Zero is an interesting number. Due to the way multiplication work, it can be sticky, making it hard to get away from it. A good example is investments: they need a kickstart to get going and if by chance they were to be wiped out, like kindling that went out, they’ll need another influx before having another chance. The same goes for business.
To be a worthy vehicle for capital, the point of a corporation is to increase shareholder value. However, some take a terribly limited view of that goal, sacrificing long-term viability for quick profits. A business is hard enough to get started that to optimize its lifetime value, we need to keep it alive and healthy for as long as possible. It may be a suboptimal individual investment strategy, but feels more sustainable for society.
I propose a similar philosophy for making decisions. In a business context, your number one priority should be making decisions in a way that avoids putting you out of business. Intuitively, you first have to survive if you want to maximize the lifetime profit of your business.
This week, I introduce the role of risk and regret in evaluating decisions, explain why I avoid using utility as a metric, and show why your evaluation of certain risks entirely depends on your financial situation. While it’s written with businesses in mind, the concepts can just as easily be applied to your personal life.
Why do we make decisions?
To know how to make good decisions, it’s worth taking a step back to understand what we mean by a “decision”. This will also let me lay some foundations for future articles.
When faced with a problem, we aim for a certain outcome and have to assess which alternative is likely to get us there effectively. To reduce costs, we usually don’t implement multiple alternatives. Consequently, the choice we make excludes other alternatives. Making a decision requires a level of commitment to our chosen path.
That commitment will take the form of resources that are allocated to the pursuit of that alternative. In a business context, resources could be capital, labor, infrastructure, time, reputation, etc. In the implementation of the alternative, resources could be:
tied-up to the project, in which case we need to consider the opportunity costs incurred by not having them available for other projects (e.g., a team of engineers);
spent to implement the alternative (e.g., money spent on cloud services) and may only retain some residual value after its use for the project (e.g., servers purchased).
As a guideline, executive will often require that the projected return on resource investments needs to be higher than what you could easily get on the stock market to greenlight the project. This is so the company as a whole remains a worthy alternative for the capital investment.
To tie it all together:
➰Decision An intentional commitment to allocate resources with the goal of achieving a certain outcome. ➰Outcome The result of a decision once all uncertainties are cleared up. ➰Alternatives (or choice) A set of credible solutions to achieve the desired outcome. One could easily find the best alternative if it weren’t for uncertainties in the mapping between an alternative and its resulting outcome.
As a reminder, we’re interested in effective decisions, meaning we exclude decisions that have an obvious alternative or require a negligible amount of resources to implement.
When choosing an alternative we optimize for certain, potentially competing, criteria. The criteria can be tangible (like profits) or intangible (like brand strength). In game theory, this is referred to by the term “utility”. It’s defined, for a given alternative, as the value attributed to the outcome you expect to achieve by choosing that alternative.
However, in the context of decision-making, we will use the mathematically equivalent regret rather than utility.
➰Regret (or opportunity loss, opportunity cost) The gap between a theoretically optimal alternative and the alternative under consideration. It’s a hindsight metric used to assess the quality of a decision.
Regret comes from not getting the expected outcome. In that case, we may consider some or all of the invested resources as wasted or at the very least used inefficiently.
I believe that focusing on the probability of loss helps a decision-maker tread carefully, and avoid risky strategies that have the potential to hurt them. (It also allows for a more elegant definition of both a risk-neutral and risk-averse decision strategy. We’ll explore this in a future article.)
Regret minimization assesses opportunity costs by tying the quality of the alternative we’re considering to the best one we could have chosen in hindsight.
Under this framework, we make decisions to minimize the regret felt by picking another alternative (or refusing to choose any). However, regret being a measure in hindsight means that to use it to choose the best alternative means we need tools to handle uncertainties.
Understanding Risks
I relied so far on your general understanding of the definition of “risk” i.e., a chance of bad outcomes. In decision sciences, risk is usually a bit more neutral.
➰Risk (Formal ISO definition) Effect of uncertainty on objectives. “Effect” is to be understood as a deviation from expectations, either positive or negative.
Too often, people ignore the potential downsides of their decision, especially when the consequences (e.g., brand damage) might be hard to assess. Sometimes, it leads to spectacular failures as can be read in Dr. Paul Nutt’s collection of hundreds of stories of that kind in his book “Why Decisions Fails” (2002).
