From the Readers
Recognizing Uncertainty when “Getting to the Right Answer”
President Ralph Blanchard’s column (“Getting to the Right Answer,” AR, August 2011) begins with a well-stated and important point about actuarial practice, but concludes with a self-destructive proposal for our profession. Actuaries must recognize the uncertainty in data, analytical models, and resulting answers. This uncertainty should be communicated with clients. We may become lax about this in practice, but I think almost all CAS members sincerely agree with this and place it very high among our values.
It is also true that pure mathematics, with its rigorous logical deductions from statements postulated to be true to other statements that are unambiguously either true or false, is not applicable to the real world without a deep fundamental adjustment to account for the enormity of uncertainty. There is a well established and constantly evolving discipline that incorporates uncertainty into mathematics, namely probability and statistics. However, improving our training of actuaries in probability and statistics is the opposite of what is advocated in the column.
It is bewildering when the proposed solution to hubris about the validity and accuracy of a singular answer is “teambuilding.” Teambuilding is a vitally important skill in society, and particularly in productive organizations, but to offer it in contrast to an equation that an overconfidence in the correctness of an answer is individualistic implies something else very sinister. The implication is that respect for the uncertainty of actuarial answers requires an emphasis on subjective truth in actuarial answers.
In many fields of human endeavor, such as expressive art, truth is partially or totally subjective. Subjective truth plays no role in mathematical deduction and only a small role in accounting and particularly in auditing. Actuarial work is not pure math but it is an application of probability and statistics focused on risk accounting. An emphasis on subjective truth in actuarial work, through teambuilding, is a perilous road. One stop along this road is opining that a loss reserve much smaller than standard methods indicate is reasonable because a “team player” would not diminish profit sharing for the team this year. Another is opining that a contract with only deterministic cash flows is insurance because team cohesion might be improved by a more stable income statement.
—Jon Evans, FCAS
2011 CAS President Ralph Blanchard III responds:
Unfortunately the response from Jon Evans misses the point with regard to the precision of actuarial science. Actuarial models are only approximations of reality, and stochastic models are merely more sophisticated approximations. They may claim to give precise estimates of the uncertainty, but that is frequently false precision that is not useful for many decision makers. As one example, RMS had a stochastic catastrophe model that quantified the risk from hurricanes striking inland properties. Then they updated their old stochastic model to a new one, version 11. The result was a materially different view of the risk. Hence the use of a stochastic model didn’t make the uncertainty go away; in the RMS case it only added another source of uncertainty.
The point that Jon’s response missed was that, instead of trying to determine a mathematical model of the uncertainty, it is often more valuable to identify the issues that drive the uncertainty. This is more useful to the decision maker. It also requires working with others to understand what is useful to them, and to explain the uncertainty in a manner that recognizes that we are dealing with real-life issues, not mathematical puzzles. In short, we need to understand our nonactuarial audiences in order to communicate effectively with them. Insisting that our stochastic models are another form of “truth” is not communication and will not further the profession.