Do you know the difference between risk and uncertainty?
An experiment by author Daniel Ellsberg on the popular podcast “The Hidden Brain” highlights the difference between the two. Imagine, Ellsberg said, an urn that prohibits you from seeing inside but that contains 50 red balls and 50 black balls. One ball will be picked; if you guess the color correctly, you will receive $100. Guess incorrectly, and you will receive nothing. How much would you pay to play this game? The expected outcome is calculated at $50, because you have a 50% chance of winning $100. It is then up to you to decide whether to play based on your risk appetite.
Now imagine a second urn that contains an unknown number of red and black balls as well as yellow and green balls. Again, one ball will be picked, and you will get $100 if you guess correctly, and nothing if you guess incorrectly. Which color will you choose? Will there be any of that color at all in the urn? It is impossible to estimate the outcome, and therefore people are reticent to put up their own money to play this game.
The point Ellsberg was trying to make was that the market (people, companies, etc.) can estimate risk and therefore factor it into decisions, but they will generally walk away from uncertainty. Visionary entrepreneurs have built great businesses and fortunes by wading into uncertainty and creating markets that did not exist previously. Think Microsoft and personal computing in the 1970s, Amazon and online retailing in the 1990s.
What risk versus uncertainty means for FP&A
Let’s look at this through the lens of common FP&A activities. Financial planning and analysis tends to follow risk practices in applying risk and uncertainty to the forecasts. Those with a high probability of occurring lead to reserves (loan losses) or expected losses (returns or breakage). Lower-probability but still significant risks may be managed by mitigating off risk via insurance or hedges. The unimaginable ones—nuclear war, federal default—are assumed to be big enough and of a low probability to be so catastrophic that the cost of protecting against them would outweigh the capital of the company to stay in business.
Try answering this question: How much capital do you need to store on your balance sheet to overcome a giant asteroid smashing into the earth and throwing up enough dust to blot out the sun? You can’t, because you cannot compute them.
What about building a business case for a new product or service where precedent does not exist? This is hard; the first 20 potential investors that Airbnb’s founders approached rebuffed them, showing that professionals who specialize in uncertainty have trouble valuing new ideas. FP&A practitioners already know the standard practices that come with trying to squeeze the unknown into quantitative models: identify the variables and uncertainties, make best estimates based on information available, refresh the model quickly, and present sensitivities and scenarios.
When dealing with true uncertainties, FP&A should recognize which cases are at the fringe of our capability to provide confident estimates. These decisions need to be made based on strategic considerations more than trusted quantitative ones. They may be excluded from the forecast all together, considered research and development for the business rather than a true investment, or factor in only the costs of the exercise without potential benefits. This is also where taking a portfolio approach to investments can be useful, where some are exploratory or revolutionary bets while others have more certain payoffs.
One final thought: When you review your models in these uncertain exercises, it is also useful to maintain a dose of humility and realize the limits of your quantification. It is hard to estimate all the ways that you will diverge from actuals!
How can we mitigate the risks of artificial intelligence in the financial industry? Read AI In The Financial Sector: The Ethics Of Algorithms (Part 1).