There is an important point to note in using "Deviation" formulas in Sage MAS 500 to calculate Safety Stock. The "Deviations" here are the differences between your "Forecast Demand" and your "Actual Demand," and your "Forecast Demand" may include data that could skew the results (e.g., Demand Adjustments). Furthermore, any differences where "Forecast Demand" exceeds "Actual Demand" in any given period are omitted from the averaging process, so negative variations are not considered at all. (I assume that Sage took this approach because they do not use squared values -- as the standard deviation formula would -- which would eliminate negatives, and also, perhaps they deemed that if your Forecast exceeds your Actual Demand, you have covered your requirements.)

Note that a statistical "Standard Deviation" for a sample or population (in this case, one would use the sample formula) is calculated on the variance from the statistical mean of the sample data, so the MAS 500 "Deviation" is not a "Standard Deviation" in "Actual Demand." The result is that setting your "Multiplier" to the value of 2.0000 does not necessarily cover 2 standard deviations in "Actual Demand."

I set up a simulation to compare the results between the MAS 500 "Deviation" formula to calculate Safety Stock and a true standard deviation in "Actual Demand." In my particular simulation (with a look-back over 12 Periods), I found that the MAS 500 formula was slightly more conservative -- tending to estimate below the value suggested by the standard deviation in "Actual Demand" more often than not.

In general I would say that the MAS 500 formula approximates coverage of the Multiplier number of standard deviations in Actual Demand, but it is not precise. Much would depend on the set of data against which the calculations are run and this, of course, means that the results will vary for each Item in a MAS 500 company's inventory.

It may have been helpful if MAS 500 had employed standard deviations against Actual Demand so that, for example, when one selected 2.000 for a multiplier, one could assert that you are setting a 95.45% customer service level on Items that employ that particular Safety Stock Formula (in accordance with the following chart).

CONFIDENCE INTERVAL FACTORS for Various Multiples of Standard Deviations:

zσ = Percentage

=========================

1σ = 68.27%

1.645σ = 90%

1.960σ = 95%

2σ = 95.450%

2.576σ = 99%

3σ = 99.7300%

3.2906σ = 99.9%

4σ = 99.993666%

5σ = 99.99994267%

6σ = 99.9999998027%

7σ = 99.9999999997440%

=========================

I hope this is of some help.