By Sam Gutterman
Regular readers (thank you, of course) of this column have gotten used to my focus on uncertainties and risks, key aspects of a risk management or estimation process. There is a range of possible outcomes in the future.
These discussions remind me of two famous, oft-quoted sayings of the American philosopher Yogi Berra (he was also a catcher for the Yankees): “It’s tough to make predictions, especially about the future” and “The future ain’t what it used to be.” How relevant!
Projecting what has not yet happened has to be approached with some humility. Not everything is as straightforward as determining the results of a well-behaved pair of dice. Two key areas to consider are differences in (1) risk characteristics of the populations in the historical database and that expected in the future and (2) expected changes in the future context.
Although classic “uncertainty” involves unknown unknowns, I usually apply a somewhat broader brush that incorporates any deviation from expected that isn’t solely based on random noise or process risk. This can include non-linearities and structural changes, as well as cliffs and tipping points.
Concern about the uncertainty of the future plays a key role not just in economics and actuarial science, but in every aspect of life, including the media.
For instance, at the time this was written, it appears that the coronavirus from Wuhan may become a full-scale pandemic. Its uncertainty has already affected capital markets worldwide and disrupted supply chains (as well as causing my wife to wear a surgical mask on a plane tomorrow). But by the time you read this End Paper, a better idea as to its trajectory will likely be available, as some of my future will be your past.
An example of uncertainty and how it makes decision-making difficult is climate change. There are significant uncertainties with respect to many aspects of the climate risk cycle. Nevertheless, the range of possible futures may lead to a conclusion that significant steps need to be taken to forestall significant costs.
Another example is the future trajectory of interest rates. The “Japanese” scenario (extremely low rates for multiple decades) was thought to be an outlier, particular to the Japanese situation—now it may not be. Yet another is the life expectancy of the Japanese female, the longest country/gender combination in the world. Can we be sure that these results will not be widespread in the future?
Because there is uncertainty about the future, materiality is a consideration. How sensitive are its consequences to alternative futures? This will have a bearing on how accurate an estimate is needed and what actions are warranted to avoid the most seriously adverse ones.
Natural laws of probability can be important (every actuary is a fan of the law of large numbers), but these can be outweighed by lack of relevant available experience data or model risks. Human behavior and external forces do not always follow such laws. Thus, the sensitivity of the future to differences in the inherent risk characteristics and external forces is important. Assumptions need to be regularly challenged. For instance, during most of my career it was assumed that nominal interest rates were floored at zero—now that some countries have lived with subzero rates awhile, it may be about time to discard such a constraint and understand its ramifications.
Even when we understand relationships and patterns, we may not be confident that historical trends will continue. Indeed, we may not be able to discern cause-and-effect relationships—lack of such knowledge can doom forecasts. But nevertheless we have to soldier on.
A recognition that there is uncertainty can help us derive sensible choices or recommendations. As actuaries, we need to regularly remind end-users of our advice of the implications of alternative futures and possible steps to effectively manage significant adverse deviations. Multiple viewpoints and risk preferences, possibly obtained through peer review, can be valuable. Powerful computers and Big Data can help analyze huge amount of historical data accurately. But if future risks or conditions change, unless care is taken, they may prove at best misleading.
Recognition and risk management can, to some extent, tame the effects of our uncertain world—even though, as Yogi also once said, “The game ain’t over ’til it’s over.” We will always face the future, and the future will always be uncertain. It is our role to try to manage its worst effects.