Introduction

FFTCalc by Glenn Meyers

FFTCalc is a spreadsheet that can be used to calculate an aggregate loss distribution with correlated
frequency distributions. The method is based on Section 10.2 of the report "Aggregation of Correlated
Risk Portfolios: Models & Algorithms" which was prepared by Shaun Wang for the CAS Committee on
the Theory of Risk.
Running FFTCalc
1. Input the parameters of the negative binomial claim count distributions. I have taken the liberty of
parametrizing the claim count distribution in terms of the expected claim count and the contagion
parameter, c, instead of Shaun's alpha and beta. If you want to use his parametrization, use the
transformations: Expected Claim Count = alpha*beta and c = 1/alpha.
2. Input the probabilities of the claim severity distributions. Because the aggregate calculation is done
with the FFT, it is necessary to express the claim severity distributions as a discrete distribution. I have
chosen units of one. If you change the units, be sure to adjust the means, variances and the
covariances of the aggregate loss distribution and the claim severity distributions.
3. Input the omega parameter. Note that Shaun's Theorem 10.2 put restrictions on omega. FFTCalc
warns you if your choice of a positive omega is not permissible.
4. Invoke the macro "AggCalc" with the command "Crtl A". Be sure you have the Excel "Analysis
ToolPak" and the "Analysis ToolPak - VBA" add-ins installed on your Excel program.
Calculation Steps
1. Calculate the FFT of each claim severity distribution. Put the results below in the same column under
the heading "FFT of Claim Severity Distribution". Note that the FFT is a vector of complex numbers.
Excel treats complex numbers as labels.
2. Calculate the expression (1-beta(j)*(FFT(j) -1))^(alpha(j)*omega) for each line of insurance. Put the
results below the FFT's under the heading "Operations on Severity FFT."
3. Sum the results of Step 2 and place them to the right under the heading "Sum Ops".
4. Calculate the FFT of the aggregate loss distribution = (Step 3 result - # Lines +1)^(-1/omega). Put the
results to the right of the Step 1 results under the heading "Agg FFT"
5. Invert the Aggregate FFT to obtain the aggregate probabilities. Put the results immediately to the left
of Step 4 under the heading "InvAggFFT". The imaginary part of the inverse FFT will be zero. The real
part then "unlabled" and put to the right of the claim severity distributions under the heading "Aggregate
Probability".
The Cycle Problem
The FFT has a "Cycle" property that can cause incorrect calculated aggregate probabilities. To illustrate,
the highest aggregate probability calculated on this spreadsheet is 127. If the parameters you select
indicate a positive probability for higher aggregate losses, Pr{128} will be added to the indicated Pr{0},
Pr{129} will be added to the indicated Pr{1}, and so on. This spreadsheet offers ways to check if this is
happening. First if Pr{127} = 0, it will usually be the case that the probability of higher losses is also 0
(unless only even claims are permissible). Next it compares the calculated mean and variance of the
aggregate loss distribution with the indicated theoretical mean and variance. If they are the same, it is
unlikely that there is a cycle problem.

Page 1

FFTCalc
Calculate Aggregate Loss Distribution using the Fast Fourier Transform
Line Statistics
Aggregate Statistics
Count Covariance Matrix
Line 1
Line 2
Omega
0.090000
Line 1
Line 2
Alpha
10.00
10.00 OmegaOK?
Yes
Line 1
7.500000
2.250000
Beta
0.500000
0.500000
E[Agg]
21.000000
Line 2
2.250000
7.500000
E[Count]
5.000000
5.000000
V[Agg]
90.895000
V[Count]
7.500000
7.500000 CalcE[Agg]
21.000000
Count Correlation Matrix
c
0.100000
0.100000 CalcV[Agg]
90.894996
Line 1
Line 2
E[Sev]
2.100000
2.100000
LastProb
0.00E+00
Line 1
1.000000
0.300000
V[Sev]
0.490000
0.490000
Line 2
0.300000
1.000000
V[LAgg]
35.525000 35.525000

