Forecasting

Forecasting accurately how much of an item you are going to need at some point in the future can be very important to a company's profits.  If the forecast for an item is too high, too much of it may be produced and then have to be kept in inventory until it is consumed.  This ties up capital in the inventory investment, and costs you further money in carrying costs.  But if the forecast turns out to be too low, and customer demand exceeds the amount of the item you have produced, you can again lose money.

No one, so far, has come up with a sure‑fire technique for forecasting inventory usage.  Some of the techniques which have been developed are almost beyond the grasp of most college graduates (unless they majored in math), and these very sophisticated systems of prediction usually require more accuracy in the data input to them than can be easily obtained.

Elliott's Inventory Management package uses an easy to understand method of forecasting the next period's usage of an item based on the item's selling history.  The technique has the rather scientific sounding name of exponential smoothing, but is basically pretty simple.  It uses a weighted moving average to calculate next period's expected usage level for an item.  Let us take a few examples to illustrate how this calculation works.  We will assume that the forecasting period is a month.  This first example will use six month's worth of sales history.

Figure 1  

     65   50   55   70   50  40       Average Usage = 333 = 55

                                                       6

Figure 2

     50   55   70   50  40  155       Average Usage = 420 = 70

                                                       6

In Figure 1, we have a diagram showing the usage of an item for six prior months.  The average usage for these six months is 55.  Then in Figure 2, we have dropped the usage for month number 1, and added the usage for the current month, month number 7, to the end, and recalculated a new average usage for the item.  This new average usage is our forecast for next month.  As you can see, the suddenly higher usage in month 7 caused the average usage to increase quite a bit over the previous average usage.

 Here is another example of this technique using a longer period of sales history, 12 months.

Figure 3

     65  50  55  70  50  40  35  60  70  50  65   50

               Average Usage = 660 = 55

                               12

Figure 4 

     50  55  70  50  40  35  60  70  50  65  50  155

               Average Usage = 750 = 63

                               12

In Figure 3, we have 12 months of sales history, again with an average usage of 55.  In Figure 4, the usage for month 1 has been dropped, and the usage for the month just ended has been added.  Even though the usage for the most recent month jumped to 155, as in the previous example, the new average usage is only 63.  As you use a longer period of sales history, a sudden increase or decrease in any one month will be dampened more than it is dampened when you only use a few months of history.  In other words, the new forecast is more responsive to sudden increases and decreases in usage when only a few months of history are used, whereas, a more stable forecast which does not fluctuate as much is obtained by using a longer period of sales history.

 As it turns out, you do not actually have to have this sales history available to the program in order to do this calculation of the new forecast.  All you have to do is specify how much weight to place on the usage figures for the prior period.  This can be done using this table.

            Periods of             Usage Weighing

          Sales History                 Factor    

                3                        .50

                6                        .29

                9                        .20

               12                        .15

               18                        .11

As you can see from this table, as you use a larger number of periods in sales history, the importance placed on this period's usage decreases.

You specify the usage-weighing factor for each inventory item as part of Item File.

For high volume items whose sales can be very volatile, you may want to use a short period of sales history, so that new forecasts are very responsive to the current demand for the product.  For other items, you may want to use a long sales history period, so that forecasts do not fluctuate as much as sales fluctuate.

This new forecast is calculated whenever you run the Recalculate Reorder Fields application.  Further detail on the actual calculations can be found in the Recalculate Reorder Fields section of this manual.

Safety Stock

Safety Stock is the quantity of an item to be kept on‑hand in case of sudden demand. It serves to cushion your inventory against increases beyond your ability to meet an unanticipated demand for the item. Initially, you should decide how much safety stock you should keep on hand for each item.  Later on, each time you run the Recalculate Reorder Fields application, the optimum value of the Safety Stock field will be recalculated.

This calculation is based on how far the forecast is deviating from the actual usage.  If the forecast is consistently running lower than the actual usage (i.e., usage is exceeding the forecast) the amount of safety stock to keep on‑hand will increase.