For effective decisions-making, risks need to be understood and included in our decision-making process.
Estimating risks
To incorporate risks in a decision analysis, we need the tools that allow us to efficiently estimate risks. As I alluded to last week, too often people with a scientific background fall into the trap of “if it can’t be perfect, it’s not worth doing”. But even flawed estimates could point unequivocally to a single alternative. Generally, people need less data than they would assume.
Risks can be estimated through various means. Examples, in a rough order of cost and precision:
Your own SWAG estimates;
Subject-matter expert opinions;
Previous studies that explore similar situations;
Data you collect (surveys, previous similar cases, etc.);
Prototypes and test runs.
Typically, we’d start from the top and work our way down as needed. A key concept is that even the most uninformed estimate, if accurately representing your level of uncertainty about the risk, can help you confidently pick an alternative over the others. You don’t need to know with any more precision whether the chance of rain for tomorrow is 70% or 90%: in either case you should pack a rain jacket.
The goal isn’t to eliminate risk, but to remove just enough uncertainty to make a decision with confidence.
Mitigating risks
When allocating resources towards some objective, there's a chance that the objective won’t be met, potentially wasting time and the allocated resources. We often shorthand this by saying that the resources are at-risk.
However, it’s rare that the entire investment is at-risk:
You may only have lost time and the tied-up resources aren’t spent;
The outcome is still serviceable, but less valuable than anticipated;
As the implementation of the decision unfolds, often it is possible to anticipate a bad outcome and mitigate the loss by cutting the project short.
In any case, your awareness of the risks associated with an alternative enables you to put contingencies in place ahead of time to mitigate those risks. This way, you might deliberately make riskier bets that allow for better payoffs. The point is to do it mindfully and carefully rather than arbitrarily taking riskier approach in the hope it’ll work out.
Prospect Theory
Another reason to focus on the negative consequences of a decision is tied to loss aversion. As you probably know, people tend to suffer more from losses more than the enjoyment they get from an equivalent gain. Estimates (Tversky & Kahneman, 1992) place the increased value we put on losses to sit between 1x and 4x higher. The difficulty in getting a more accurate estimate stems from the fact that the exact amount depends on one’s wealth.
For example, consider the following game. I flip a fair coin:
Tail: I give you $3;
Head: You give me $1.
That game probably sounds like a no-brainer. Your expected gain is $1 and the most you risk is $1.
Now, an analogous game. I flip a fair coin:
Tail: I give you $100,002;
Head: You give me $100,000.
While the expected gain is still $1, the game suddenly looks a lot riskier. According to utility theory, you should still play that game as your expected outcome is positive, but the potential loss of $100,000 is hard to shrug off for most people.
Prospect Theory studies the value of gains and losses as a function of their magnitude, relative to a reference point that is unique to every individual. (Prospect theory results demonstrate that reality is even more pernicious and explains why poorer people have a tendency to participate more in lotteries.) For our context, I’ll adapt this in the following way:
➰Prospect Theory Corollary The more a potential loss leads to a position that threatens the survival of the business, the higher is the regret associated with that loss.
If a certain alternative could threaten your ability to stay in business, its regret should have their weight increased to reflect that potentially fatal risk to the business. Conversely, the less you’re threatened, the more you should evaluate your options in accordance to utility theory, ignoring loss aversion.
Humanity seems to have internalized this rule for the survival of the species, yet it’s only recently been considered as a complement to utility theory. We seldom play a game only once. You can’t keep playing if you’re out of money. You can’t make profits if you’re out of business.
In practice, however, this can be hard to get right unless you have a good understanding of the risk appetite of the business, which is itself challenging to quantitatively measure. Unless you’re just starting a business, a good rule of thumb proposed by David Charlesworth in “Decision Analysis for Managers” is to pay careful attention to any alternative that may put more than a fifth of a business’s assets at-risk. In those cases, you may add weight to its associated regret to penalize that alternative in line with the increased risk. If that is too onerous, you should at least make sure this is flagged to the decision-maker so they’re fully aware of the potential consequences of picking that alternative.
Avoid going bankrupt.
Next week, I’ll present a general purpose decision-making framework. Subscribe to the newsletter so you’re notified when it comes out!
I’m also collecting stories about important decisions my readers have made in the past, big or small! If you want to share, either leave a comment or send me a DM!
Comments