Claim Severity Distribution

0
0
0
1
0.200000
0.200000
2
0.500000
0.500000
3
0.300000
0.300000
4
0.000000
0.000000
5
0.000000
0.000000
6
0.000000
0.000000
7
0.000000
0.000000
8
0.000000
0.000000
9
0.000000
0.000000
10
0.000000
0.000000
11
0.000000
0.000000
12
0.000000
0.000000
13
0.000000
0.000000
14
0.000000
0.000000
15
0.000000
0.000000
16
0.000000
0.000000
17
0.000000
0.000000
18
0.000000
0.000000
19
0.000000
0.000000
20
0.000000
0.000000
21
0.000000
0.000000
22
0.000000
0.000000
23
0.000000
0.000000
24
0.000000
0.000000
25
0.000000
0.000000
26
0.000000
0.000000
27
0.000000
0.000000
28
0.000000
0.000000
29
0.000000
0.000000
30
0.000000
0.000000
31
0.000000
0.000000
32
0.000000
0.000000
33
0.000000
0.000000
34
0.000000
0.000000
35
0.000000
0.000000

Aggregate
Limited
Probability Pure Prem.
0.000895
0.000000
0.000914
0.999105
0.002796
1.997296
0.004136
2.992692
0.006358
3.983951
0.009369
4.968851
0.012290
5.944383
0.016020
6.907624
0.019780
7.854846
0.023607
8.782288
0.027501
9.686123
0.031095
10.562457
0.034424
11.407695
0.037334
12.218510
0.039726
12.991991
0.041590
13.725745
0.042870
14.417910
0.043575
15.067205
0.043725
15.672925
0.043354
16.234920
0.042513
16.753561
0.041262
17.229690
0.039664
17.664556
0.037787
18.059758
0.035697
18.417173
0.033456
18.738892
0.031123
19.027154
0.028750
19.284293
0.026381
19.512682
0.024055
19.714691
0.021804
19.892644
0.019651
20.048793
0.017616
20.185290
0.015710
20.304172
0.013942
20.407343
0.012315
20.496572

Page 2

Aggregate Covariance Matrix

Line 1
Line 2
Line 1
35.525000
9.922500
Line 2
9.922500 35.525000
Aggregate Correlation Matrix
Line 1
Line 2
Line 1
1.000000 0.27931034
Line 2 0.27931034
1.000000

FFTCalc
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85

0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000

0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000

0.010829
0.009482
0.008269
0.007182
0.006215
0.005358
0.004603
0.003942
0.003364
0.002863
0.002428
0.002054
0.001733
0.001457
0.001223
0.001023
0.000854
0.000712
0.000591
0.000490
0.000406
0.000335
0.000276
0.000227
0.000186
0.000152
0.000125
0.000102
0.000083
0.000067
0.000055
0.000044
0.000036
0.000029
0.000023
0.000019
0.000015
0.000012
0.000010
0.000008
0.000006
0.000005
0.000004
0.000003
0.000003
0.000002
0.000002
0.000001
0.000001
0.000001

20.573486
20.639571
20.696173
20.744507
20.785659
20.820596
20.850176
20.875152
20.896186
20.913856
20.928663
20.941042
20.951367
20.959959
20.967093
20.973005
20.977894
20.981928
20.985251
20.987982
20.990223
20.992059
20.993559
20.994784
20.995782
20.996594
20.997254
20.997788
20.998221
20.998572
20.998854
20.999082
20.999265
20.999413
20.999531
20.999626
20.999702
20.999763
20.999812
20.999851
20.999882
20.999906
20.999926
20.999941
20.999954
20.999963
20.999971
20.999977
20.999982
20.999986

Page 3

FFTCalc
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127

0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000

0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000

FFT of Claim Severity Distribution

0
1
1
1
0.9941044070.994104407
2
0.9765116860.976511686
3
0.9475022570.947502257
4
0.9075377060.907537706
5
0.8572522200.857252220

0.000001
0.000001
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000

20.999989
20.999991
20.999993
20.999995
20.999996
20.999997
20.999997
20.999998
20.999998
20.999999
20.999999
20.999999
20.999999
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000
21.000000

Agg FFT

InvAggFFT
1
8.9488120911
0.4700343609.1378734816
-0.25844533 2.7960499404
-0.38742858 4.1363035888
-0.21085723 6.3583424024
-0.07068254 9.3690924122

