What constitutes a good decision? This appears to be a relatively straightforward question, and most people I work with come up with an answer to this question without any difficulty. They say a good decision is one that results in a good outcome.
In one sense, this is perfectly true. Decisions are generally defined as making a choice between different options. I decide what I order at a restaurant from different menu items; I decide who to hire among different candidates; I decide which strategy option to execute given my competitive environment. A good decision is when I choose the right menu item, hire the right candidate or pick the right strategy. In all of these instances, ‘right’ simply means the one that leads to the best outcome.
But think about this more carefully. When I choose an item from a restaurant, there can be different elements, which can result in a more difficult, or easy, decision. If I am in a restaurant where I know all the dishes will be of the same quality and I know what I want, then my choice is easy. However, what if I am unsure of the quality of the different dishes? Also, what if I am unsure of what I want to eat? These are two elements that can make the choice more difficult.
The same elements make hiring the ‘right’ candidate a difficult choice. I am not sure of the future performance of potential employees, or whether their skillset will be what I actually require going forward. Choosing the right strategy also involves a large degree of uncertainty. How will my competitors respond? What will the demand/supply/price/tax regime/competitor response in the future be? Will my choice of strategy today bind me to a course of action that I might end up not wanting to pursue?
If we reward or punish risky decisions based only on their outcomes, we give people an incentive to avoid making decisions
To make these types of choices easier, we often seek more information. We read restaurant reviews, carefully examine the qualifications and interview the candidates and conduct extensive strategic analysis.
However, even after we do all these things, residual uncertainty often remains. Add into the mix our different appetites for risk. Do we go after the safe, mediocre candidate, or take a chance on a potential genius? We are forced to deal with risk and uncertainty, which means we may make a great choice based on the information we have, but it could still lead to a bad outcome.
Or we could have flawed analysis, flawed assessments of risks/rewards, flawed evaluation of our future needs and desires and still make a decision that led to a good outcome. To put it another way, we can make good a decision with a bad outcome and we can make bad decisions with good outcomes. I would prefer the latter outcome, but I would not want to punish the decision-maker in the first instance and reward the decision-maker in the second.
In fact, if we begin to reward or punish uncertain and risky decisions based only on their outcomes, we give people an incentive to try to avoid making decisions. After all, I might make a good decision and be punished by a bad outcome. So, better to ask the boss what to do and avoid responsibility! Form a committee and spread responsibility! This does not lead to the agile, quick-response businesses we need.
To make matters worse, we fall prey to several predictable errors when making decisions under uncertainty. Our appetite for risk can be shaped by the way choices are presented. We are consistently overconfident in our own abilities to understand the past and to therefore predict the future. We consistently attribute too much causation to fixed effects and too little to random ones. We consistently ignore base-rate information when evaluating the future performance of individuals or individual projects. We have little intuitive understanding of the power of regression towards mean performance. Our confidence in causal explanations often has very little to do with the evidence supporting it.
However, this is not because we do not try hard enough, or are not educated enough, or are too careless. Rather, and somewhat more disturbing because it cannot be fixed easily, it is because of the architecture of our minds. The way our brains are designed to think creates illusions – visual, audible and cognitive – that we cannot avoid. That is the bad news. The good news is that individuals and organisations can take steps to avoid many of these predictable errors and mitigate bad outcomes if and when they happen. Creating better decision-making processes is the key to avoiding these traps and is something we work together with different organisations to do.
Let us assume that we have fixed all our predictable errors; we are still left with our uncertainty problem to the extent that we have to make choices now and will not know until later if they are the ‘right’ one. We still have a disconnect between the quality of the decision and the outcome.
Imagine you lead an office of 100 investment analysts. You have designed an incentive programme to reward your best analysts as follows: at the beginning of each quarter, each analyst must submit the name of a single publicly traded company that she/he believes will gain more than the overall market growth by the end of the quarter.
For example, let us say Analyst 1 chooses Company A at the beginning of the quarter. At the end of the same quarter, you look at the change in the value of the stock in Company A. If it has grown more than the market index has increased, then Analyst 1 has a positive review for the quarter. If at the end of the year you have received all positive quarterly reviews, then you give the analysts a bonus. At the end of the first year, you have six analysts who have all positive reviews and they are duly rewarded.
The bonus is based on clear evidence that they made good analytical decisions on their company picks. But is there really such evidence? To put it another way, how do we know when a decision is good rather than just lucky?
Determining what makes a good decision is often as difficult as the decision itself
One way to tackle this problem is to think about what the world would look like if the decision-makers had no ability and all the success was due to luck. After the first quarter, if there was no skill in your analysts at all, how many would have been lucky and picked a company that grew faster than the market? Our best guess would be about half. Of this 50, how many would expect to be lucky in the second quarter? About 25. In the third and fourth quarters, about 12.5 and 6.25 respectively. So we would expect about six to have all positive reviews just by chance.
Do we really want to reward people for being lucky? Creating key performance indicators that avoid these types of problems is another area that is rapidly developing within universities and the most innovative organisations.
Determining what makes a good decision is often as difficult as the decision itself. By carefully thinking about the problem, we can often avoid predictable mistakes. We can also integrate a more sophisticated understanding of luck and skill into our decision-making practice, which results in rewarding the right things and avoiding punishing the wrong ones. To do this requires care, practice and knowledge of the problems. That is where effective decision-making training can be of great help.
|About the author
Dr. Matthew Mulford is an affiliate professor at HEC-Paris, a senior research fellow at the London School of Economics and a visiting faculty member at the European School of Management and Technology in Berlin. He teaches in various HEC executive education programmes, such as the HEC Executive MBA, with research interests that include the psychology of judgment and decision-making in inter-dependent interactions.