Actual Observed Sample Sample

Year X bar
decrement decrement A/E mean variance

2015 143,767,446 121,774,575 84.7%

2016 181,244,022 161,890,816 89.3%
2017 178,138,758 181,177,802 101.7%
2018 152,604,107 191,588,138 125.5%
2019 170,349,129 189,353,332 111.2% 102.5% 2.743% 102%
2020 186,216,000 195,040,124 104.7% 106.5% 1.764% 106%
2021 191,499,664 246,163,957 128.5% 114.3% 1.473% 114%
2022 187,968,157 250,548,243 133.3% 120.7% 1.474% 121%
1,537,536,987

Actual vs. Expected Decrements

140.0%
130.0%
120.0%
Actual vs Expected

110.0% Column Z
100.0% 1 in 10 years
90.0% 1 in 20 years
80.0% Observed 5 year A/E
70.0%
60.0%
50.0%
2019 2020 2021 2022
Year

Comments:
1) Gradually lapse rates are rising (especially in recent years).
2) Observed lapse rates for 2021 and 2022 are outside the 1 in 10 and 1 in 20 confidence intervals, and just within 1 in 100
3) Therefore, the current fixed lapse rate of 3.8% is not appropriate and needs to be revised upwards.
4) Considering the last eight years of data, there has been an average lapse rate of 4.2% (i.e., total decrements/total numb
5) Potentially, we could consider having a variable lapse rate assumption going forward as the lapse rates have been increa
 1/10 1/10 1/20 1/20 1/100 1/100
100% 86.0% 94.0% 94.0% 96.0% 98.5% 99.5%

lower upper lower upper lower upper

Lapse Assumption df
bound bound bound bound bound bound

3.8%
3.8%
3.8%
3.8%
3.8% 4 90.8% 114.6% 85.4% 117.3% 75.6% 134.1%
3.8% 4 92.6% 111.7% 88.3% 113.9% 80.4% 127.3%
3.8% 4 93.2% 110.7% 89.3% 112.7% 82.1% 125.0%
3.8% 4 93.2% 110.7% 89.3% 112.7% 82.1% 125.0%
4.20%

ents

Column Z
1 in 10 years
1 in 20 years
Observed 5 year A/E

ntervals, and just within 1 in 100 years confidence interval.

e., total decrements/total number of policies). Therefore, the lapse rate assumption should at least be increased to 4.2% per annum.
the lapse rates have been increasing. Albeit, this might increase calculation complexities.
 Bias adj. 1/10 1/10 1/20 1/20 1/100 1/100
98.5% 99.5%
86% lower 94% upper 94% lower 96% upper lower upper
Year 5 Year A/E bound bound bound bound bound bound

2019 102% 75.6% 9.8% 5.4% 23.8% 2.7% 16.8%

2020 106% 80.4% 7.9% 4.3% 19.1% 2.1% 13.5%
2021 114% 82.1% 7.2% 3.9% 17.5% 2.0% 12.3%
2022 121% 82.1% 7.2% 3.9% 17.5% 2.0% 12.3%
0

ncreased to 4.2% per annum.

 Lapses 7 year 5 year
102.6% 109.2%

Number of Expected Actual Observed

Year Experience
policies decrement decrement A/E

2015 3,783,353,838 (37,126,286) 158,900,861 121,774,575 76.6%

2016 4,769,579,523 (38,431,524) 200,322,340 161,890,816 80.8%
2017 4,687,862,048 (15,712,404) 196,890,206 181,177,802 92.0%
2018 4,015,897,557 22,920,440 168,667,697 191,588,138 113.6%
2019 4,482,871,803 1,072,717 188,280,616 189,353,332 100.6%
2020 4,900,421,059 (10,777,561) 205,817,684 195,040,124 94.8%
2021 5,039,464,840 34,506,433 211,657,523 246,163,957 116.3%
2022 4,946,530,442 42,793,964 207,754,279 250,548,243 120.6%
36,625,981,109 1,537,536,987

Actual vs. Expected Decrements

140.0%
130.0%
120.0%
Actual vs Expected

110.0% Column Z
100.0% 1 in 10 years
90.0% 1 in 20 years
80.0% Observed 5 year A/E
70.0%
60.0%
50.0%
2019 2020 2021 2022
Year

Comments:
1) The change in lapse assumption rate from 3.8% to 4.2% has led to a more accurate modelling of lapse rates.
2) This has brought back the ratio ("the Ratio") of the actual decrements over the expected decrements within 1 in 10, 1 in
3) This has led to a drop in Sample mean and Sample Variance of the Ratio given more accurate modelling of lapse rates.
4) Given the Ratio has been increasing in recent years, the Ratio is at the lower end of the confidence intervals in the earlie
5) The expected decrements have increased by roughly 10.5% (4.2%/3.8%), which is in line with our expectations.
6) Overall, the above movements in results are consistent with the increase in lapse rate assumption from 3.8% to 4.2%.
 2 year 1 year 1 std dev 1/10 1/10 1/20
118.4% 120.6% 11.0% 100% 86.0% 94.0% 94.0%

Sample Sample lower upper lower

X bar Lapse Assumption df
mean variance bound bound bound

4.2%
4.2%
4.2%
4.2%
92.7% 2.245% 93% 4.2% 4 91.6% 113.2% 86.8%
96.4% 1.444% 96% 4.2% 4 93.3% 110.6% 89.4%
103.4% 1.206% 103% 4.2% 4 93.9% 109.7% 90.3%
109.2% 1.206% 109% 4.2% 4 93.9% 109.7% 90.3%
4.20%

umn Z
10 years
20 years
erved 5 year A/E

of lapse rates.
ments within 1 in 10, 1 in 20 and 1 in 100 confidence intervals.
modelling of lapse rates.
ence intervals in the earlier years (2019 and 2020) and has gradually moved to the upper end (2021 and 2022).
ur expectations.
tion from 3.8% to 4.2%.
1/20 1/100 1/100
96.0% 98.5% 99.5% Bias adj. 1/10 1/10 1/20

upper lower upper

bound bound bound 86% lower 94% upper 94% lower
Year 5 Year A/E bound bound bound

115.6% 77.9% 130.9% 2019 93% 77.9% 8.9% 4.8%

112.5% 82.3% 124.7% 2020 96% 82.3% 7.1% 3.9%
111.5% 83.8% 122.6% 2021 103% 83.8% 6.5% 3.6%
111.5% 83.8% 122.6% 2022 109% 83.8% 6.5% 3.6%
1/20 1/100 1/100
98.5% 99.5%
96% upper lower upper
bound bound bound

21.6% 2.4% 15.2%

17.3% 1.9% 12.2%
15.8% 1.8% 11.2%
15.8% 1.8% 11.2%