Category: Risk Assessment and Management

Differing definitions of risk

There are many definitions of risk that vary by specific application and situational context. The widely inconsistent and ambiguous use of the word is one of several current criticisms of the methods to manage risk.

Some of the more popular definitions:

  1. Classical/Scientific
    Expected Loss due to an uncertain event
    likelihood times impact of loss
  2. Knightean Uncertainty:
    Knight defined risk as a measurable uncertainty. According to this definition, no consequence is required for an event to be a risk.
  3. Financial Risk
  4. ISO
  5. PMI
  6. DHS
  7. COSO

1) Classical/Scientific

Risk concerns the expected value of one or more results (consequences) of one or more future events. Stated a bit differently, risk is an uncertain event that matters.
Technically, the value of those results (consequences) may be positive or negative.
However, general usage tends to focus only on potential harm that may arise from a future event.
The harm is a failure to achieve one or more objectives.

2) Knightean Uncertainty

Frank H. Knight (1885-1972) was a Professor of economics at the University of Chicago.
Knight tried to distinguish measurable uncertainty (probability distributions) from uncertainties that are not measured.
Loss is not a part of Knight’s definition.
According to Knight’s definition, someone who watching a roulette wheel at a casino without wagering any money is a risk. This definition is contrary to the notion that a risk is an uncertain event that matters.

more to come…




Are humans risk averse? Not Necessarily…

You may have been told that humans are risk averse, but that belief is no longer true.

Humans are loss averse:

a) The fight-or-flight response (also called hyperarousal, or the acute stress response) is a physiological reaction that occurs in response to a perceived harmful event, attack, or threat to survival.[1]
b) It’s in our DNA. Our distant ancestors, when faced with a fight or flight situation – for example being attacked by a tiger while attacking a deer –chose to avoid the loss and fled.  Many/most of those humans who chose to fight instead of flight did not survive to pass on their DNA.
c) Kahneman and Tversky’s Prospect Theory (for which Kahneman was awarded a Nobel prize) showed that to humans, losses loom larger than gains.  This is reflected by the slope for of the typical value curves being greater for losses than the slope for gains:

https://en.wikipedia.org/wiki/Prospect_theory

But we are not always risk averse!

We exhibit risk-aversion when taking risks

that is, when seeking gains. For example, most people would not gamble $100 or $1000 on a flip of a coin — double or nothing.

But we exhibit risk-seeking behavior when facing risks

that is when faced with losses. For example, when facing a certain loss of $100 or $1000, most people would gamble double or nothing on the flip of a coin.

Summary

Humans are loss averse, but not necessarily risk averse.

When taking risks, humans are generally risk averse.

But when facing risks, humans are generally risk seeking. We have a natural tendency to gamble that risk events will not occur rather than invest in controls to reduce the risks.

Why 5 by 5 risk maps are meaningless and perhaps even dangerous

A 5 x 5 risk maps (sometimes called heat maps) for risks that are represented by the product of likelihood and impact ratings on one to 5 scales is meaningless because multiplication of ordinal measures is mathematically meaningless. Ordinal ratings do not have interval or ratio meanings. So, for example, the difference between a 5 and 4 does not have the same meaning as the difference between a 3 and 2. More importantly, the ratio of a 4 to a 2 does not have the same meaning as the ratio between a 2 and a 1.  

Since multiplication of ordinal numbers is mathematically meaningless, the position of the points on the typical risk map resulting from the multiplication of two ordinal ordinal 1-5 axes are meaningless.  This is reflected in the rectangular regions of the typical risk map, rather than a curved (hyperbolic regions) that represent products of ratio scale measures of likelihood and impact.

Other drawbacks of traditional heat maps based on 1 to 5 estimates of likelihood and impact are:
Risks in a yellow region might actually be higher than those in a red region;
Risks in a green region might actually be higher than those in a yellow region;