ABC Analysis

 An ABC Analysis can be a useful tool for categorizing your inventory items.  It is based on the general principle that a small percentage (about 15‑20%) of your inventory items will be found to produce a large percentage (about 70‑80%) of your income (Category A items), a larger portion of your inventory items (about 30‑40%) will be found to produce about 15‑20% of your income, and that the remaining 40‑60% of your inventory items will account for only the remaining 5‑10% of income.

     Usage in Dollars         Inventory Items     Class

          70‑80%                   15‑20%           A

          15‑20%                   30‑40%           B

           5‑10%                   40‑60%           C

You may find it of great benefit to tightly control the inventory levels of the relatively few Class A items, since these account for a higher percentage of activity and bring in a higher proportion of your income.  On the other hand, those items which are relatively low‑activity items can be managed by a looser Inventory Management package on a more casual basis.

When you first set up the Inventory Item File, you may not have a breakdown of these categories available.  If not, you can leave the ABC Analysis code (called the Inventory Class code on screen 3 of the Item File) blank.  Then later, after some sales history has accumulated for your inventory items, run the Print ABC Analysis Report application.  After the report has been run, and you are satisfied with the results obtained, you may have the program go through the inventory items, setting their inventory class.

Stocked vs. Non‑Stocked, Controlled vs. Non‑Controlled

   There are two fields in the Inventory Item record, which will be discussed here, the Stocked Flag and the Controlled Flag.

The stocked flag can have one of two values, either Y = Stocked, or N = Non‑Stocked.

A stocked item is one, which you plan to keep on the shelf either for sales to customers or for use in your manufacturing plan.  A non‑stocked item is one which is never kept on the shelf as a finished end item, ready for sale or use in the plant, even though its components may be kept on stock at all times.  A non‑stocked item may be manufactured or assembled to customer order, and so is not kept on the shelf itself.

The controlled flag can have one of two values, either Y = Controlled or N = Non‑Controlled.

 A controlled item has its quantity in inventory allocated when a customer order or shop order is issued which requires a quantity of the item, and this quantity of the item is de‑allocated when the customer order is shipped or the materials are issued to the shop.  This allocation and de‑allocation does not occur for a non‑controlled item.

Before covering how these fields are used by the other Manufacturing packages, let us take a few examples of items, which illustrate the possible combinations, which can occur using these two fields.  We will use a company, which manufactures bicycles for the example: 

1.    A stocked and controlled item. This would be an item, which is kept in stock and which is allocated when ordered, and de‑allocated when used.  An example of this would be the handlebars for the bicycle. It may be the company's policy to always keep these handlebars in stock, ready for issue when needed for assembly. But it is also important to know how much of the quantity on‑hand has already been allocated to orders, which currently exist.  Thus the handlebars are a controlled item as well.

2.    A stocked but non‑controlled item.  This would be an item which is always kept in inventory, but which is not allocated or de‑allocated by the processing of orders.  An example of this might be the nuts used for holding parts of the bicycle together.  These nuts are usually made available in boxes in appropriate areas of the shop, and they are used as needed.  They are replenished when a visual review or a two‑bin system shows that there is a need.

3.    A non‑stocked but controlled item.  This item is not kept in stock for regular orders but instead is purchased or manufactured for a particular customer order.  But once it is made, it is definitely controlled.  An example of this might be a particular seat assembly that is made for one particular customer.  Once it is made, you definitely want to have its use controlled.

4.    A non‑stocked and non‑controlled item.  This might be an item, which only exists as a temporary sub‑assembly at some point in the assembly procedure, such as a particular gear assembly.  This gear assembly might have engineering drawings associated with it, and the company may want to be able to determine how many of them have been made, even though the item never goes into stock and is not allocated or de‑allocated.  This type of item is often referred to as a phantom subassembly.

An understanding of these terms can be important if you plan to use the Customer Order Processing package, or any of the other Elliott packages which use Bill of Material Processor.  For example, when a customer orders a part which is non‑stocked but controlled, and for which a Bill of Material exists, the item itself is allocated.  The program then explodes through the Bill of Material and allocates those components, which are stocked, if the components are also controlled.