Page 4

0.200000 0.99879546
0.500000 0.99518473
0.300000 0.98917651
0.99410441

FFTCalc
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55

0.7974408580.797440858
0.7290448620.729044862
0.6531343260.653134326
0.5708885540.570888554
0.4835745110.483574511
0.3925237880.392523788
0.2991085420.299108542
0.2047168880.204716888
0.1107282300.110728230
1.8489003881.848900388
-7.07106781 -7.07106781
-0.15565903 -0.15565903
-0.23524288 -0.23524288
-0.30846569 -0.30846569
-0.37446325 -0.37446325
-0.43251645 -0.43251645
-0.48206118 -0.48206118
-0.52269499 -0.52269499
-0.55418056 -0.55418056
-0.57644583 -0.57644583
-0.58958100 -0.58958100
-0.59383228 -0.59383228
-0.58959277 -0.58959277
-0.57739060 -0.57739060
-0.55787461 -0.55787461
-0.53179797 -0.53179797
-0.5+1E-001i-0.5+1E-001i
-0.46338675 -0.46338675
-0.42291066 -0.42291066
-0.37954973 -0.37954973
-0.33428676 -0.33428676
-0.28808898 -0.28808898
-0.24188860 -0.24188860
-0.19656461 -0.19656461
-0.15292621 -0.15292621
-0.11169828 -0.11169828
-7.35090458 -7.35090458
-3.88802803 -3.88802803
-8.22017866 -8.22017866
1.8181020021.818102002
4.0152561464.015256146
5.7641898315.764189831
7.0710678117.071067811
7.9528136447.952813644
8.4362091388.436209138
8.5567788348.556778834
8.3574890328.357489032
7.8872948717.887294871
7.1995721547.199572154
6.3504729486.350472948

-0.01405147 1.2290098913
0.0011016451.6019684063
0.0030815631.9779569124
0.0022110092.3607371518
0.0012096072.7501192696
0.0005391643.1095104307
0.0001544453.4424068610
-4.78821157 3.7333565559
-0.00014408 3.9726398627
-0.00018085 4.1589672669
-0.00018515 4.2869833675
-0.00017225 4.3574570921
-0.00015079 4.3725481906
-0.00012563 4.3353711164
-9.95215967 4.2513160828
-7.39732958 4.1261751393
-4.98153053 3.9663924988
-2.74693161 3.7787143137
-7.12766907 3.5696901041
1.1144974473.3456219644
2.7351090393.1123157865
4.1523958402.8749637141
5.3705906862.6380966843
6.3935760562.4055350051
7.2242793652.1804000812
7.8646446911.9651441342
8.3162877091.7615965572
8.5821504461.5710272964
8.6697465321.3942147441
8.5969413171.2315171362
8.4016048331.0829434468
8.1567385029.4822050661
7.9923876258.2685518265
8.1240227487.1819016450
8.8830103116.2145279939
0.0001073755.3579688201
0.0001428264.6033751321
0.0002017223.9417946090
0.0002897173.3643958340
0.0004093932.8626396165
0.0005579912.4284043476
0.0007261142.0540722493
0.0008984741.7325831001
0.0010569811.4574615370
0.0011852001.2228234179
0.0012723881.0233660768
0.0013154998.5434662710
0.0013186287.1155181607
0.0012906225.9126232375
0.0012421324.9021384298

Page 5

FFTCalc
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105

5.3972454385.397245438
4.3965590384.396559038
3.4028753753.402875375
2.4669043922.466904392
1.6341826481.634182648
9.4380779369.438077936
4.2735941474.273594147
1.0803191051.080319105
0
0
1.0803191051.080319105
4.2735941474.273594147
9.4380779369.438077936
1.6341826481.634182648
2.4669043922.466904392
3.4028753753.402875375
4.3965590384.396559038
5.3972454385.397245438
6.3504729486.350472948
7.1995721547.199572154
7.8872948717.887294871
8.3574890328.357489032
8.5567788348.556778834
8.4362091388.436209138
7.9528136447.952813644
7.0710678117.071067811
5.7641898315.764189831
4.0152561464.015256146
1.8181020021.818102002
-8.22017866 -8.22017866
-3.88802803 -3.88802803
-7.35090458 -7.35090458
-0.11169828 -0.11169828
-0.15292621 -0.15292621
-0.19656461 -0.19656461
-0.24188860 -0.24188860
-0.28808898 -0.28808898
-0.33428676 -0.33428676
-0.37954973 -0.37954973
-0.42291066 -0.42291066
-0.46338675 -0.46338675
-0.5-1E-001i -0.5-1E-001i
-0.53179797 -0.53179797
-0.55787461 -0.55787461
-0.57739060 -0.57739060
-0.58959277 -0.58959277
-0.59383228 -0.59383228
-0.58958100 -0.58958100
-0.57644583 -0.57644583
-0.55418056 -0.55418056
-0.52269499 -0.52269499

