An Estimate by Any Other Name
By Paul E. Lacko
What words do we actuaries use among ourselves to refer to the work we have done and its results? What happens when we use the same language with non-actuaries? How do our words differ from the words other people use? Actuaries will generally speak the technical lingo of the trade, of course, among themselves. Experienced actuaries know full well that most non-actuaries just don’t comprehend this foreign language, which confusingly sounds a lot like the native tongue. New actuaries often have a hard time comprehending that such difficulty in comprehension even exists, and a harder time comprehending that rectifying the situation usually falls to the actuary. And the standard methods of helping a non-native speaker aren’t effective: speak m-o-r-e s-l-o-w-l-y; enunciate key words MORE LOUDLY; repeat important phrases, repeat important phrases, repeat important phrases.
Both groups, the actuaries and the non-actuaries, experience the communication, or lack thereof, as dysfunctional. Each group perceives that the problem lies with, and should be solved by, the other group. The larger group, of course, dictates the terms of the compromise. It falls to the actuaries to do whatever is necessary in order to express themselves in the native language as spoken and understood by the majority. The non-actuaries will not sit for an intensive course in conversational “actuarese.”
This might be a serious challenge, judging by a letter published recently in Contingencies. The writer argues that “the actuary doesn’t do any predicting. If we did, we’d be judged by how well we predict… The quality of the actuary’s work doesn’t depend on how close the projection was to the actual experience.”
Speakers of the native language would certainly reply that actuaries do so “predict,” and you can look it up in any standard dictionary! Here are a few definitions from mine:
Predict (v): to declare in advance; to foretell on the basis of observation, experience, or scientific reason.
Prediction (n): forecast
Project (v): to plan, figure, or estimate the future; to communicate vividly esp. to an audience.
Projection (n): an estimate of future possibilities based on current trends.
Estimate (v): to judge tentatively or approximately the value, worth, or significance of; to determine roughly the size, extent or nature of.
Estimate (n): a rough or approximate calculation; a numerical value obtained from a statistical sample and assigned to a population parameter.
Well, gee, these words certainly sound appropriate to describe actuarial work. What’s the point in quibbling with native speakers (who outrank us) over small connotative distinctions? As far as native speakers are concerned, actuaries estimate, predict, project, forecast, and probably a lot else besides.
Let’s think a bit about “quality,” as well. The “quality” of my work is reviewed frequently, and it undergoes a formal review once each year. As defined by the non-actuaries to whom I report, “quality” most definitely includes how close my projections are to the actual experience. This is not measured by how close any individual projection is to the actual experience that emerges, but overall I need to maintain a decent batting average. Batting average might not be the best analogy here; a professional baseball player who can maintain a .350 average is considered top-rank. Batting .350 gets me a grade of “Needs Improvement.”
What do the writer’s words say to non-actuaries in their native language? And why should we feel surprise, even resentment, when non-actuaries in senior management complain that actuaries lack “business sense”? Actuarial work—any work, for that matter—is always judged retrospectively in view of results. “Quality” and “value” are essentially synonyms to most native speakers. Senior managers want value for the dollars they spend on actuarial services.
I think I understand what the writer is trying to say, and I’m sure you do, too. Frank Schmid and Jonathan Evans express it better in their Winter 2007 Forum paper, “Forecasting Workers Compensation Severity and Frequency Using the Kalman Filter.”
Forecasting is a signal extraction and signal extrapolation exercise. Signal extraction is the process of filtering out measurement errors from empirical data. Measurement errors include the total impact from all sources of noise, deviations of the empirical data from the underlying signal that do not affect the expected values of future observations. In forecasting, the signal is the quantity of interest, because it is the signal that determines the expected values of future observations…Specifically, it is the objective of a forecasting model to elicit from historical observations the process that generates the (unobservable) signal. Because the forecasting model replicates the data-generating process of the signal (instead of fitting historical observations), the quality of these models cannot be judged by the fit to the observed data….
If you and I can translate this technical language into the vernacular, we might just score a few quality points that boost our batting averages.
Senior managers can be made to understand that actuarial models are concerned with “right methods” as opposed to “right answers.” They can understand that actuaries provide signals, and, perhaps more importantly, that actuaries can provide useful information about the noise. Noise is very important to senior managers. They cannot ignore or eliminate the noise. It means risk and uncertainty, which can be retained, managed, hedged, or transferred, but never eliminated or ignored. Still, “right methods” have no value unless the actuaries who apply them season after season show consistently high batting averages.
Ultimately, I fear, differences in backgrounds, cultures, and work environments will always cause our audiences to derive meanings from our communications that we did not intend. Language can only indicate the signal of our intentions. Noise, inherent and unavoidable, distorts the signal. But I believe we can reduce the noise to a tolerable level. When in Rome, speak as the Romans.
This brings to mind a song by a gifted singer/songwriter from the late 1960s and early 1970s you probably never heard of, Tom Rapp. He recorded “Song About a Rose” with his group, Pearls Before Swine, on an album called “The Use of Ashes.” I leave you with the last two lines of the song:
And even God can only guess why or where or when or if the answers all belong And you and I, we sing our song about a rose or perhaps the shadow of a rose.