The special role of common and expected uncertainty for management

After what we recently went through, the area of risk management gets naturally more attention.  The question is centered on what an organization can do to face very big risks; many of them come from outside the organization.

What about the known small risks managers face all the time?

I suggest distinguishing between two different types of uncertainty/risks, which call for distinct methods of handling.  One is what we usually refer to as risks, meaning possible occurrences that generate big damage.  This kind of uncertain event is viewed as something we strive to avoid, and when we are unable to we try to minimize the damage.

The other type, which I call ‘common and expected uncertainty’, is simply everything we cannot accurately predict, but we know well the reasonable range of possible results.   The various results are sometimes positive and sometimes negative, but not to the degree that one such event would destroy the organization.  The emphasis on ‘common and expected’ is that none of the possible actual outcomes should come as a surprise.  While the actual outcome frequently causes some damage, true significant damage could come only from the accumulation of many such uncertain outcomes, and this is usually rare. So, losing one bid might not be disastrous, but losing ten in a row might be.

This article claims that there is a basic difference in handling the two types of uncertainty.  While both impact decisions and both call for protective mechanisms, the objective of those mechanisms and the practice of managing them is quite different.

The economic impact of ‘common and expected uncertainty’ is by far underappreciated by most decision-makers.

Hence the value of improving the method for dealing with ‘common and expected uncertainty’ is much higher than expected.

A big risk is something to be prepared for, but the means have to be carefully evaluated.  For instance, dealing with the risk of earthquakes involves economic considerations.  It is definitely required to apply standards of safety in the construction of buildings, roads, and bridges, but the costs, and the impact on the lead time, have to be considered.  Another common protection against the damage of earthquakes is given by insurance, which again raises the issue of financial implications.

Some risks are very hard to prepare for.  What could have the airlines do to prepare for the Coronavirus other than carrying enough cash reserves?  Airlines invest a lot in preventing fatal accidents and have procedures to deal with such events.  But, there are risks for which preparations, or insurance, don’t really help.  Every time I go on a flight I’m aware that there is a certain risk for which I have no meaningful protection.  So, I accept the risk and just hope that it’ll never occur.

Ignoring common and expected uncertainty is not reasonable!  However, it is practically ignored by too many organizations, which pretend they are able to predict the future accurately and base their planning on it.  This illogical behavior creates an edge for organizations with better capability to deal with common and expected uncertainty and generate very high business value based on reliable and fast service to customers.  That capability leads also to built-in flexibility that quickly adapts to the changing tastes of the market.  Isn’t this a basic capability for facing the new market behavior resulting from the Coronavirus crisis?  The burst of the epidemic changed the common and expected uncertainty, but by now we should be used to its new behavior, making it more “expected” than it was in March 2020.

Failing to deal with the common and expected uncertainty is especially noted in supply chain management.

For instance, a past CEO of a supermarket chain admitted to me that at any given time the rate of shortages on the shelf is, at least, 15%.  The damage of 15% shortages is definitely significant, as it means that many of the customers, coming to a supermarket store with a list of items to buy, don’t go home with the full list fulfilled.  When this is an ongoing situation then some customers might decide to move to another store.  As long as all the chains suffer from the same level of shortages this move of customers is not so damaging.  But, if a specific chain would significantly reduce the shortages it would steal customers from the other chains.

Given the common and expected uncertainty in both the demand and the supply is there a better way to manage the supply chain in a much more reliable way?

To establish a superior way the basic flaw(s) in the current practice should be clearly verbalized.

The current flawed managerial use of forecasts points to an even deeper core problem.  Mathematically a forecast is a stochastic function exposed to significant variability and thus should be described by a minimum of two parameters: an average and a measure of the spread around that average.  The norm for forecasting is using the forecast itself as an average and the forecasting error that points to an average absolute deviation from the average.  The forecasting error, like the forecast itself, is deduced from the past results.

The use of just ONE number forecasts in most management reports demonstrates how managers pretend “knowing” what the future should be, ignoring the expected spread around the average. When the forecast is not achieved it is the fault of employees who failed, and this is many times a distortion of what truly happened.  Once the employees learn the lesson they know to maneuver the forecast to secure their performance.  The organization loses from that behavior.

When the MRP algorithm in the ERP software takes the forecast and calculates the required materials the organization doesn’t really get what might be needed!  Safety stock without reference to the forecasting errors is too arbitrary to fix the situation.

A decision-maker viewing an uncertain situation needs to have two different estimations in order to make a reasonable decision:

1. What could be the situation in a reasonable best-case scenario?
2. What might be the reasonable worst-case situation?

The way to handle uncertainty is to forecast a reasonable range of what we try to predict.  Forecasting sales is the most common way to determine what Operations should be prepared to do.  Other cases where reasonable ranges should be used include considering the time to complete a project or just a manufacturing order.  The need for the range is to support the promise for completion, leaving also room for delays due to common and expected uncertainty.  The budget for a project, or a function within the organization, is an uncertain variable that should be handled by predicting a reasonable range.

The size of the range provides the option of using a buffer, the protection mechanism against common and expected uncertainty.  While one size of the reasonable range expresses a minimum assessment, where an actual result of less than that number seems “unreasonable” based on what we know.  The other side expresses the maximum reasonable assessment.  If you choose to protect from the possible reality of being close to the maximum assessment, like when you strive to prevent any shortage, then you need to tolerate too high stock, time, money, or any other entity that constitutes a buffer.  In cases where the cost of the buffer is high, then the financial consequences of losing sales due to shortages have to be considered.

One truly critical variable in the supply chain is the forecasting horizon.  Cost considerations can push planners to use too long horizons, which increases the level of uncertainty in an exponential way.  When it comes to managing the supply chain, which is all about managing the common and expected uncertainty, the horizon of the demand forecast should reflect the reliable supply time and not beyond that value.

Buffer Management is an unbelievably important concept, developed by Dr. Goldratt, which is invaluable for managing common and expected uncertainty during the execution phase, and also helps to identify emerging situations where the buffers, based on the predicted reasonable ranges, fail to function properly.  The idea is simple:  as long you are using a buffer against a stream of fluctuations, the state of the buffer tells you the real current level of urgency of the particular item, order, or even the state of cash.  Buffer management uses the well-known code of Green, Yellow, and Red to radiate what is more urgent, and this provides the best behavior model for dealing with common and expected uncertainty.

The big obstacle for becoming much more effective is to recognize the impact of both risks and common and expected uncertainty.  The difficulty in recognizing the obvious is how can the boss know when the subordinate does a good job?  The inherent uncertainty is an easy explanation for any failure to meet targets.  Problem is:  shutting our eyes does not help to improve the situation.  So, it is the need for managers to constantly measure the performance of every employee and demand accountability for results, for which the employees have just partial impact, is the ultimate cause for most managers to ignore common and expected uncertainty.