0.0011832084.0555678601
0.0011219123.3481603513
0.0010639142.7585179198
0.0010127142.2682227692
0.0009701701.8614878249
0.0009370581.5248338609
0.0009135451.2467947169
0.0008995321.0176509151
0.0008948818.2919110348
0.0008995326.7450012983
0.0009135455.4777212554
0.0009370584.4414672260
0.0009701703.5956639475
0.0010127142.9065287775
0.0010639142.3460065817
0.0011219121.8908560400
0.0011832081.5218692766
0.0012421321.2232080912
0.0012906229.8184154238
0.0013186287.8707113660
0.0013154996.3013135255
0.0012723885.0385463787
0.0011852004.0239134069
0.0010569813.2097625451
0.0008984742.5573456286
0.0007261142.0352096545
0.0005579911.6178665306
0.0004093931.2846957975
0.0002897171.0190416481
0.0002017228.0747151787
0.0001428266.3916865748
0.0001073755.0543551496
8.8830103113.9928853169
8.1240227483.1512816789
7.9923876252.4847069468
8.1567385021.9573058703
8.4016048331.5404428029
8.5969413171.2112767084
8.6697465329.5161089343
8.5821504467.4696598536
8.3162877095.8583398462
7.8646446914.5907892230
7.2242793653.5945601596
6.3935760562.8122643938
5.3705906862.1984912122
4.1523958401.7173450156
2.7351090391.3404804601
1.1144974471.0455366522
-7.12766907 8.1489087348
-2.74693161 6.3466784622

Page 6

FFTCalc
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127

-0.48206118 -0.48206118
-0.43251645 -0.43251645
-0.37446325 -0.37446325
-0.30846569 -0.30846569
-0.23524288 -0.23524288
-0.15565903 -0.15565903
-7.07106781 -7.07106781
1.8489003881.848900388
0.1107282300.110728230
0.2047168880.204716888
0.2991085420.299108542
0.3925237880.392523788
0.4835745110.483574511
0.5708885540.570888554
0.6531343260.653134326
0.7290448620.729044862
0.7974408580.797440858
0.8572522200.857252220
0.9075377060.907537706
0.9475022570.947502257
0.9765116860.976511686
0.9941044070.994104407

Operations on Severity FFT

0
1
1
1
1.0027711651.002771165
2
1.0110259221.011025922
3
1.0245913201.024591320
4
1.0431892961.043189296
5
1.0664499481.066449948
6
1.0939268881.093926888
7
1.1251129391.125112939
8
1.1594550401.159455040
9
1.1963678571.196367857
10
1.2352460251.235246025
11
1.2754751951.275475195
12
1.3164420781.316442078
13
1.3575436931.357543693
14
1.3981959201.398195920
15
1.4378414461.437841446
16
1.4759570741.475957074
17
1.5120603951.512060395
18
1.5457157551.545715755
19
1.5765394481.576539448
20
1.6042040831.604204083
21
1.6284420541.628442054
22
1.6490480651.649048065
23
1.6658806701.665880670
24
1.6788628061.678862806
25
1.6879813131.687981313

