Chaman L. Jain, Ph.D

Chaman L. Jain, Ph.D

I recently came across this white paper with the above title, written by a company called AGENTRICS, which I found  very strange and contradictory.

Here are four reasons why:

1. The paper claims since we cannot forecast (meaning 100% accurately), forecasting is unnecessary. The practice of forecasting has caused and will continue to cause too much inventory and thus fails to provide the degree of availability that the market demands. It is a strange belief for a number of reasons: a) We need forecasts even if they are less than perfect, because every plan is based on them. Of course, the more accurate the better. b) forecasts do improve to some degree if we work on them.

2. The paper emphasizes only the modeling aspect of forecasting. Models are only one part of forecasting, the other part is the process—the Consensus Forecasting, S&OP and CPFR. I have seen study after study showing how much improvement they made in forecasts and inventory by using such processes. Furthermore, the fact 79% of the companies use Consensus Forecasting, 70% of them S&OP and 36% of them CPFR, they must have found something to justify. (These numbers are based on the survey done by the Institute of Business Forecasting and Planning—IBF) In fact, I don’t know any company where forecasts deteriorated and inventory levels increased by using these processes.

3. The statement of “More inventory equals less availability” does not make sense. This implies that demand planners don’t know what they are doing. Furthermore, as I said before, forecasts do improve when you work on them. With that you not only get the right inventory, but also manage to reduce it.

4. I also found the approach echoed in the paper self contradictory. Its first step to determine inventory is to determine display stock levels (stock the retailer must hold to serve a customer best), using marketing considerations. How can one determine such levels without forecasting? It further states that “each link in the supply chain should strive to produce only what the downstream link consumes.” It is great, but again how can one determine what downstream link consumes (which I believe is sales to end consumers) without forecasting? Of course, one does not need to forecast if the lead-time or cycle time is zero, which is often not the case.

Your thoughts, comments, similar experiences are welcome.

Chaman L. Jain, Ph.D
Professor, St. Johns University – New York
Chief Editor, IBF’s Journal of Business Forecasting