Decision-making under uncertainty

This is an excerpt of the original article. It was written for the March-April 2021 edition of Supply Chain Management Review. The full article is available to current subscribers.

March-April 2021

Last night, my wife and I shared a socially distanced bonfire with a few friends. One was a retired physician who is spearheading the vaccination effort in the small New Hampshire city where I live. New Hampshire has had its challenges getting needles into arms like everywhere else, but it seems as if we’re breaking through the log jam. For example, between week 1 and week 3, they’ve tripled the number of people they can vaccinate in a day, and they’ve expanded from five days a week to seven days a week. At least for now, there has not been a shortage of vaccines. I know there is a long way to go, but you can feel it picking up speed. Call me…

Browse this issue archive.

Already a subscriber? Access full edition now.

Need Help?

Contact customer service

847-559-7581   More options

Not a subscriber? Start your magazine subscription.

Seems like everything these days is COVID-19. Recently the editor of the Journal of Business Forecasting asked me to write an article on demand planning under uncertainty.* Initially it sounded redundant. I asked: “Doesn’t all demand planning (as well as demand forecasting) always deal with uncertainty? After all, customers are fickle.”

Moreover, I have been teaching introductory business analytics in which I cover decision-making under risk vis-a-vis uncertainty. So, did the editor also mean risk in addition to or separate from uncertainty? This Insights column is a slightly revised version of the article I wrote clarifying the terms risk versus uncertainty. It focuses mainly on the latter because most formal decision-making—dealing with “apparent” randomness—is statistically- and probability-based.

A textbook view of uncertainty and risk

I teach a quantitative undergraduate business-school class each semester at the University of Massachusetts in Lowell. The course I have been teaching of late is called “Introduction to Business Analytics,” a required course for undergraduate business majors. The course textbook is “Quantitative Analysis for Management” by Barry Render, Ralph M. Stair, Jr., and Michael E. Hanna. Chapter three deals with decision analysis and talks about three types of decision-making environments.

1. Decision making under certainty. “A decision-making environment in which future outcomes or states of nature are known.” This one may seem like demand-Nirvana to forecasters and supply-chain (SC) planners. However, it is not, for two reasons. The first is that if everything is known, resulting in 100% demand forecasting and planning accuracy, we would find our skills surplus to requirements. The second is that if demand is known, it might be so because the marketing and sales functions are not promoting nor developing significant new products; or their promotions are just plain ineffectual. Regarding the second point, more than likely their company’s future is not bright, and might indeed eventually die due to successful competitive threats.
2. Decision making under uncertainty. “A decision-making environment in which several outcomes or states of nature may occur. The probabilities of these outcomes, however, are not known.” Knowing what might happen in these environments may sound comforting to a planner, but not too much. Just knowing that a known set of pandemics or earthquakes will occur doesn’t help us predict when they will happen, nor can we predict their severity. A new product launch that just resembles other product launches is only slightly more comforting. Moreover, comparing COVID-19 to arguably the closest thing to it, the flu, hasn’t helped the world successfully manage the outbreak.
3. Decision making under risk. “A decision-making environment in which several outcomes or states of nature may occur as a result of a decision or alternative. The probabilities of the outcomes or states of nature are known.” Most demand forecasting and supply chain planning practices are meant for these types of environments. Because historical data is available, statistical methods can be used to estimate various probabilities of the states of nature and statistical-based inventory management practices (for example) are well known, for example, for setting safety-stock levels.

SC
MR