-4.98153053 4.9395306616
-7.39732958 3.8416887947
-9.95215967 2.9858020754
-0.00012563 2.3190338772
-0.00015079 1.7999697184
-0.00017225 1.3961754484
-0.00018515 1.0822710611
-0.00018085 8.3841256528
-0.00014408 6.4909689233
-4.78821157 5.0222175094
0.0001544453.8834693794
0.0005391643.0011424633
0.0012096072.3179265332
0.0022110091.7892159563
0.0030815631.3803199433
0.0011016451.0642754849
-0.01405147 0
-0.07068254 0
-0.21085723 0
-0.38742858 0
-0.25844533 0
0.4700343600
Sum Ops
2
2.005542330
2.022051845
2.049182640
2.086378593
2.132899897
2.187853777
2.250225878
2.318910081
2.392735714
2.470492051
2.550950391
2.632884157
2.715087386
2.796391841
2.875682893
2.951914148
3.024120791
3.091431510
3.153078896
3.208408166
3.256884109
3.298096131
3.331761341
3.357725613
3.375962627

Page 7

FFTCalc
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75

1.6932854451.693285445
1.6948843981.694884398
1.6929439011.692943901
1.6876819161.687681916
1.6793635321.679363532
1.6682951171.668295117
1.6548178411.654817841
1.6393006611.639300661
1.6221328871.622132887
1.6037164491.603716449
1.5844579951.584457995
1.5647609471.564760947
1.5450176541.545017654
1.5256018041.525601804
1.5068612171.506861217
1.489111203 1.489111203
1.4726286211.472628621
1.4576468201.457646820
1.4443515951.444351595
1.4328783141.432878314
1.4233103241.423310324
1.4156787141.415678714
1.4099634851.409963485
1.4060960791.406096079
1.4039632221.403963222
1.4034119131.403411913
1.4042554151.404255415
1.4062800011.406280001
1.4092522451.409252245
1.4129266111.412926611
1.4170531071.417053107
1.4213847981.421384798
1.4256849851.425684985
1.4297338911.429733891
1.4333347011.433334701
1.4363188591.436318859
1.4385505211.438550521
1.4399300841.439930084
1.4403967511.440396751
1.4399300841.439930084
1.4385505211.438550521
1.4363188591.436318859
1.4333347011.433334701
1.4297338911.429733891
1.4256849851.425684985
1.4213847981.421384798
1.4170531071.417053107
1.4129266111.412926611
1.4092522451.409252245
1.4062800011.406280001

3.386570891
3.389768797
3.385887802
3.375363833
3.358727064
3.336590235
3.309635683
3.278601323
3.244265774
3.207432898
3.168915991
3.129521894
3.090035309
3.051203608
3.013722435
2.978222406
2.945257242
2.915293640
2.888703191
2.865756629
2.846620648
2.831357428
2.819926970
2.812192159
2.807926444
2.806823827
2.808510831
2.812560002
2.818504491
2.825853223
2.834106215
2.842769596
2.851369971
2.859467783
2.866669402
2.872637719
2.877101042
2.879860168
2.880793502
2.879860168
2.877101042
2.872637719
2.866669402
2.859467783
2.851369971
2.842769596
2.834106215
2.825853223
2.818504491
2.812560002

Page 8

FFTCalc
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125

1.4042554151.404255415
1.4034119131.403411913
1.4039632221.403963222
1.4060960791.406096079
1.4099634851.409963485
1.4156787141.415678714
1.4233103241.423310324
1.4328783141.432878314
1.4443515951.444351595
1.4576468201.457646820
1.4726286211.472628621
1.489111203 1.489111203
1.5068612171.506861217
1.5256018041.525601804
1.5450176541.545017654
1.5647609471.564760947
1.5844579951.584457995
1.6037164491.603716449
1.6221328871.622132887
1.6393006611.639300661
1.6548178411.654817841
1.6682951171.668295117
1.6793635321.679363532
1.6876819161.687681916
1.6929439011.692943901
1.6948843981.694884398
1.6932854451.693285445
1.6879813131.687981313
1.6788628061.678862806
1.6658806701.665880670
1.6490480651.649048065
1.6284420541.628442054
1.6042040831.604204083
1.5765394481.576539448
1.5457157551.545715755
1.5120603951.512060395
1.4759570741.475957074
1.4378414461.437841446
1.3981959201.398195920
1.3575436931.357543693
1.3164420781.316442078
1.2754751951.275475195
1.2352460251.235246025
1.1963678571.196367857
1.1594550401.159455040
1.1251129391.125112939
1.0939268881.093926888
1.0664499481.066449948
1.0431892961.043189296
1.0245913201.024591320

