Risk management - Cost model
Cost model
We have seen before [see Simple estimating of risk] the use of a simple 3 point estimate and extended this to refine the estimates to push out the boundaries.
The remainder of these examples will just employ the simple 3 point estimate with a straight line relationship between the values.
These are also known as ‘triangular distributions’.
If we had to guess the COMPLETION DATE and the TOTAL COST of a project with less than one hour of consideration it would be difficult, to say the least.
Even if the project team were made up of highly experienced personnel, the best in their field, it would be very difficult.
If each project team member had a guess then there would be a wide range of opinions.
It becomes obvious to break a project down into smaller parts, where focussed expertise can better predict the COMPLETION DATE and the TOTAL COST and then to ‘add’ them, in some fashion, to get a reasonable summary of the total picture.
In order to assess the risk for each of the above tasks (for COST) we could predict a range of potential values of cost. For example:
MINIMUM value:
LIKEKLY value:
MAXIMUM value:
That is, a 3 point estimate.
We would get:
| Cost | ||||
|---|---|---|---|---|
| Task | Min | Likely | Max | Distribution |
| A | 2 | 4 | 5 | 3.7 |
| B | 1 | 3 | 6 | 3.3 |
| C | 2 | 3 | 4 | 3 |
| D | 5 | 7 | 9 | 7 |
| E | 3 | 6 | 7 | 5.3 |
| F | 1 | 6 | 11 | 6 |
| G | 3 | 4 | 5 | 4 |
| H | 8 | 10 | 12 | 10 |
| Total | 25 | 43 | 59 | 41.3 |
On this basis we could assess each of the tasks.
We could then add up all of the MINIMA, LIKELY and MAXIMA costs.
This would give a simplistic indication of the total range.
In the above example we end up with:
MINIMUM total: 25
LIKEKLY total: 43
MAXIMUM total: 59
This equates to the earlier ‘full positive correlation’ [see The Risk Management process - Evaluation - positive correlation].
The final column is the DISTRIBUTION or the sum of the others divided by 3 e.g. 2+4+5 =11. divided by 3 = 3.7.
Each of these are a simple ‘mean’ which we know previously to be additive.
However, this is not an entirely useful model as the chances of getting the MINIMUM or MAXIMUM values is not the same as for the ‘LIKELY’ value. That is the ‘likelihoods’ are different.
This not only applies to the individual range of costs but also the individual tasks.
For example, the value of ‘A’ could be towards the HIGH end while ‘B’ could be towards the LOW end.
It is unlikely that they will all be LOW (total 25) or all be HIGH (total 59) [see Cost model part 2].