Está en la página 1de 8

Probability Impact Grid analysis

In fact, if your matrix is not 5 x 5 you can still put in as much as will fit and see what it tells you.

Probability Impact Grid analysis


Probabilty Impact grids are very common in risk management/internal control and it is also common to assign a summary risk score by combining the 'probability' and 'impact' ratings. This is often done by multiplying (occasionally adding) row and column indices. What most people probably think that gives them is a sort of expected value for the risk e.g. 50% likely to happen and will cost 1,000 units if it does so the expected value is 0.50 x 1,000 = 500. However, if you have defined the ranges for each level of probability and impact so that they are not the same size you will probably find that multiplying the indices does not give you the expected value you imagined. For example if your ranges grow by a multiple each time then you would need to add the indices to get the right ranking. This kind of mathematical blunder means that risks are ranked in the wrong order, and some risks are included/excluded for upwards escalation when they should not be. If you want to see if any of this applies to your risk rating design and you've used a 5 x 5 matrix then use the 'analysis' worksheet. In fact, if your matrix is not 5 x 5 you can still put in as much as will fit and see what it tells you. Most people will find that it is surprisingly difficult to engineer a satisfactory risk summary score using indices. It is better to assign an approximate 'mid point' to each probability level and each impact level and use that instead. When choosing a 'mid' point do not take the average of the upper and lower limits of the level unless the levels are all of equal size. In the analysis spreadsheet I've used the ratio of the range's size and that of an adjacent range to fix a 'mid' point. This roughly agrees with more sophisticated calculations done by fitting beta distributions to actual risk register data.

Probabilty Impact Grid mathematics - does yours work as you expect?


Expected value using lowest corner of each cell Lower Upper Prob 0.1 1 0.01 0.1 0.001 0.01 1E-04 0.001 0 1E-04 Index 5 4 3 2 1 Index Upper Lower Imp 0 0 0 0 0 1 5 0 0.5 0.05 0.005 0.0005 0 2 50 5 5 0.5 0.05 0.005 0 3 500 50 50 5 0.5 0.05 0 4 5000 500 500 50 5 0.5 0 5 50000 5000 Rank order 17 17 17 17 17 7 11 14 16 17

Yellow areas are editable. Coloured grids below show where the ranking of risks using expected values differs from that usin

Low EV rank - multiplied indices rank 1 -2 -1 -1 -1 0 -4 0 0 -6 -2 0 -8 -6 -4

0 0 0 -1 -1

0 0 -1 -2 1

High EV rank - multiplied indices rank -5 -2 -2 -1 -1 2 0 2 0 0

Low EV rank - added indices rank 6 0 0 1 0 0 -3 -2 0 -6 -4 -2 -8 -6 -3

0 0 0 0 1

0 0 0 0 6

High EV rank - added indices rank 0 0 0 0 0 0 0 0 0 0

Expected value using highest corner of each cell Lower Upper Prob 0.1 1 0.01 0.1 0.001 0.01 1E-04 0.001 0 1E-04 Index 5 4 3 2 1 Index Upper Lower 5 0.5 0.05 0.005 0.0005 1 5 0 Imp 50 5 0.5 0.05 0.005 2 50 5 500 50 5 0.5 0.05 3 500 50 5000 500 50 5 0.5 4 5000 500 50000 5000 500 50 5 5 50000 5000

Rank order 11 16 20 23 25 7 11 16 20 23

Expected value using 'mid' points of each cell Prob Mid 0.182 0.018 0.002 2E-04 1E-05 Index 5 4 3 2 1 Index Mid 0.090909 0.009091 0.000909 0.000095 0.000005 1 0.5 1.727273 0.172727 0.017273 0.001805 0.000095 2 9.5 16.52893 1.652893 0.165289 0.017273 0.000909 3 90.90909 165.2893 16.52893 1.652893 0.172727 0.009091 4 909.0909 1652.893 165.2893 16.52893 1.727273 0.090909 5 9090.909

Rank order 15 18 21 23 25 8 11 16 20 23

Imp

Score by multiplying index numbers Lower Upper Prob 0.1 1 0.01 0.1 0.001 0.01 1E-04 0.001 0 1E-04 Index 5 4 3 2 1 Index Upper Lower Imp 5 4 3 2 1 1 5 0 10 8 6 4 2 2 50 5 15 12 9 6 3 3 500 50 20 16 12 8 4 4 5000 500 25 20 15 10 5 5 50000 5000

Rank order 16 18 21 23 25 9 12 14 18 23

Score by adding index numbers Lower Upper Prob 0.1 1 0.01 0.1 0.001 0.01 1E-04 0.001 0 1E-04 Index 5 4 3 2 1 Index Upper Lower Imp 6 5 4 3 2 1 5 0 7 6 5 4 3 2 50 5 8 7 6 5 4 3 500 50 9 8 7 6 5 4 5000 500 10 9 8 7 6 5 50000 5000

Rank order 11 16 20 23 25 7 11 16 20 23

isks using expected values differs from that using row and column indices added or multiplied.

4 7 11 14 17

2 4 7 11 17

1 2 4 7 17

High EV rank - multiplied indices rank -1 0 0 0 0 0 0 0 -1 2 -1 -2 -1 -2 -5

High EV rank - added indices rank 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

4 7 11 16 20

2 4 7 11 16

1 2 4 7 11

5 9 13 16 21

3 5 9 11 18

1 2 4 7 14

5 7 11 14 21

2 4 7 12 18

1 2 5 9 16

4 7 11 16 20

2 4 7 11 16

1 2 4 7 11

Mid EV rank - multiplied indices rank -1 -1 0 1 0 0 -1 2 1 0 0 2 2 2 -1 0 2 2 -1 -2 0 0 0 0 -2

Mid EV rank - added indices rank 4 1 1 1 0 2 0 2 1 0 1 0 2 2 0 0 0 0 0 0 0 0 1 2 3