2.808510831
2.806823827
2.807926444
2.812192159
2.819926970
2.831357428
2.846620648
2.865756629
2.888703191
2.915293640
2.945257242
2.978222406
3.013722435
3.051203608
3.090035309
3.129521894
3.168915991
3.207432898
3.244265774
3.278601323
3.309635683
3.336590235
3.358727064
3.375363833
3.385887802
3.389768797
3.386570891
3.375962627
3.357725613
3.331761341
3.298096131
3.256884109
3.208408166
3.153078896
3.091431510
3.024120791
2.951914148
2.875682893
2.796391841
2.715087386
2.632884157
2.550950391
2.470492051
2.392735714
2.318910081
2.250225878
2.187853777
2.132899897
2.086378593
2.049182640

Page 9

FFTCalc
126
127

1.0110259221.011025922 2.022051845
1.0027711651.002771165 2.005542330

Page 10

FFTCalc

Page 11

FFTCalc

Page 12

FFTCalc

0.049068
0.098017
0.14673
0.102841

Page 13

FFTCalc

Page 14

FFTCalc

Page 15

FFTCalc

Page 16

FFTCalc

Page 17

FFTCalc

Page 18

FFTCalc

Page 19

Glenn Meyers' Comments on Fourier Methods

This spreadsheet illustrates how the Fast Fourier Transform (FFT) can be used to calculate aggregate
loss distributions. While it was designed to be an educational tool, it can be used for real life actuarial
problems. In real problems you will have to expand the size of the FFT vector, and may have to add
additional lines of insurance. Excel 5.0 can handle 1024 components in the FFT, and Excel 97 can
handle 4096 components.

The continuous Fourier Transform, as implemented in the Heckman/Meyers algorithm, can be substituted
in the steps described in the introduction for the FFT method. The theories underlying the two methods
are equivalent. However there are practical differences that may be important in day-to-day use.
Fast Fourier Transform
The claim severity distribution input must be
discretized in equal intervals.

Continuous Fourier Transform

In the Heckman/Meyers algorithm, the claim severity
distribution must also be discretized, albeit in
different way. It is approximated with a piecewise
linear cumulative distribition function. But there is
no requirement that the intervals be equally spaced.

The aggregate loss distribution output must also be

discretized in equal intervals of the same size as the
claim severity distribution input. Note that the
appropriate scale for claim severity distributions is
usually in the hundreds or thousands, while the
appropriate scale for aggregate can be in the
millions, or even hundreds of millions. If this is the
case, some compromise will have to be made.

There is no discretization in the output. Various

quantities, such as the cumulative probabilitiy,
limited expected loss and the limited second
moment are available as a function of the loss level
and the Fourier transform of aggregate loss
distribution.

The FFT method is at its best in high reinsurance

layers where the usual claim severity models are
fairly flat, and the expected claim count is small.

The continuous Fourier transform method is at its

worse in this situation. It can take a long time to
crank out those last few decimal places of accuracy.
But it will eventually get the answer.

The FFT method is at its worse when the claim

The continuous Fourier transform is at its best in
severity distributions have a high proportion of small this situation. It will calculate the results very
claims, but also a long tail, and the expected claim quickly.
count is large. This will force the discretization to
have a very large number of points, and it will take a
long time for to crank out the answer.

In summary, neither method works best in all cases. The FFT has one advantage in that it can be used
on a spreadsheet with available add-ins. This is currently not the case for the Heckman/Meyers
algorithm, but should be true by March/April 1998. See www.pipeline.com/~ggmeyers/index.htm for
updates.

I clearly have a bias in my evaluation since I am the Meyers in the Heckman/Meyers algorithm and I sell a
product implementing that algorithm. I will say that we were aware of the FFT when we made the decision
to use the continuous Fourier transform over 17 years ago. Our decision to go this way was based the
discretization considerations listed above. I tend to think that the decision on methodology should be
based more on the appropriateness of model assumptions and practical considerations rather than
adherence to a particular calculation methodology. I would want any work that I do to be judged on that
basis. I would add that there are different models underlying the copula and simulation models which are
discussed in Shaun's report. These models should be evaluated.