This is a post answering some well thought questions posed by IMR reader Jack Vinson regarding a previous post of mine, What to do About Seasonality. Jack Vinson's questions and comments are quoted in bold text.
"How accurate are these forecasts?"
First of all, the degree of accuracy of forecasts can vary. Essentially, the more accurate the forecasts are, the fewer inventory you can get away with holding. A major reason that inventory needs to be held is because of such uncertainty in demand.
Inventory is essentially a safety net against varying degrees of stock out probabilities. In other words, there is always a chance that you COULD run out. Let's say you have a demand forecast that looks to be about an average of 100 units a day with a standard deviation of 10 units. So, if you want, you could produce 100 a day, but on the days when the demand deviates above 100 units, you'll run out and get back on track when you have a day under 100.
Inventory provides you with additional protection against a stock out even on those days when you deviate
in the positive direction. Depending on how safe you want to be, you
could hold more or less inventory, but even if you hold 10,000 units, remember
that there is always some percentage, even if it is very small that you could
have a REALLY big day and sell out. (Another option is to keep capacity slack in order
to fulfill big days).
How far into the future are these forecasts accurate?
Demand forecasts can constantly change. For example, perhaps you sell parts required for hybrid cars. Your demand projections may be set at 1,000 units a day with a standard deviation of 200 units, but if gas prices soar to $8 a gallon, you can appreciate that your demand would be likely to increase. This increase in demand could be long term and the forecasts resulting from it may be very inaccurate for months to come. As I hope I have demonstrated, demand can change very quickly.
How much work goes into putting them [forecasts] together?
Typically, a lot. I won't go into details, but there are a ton of variables that go into demand forecasts and this is something a marketing specialist would be more equipped to help you with. Based on my limited work with deriving a demand, I can tel you that it can be very difficult.
However, if you're lucky and demand has been steady for a long period of time and looks to be staying steady, it can be fairly simple if decent demand records have been kept.
"Let's look at the opposite end of the spectrum. In what situations would it be possible to manage without a detailed forecast?"
In response to your next set of questions, my answer is that without a detailed demand forecast, proper inventory levels cannot be accurately set. If you have no clue what your demand is going to be, one of three situations will befall your firm:
1) You under produce and are left with a large backorder log/lost business to competitors/both.
2) You overproduce and are left with more inventory than you know what to do with. Hopefully the holding costs are low and even more hopefully, your product won't become obsolete before you can sell it all.
3) You get lucky and you produce exactly what you need when you need it. Without a detailed forecast, luck is exactly what you are going to need in order to achieve this.
"What operational policies would need to be in place?"
The operational policies without a detailed forecast would likely be driven by the holding costs and the backorder costs. If you have low holding costs, and high backorder costs than it makes sense to overproduce. Worse case scenario you hold onto extra inventory. This might be considered a small price to pay to prevent the potential loss of business from the backorders. Also, if your company has a guaranteed on-time delivery clause with whom you supply units to, backorders may be a very costly option.
If however, backorders result in a minimal cost and holding costs are through the roof, then under producing is likely a better option.
"The goal, of course is to ensure the right products are in the right place at the right time."
You are right, this is the "goal" However, there are certainly exceptions to this. I believe I have explained why, but just to reiterate, sometimes the only way to ensure that the products can be delivered at the right time is to over produce, providing your firm a safety net of inventory. In the event that holding costs (which would be incurred in the event that demand is LESS than you project it to be) are very high, it is much wiser to run the risk of under producing. Yes, this will have a backorder cost. This cost needs to be compared to the holding costs and a decision needs to be made, in respect to other company policies, regarding whether or not the risk of over producing, or under producing should be made.
I hope that answers your questions, Jack, and thanks for reading.
Did you enjoy this post?
I am having a little fun here, since I have some specific things in mind. Hope you don't mind my continued questions...
What elements go into the determining how much inventory to hold at each location in a supply chain (consumption points, distribution centers, regional warehouses and central warehouse)? Which pieces of the puzzle are the most certain? Which have the most uncertainty?
Once you have set your inventory levels, why do you need to continue forecasting? Isn't it easier replace what has been consumed? If consumption increases, then you replace more. If it decreases, then you replace less.
Posted by: Jack Vinson | Sep 15, 2005 12:14:40 PM
The comments to this entry are closed.