Attention: Due to COVID-19 restrictions, seat capacity at Prometric testing centers continues to be affected. If you are unable to find a seat for your exam, click here to see your options.

Back to The Futurism


Back to the Futurism

By Steven W. Easson

THE REPORT ISSUED by the Society of Actuaries in October 2005 entitled, "A Study of the Use of the Delphi Method, A Futures Research Technique for Forecasting Selected U.S. Economic Variables and Determining Rationales for Judgments," was the end product of many months of development work that aimed to educate actuaries and other professionals about the application of futures research techniques for collecting judgments from a diverse panel of experts from multiple professional disciplines. The judgments and their associated rationales obtained in this study were a plethora of plausible future developments and their plausible impacts on the future values of selected economic variables. Our study's participants included actuaries, economists, investment managers, futurists, modellers, politicians and scientists. The study is available at . Also in the April/May 2006 edition of The Actuary, Ted Gordon (a well–known futurist who was our study's principal consultant) and I described the goals, methodologies, results and application of the study's results.

Prompted by the emergence of the current financial crisis and possible misconceptions of futures research principles and the goals of this study in particular, I couldn't resist the temptation to put in another pitch to encourage actuaries to understand and incorporate futures research techniques into their professional practices, to reflect on the goals and results of our study and to address, in general, the goals and principles of futures research techniques.

To start with, the study, recognizing its inaugural status, was perhaps more art than science. One of the study's main goals was to illustrate the application of futures research techniques with the hope that it would motivate actuaries to utilize futures research techniques for many various applications (including repeating this particular study with the benefit of hindsight). For example, our Project Oversight Group did have a healthy and lengthy debate as to whether to imagine the state of the world 10 or 20 years into the future (we decided on 20) and whether the future world should be viewed as the average of what could happen over 20 years or a point estimate 20 years from now (we chose the latter).

Some of the goals and principles in utilizing futures research techniques and in particular the study's use of the Delphi method (and Trend Impact Analysis method) were/are as follows:

  • Identification of plausible future developments and scenarios. Futures research techniques are not meant to produce "predictions."
  • One very valuable outcome of the Delphi method in particular is the identification of plausible "outliers" (or quoting Nissam Taleb, "Black Swans"), that is outcomes which might otherwise be considered to have very little chance of occurring.
  • Futures research techniques are meant to produce a convergence of various opinions, not necessarily consensuses.
  • When using futures research techniques, qualitative judgments can be equally or in many instances an even more important output than quantitative values.
  • The outcomes from futures research initiatives serve as reality checks for results from quantitative models, not necessarily inputs to, nor overrides to, quantitative models.

Futures research techniques aim to identify a "fan of plausible scenarios" that provide management with valuable insights for setting both business and risk management strategies and tactics. Note that the range of this resulting fan will vary with the particular application and, as you will see, the resulting fan was quite wide for the study's particular application. Some other applications would produce narrower ranges.

Possible misconceptions from our study are that the SOA was making predictions, came to consensuses and even perhaps inferred the results were "statistically significant." These are not goals of utilizing futures research techniques and they certainly were not the study's goals.

The anonymous nature of the Delphi survey facilitates a true debate that is independent of personalities and any fear of embarrassment associated with offering an outlier opinion. Delphi surveys involve multiple rounds whereby the results of the previous round are anonymously fed back to the participants. Surveys are redone until there is a convergence of opinion. Consensus of opinion should not be required nor expected. For example, if after the first round, half the participants felt that inflation would be 1 percent (with their rationales for that possibility) and half the participants felt that inflation would be 5 percent (with their rationales for that possibility), the goal is not to necessarily have all or even a majority agree on one of 1, 3 or 5 percent and a common set of rationales. Rather the survey continues until the latest round does not produce results significantly different from the previous round. So, if in round two of my example none of the participants in either the 1 percent camp or the 5 percent camp are swayed, then the survey stops and the study's results have produced two plausible inflation rates (1 percent and 5 percent).

Finally, I personally view the qualitative opinions and rationales for judgments as equally or even more valuable than the quantitative results in many instances. In my opinion, the application of the numeric results obtained from this study should largely be limited to serving as a reality check for the results obtained from our robust quantitative models, to influence the choice of deterministic stress tests to perform and to influence risk management and risk appetite practices and frameworks and the development of economic capital.

So, how did we do regarding some of our qualitative judgments in the study?

It is hard to choose how to answer this question from the plethora of opinions, rationales and judgments collected from our study as outlined in our study's 142–page report, so I will, with the benefit of hindsight, selectively choose some of the ones that may be reflective of our current economic state. However, I encourage you to at least glance through our report to obtain a complete view of the outcomes of the study. Recognize that round one was sent to the participants in November 2004 and round two in March 2005, a time of seemingly increasing prosperity.

  • "The Fed may not be able to effectively offset a global recession (especially a serious one), so fighting deflation may be less effective than containing renewed inflation."
  • "The Fed's credibility is eroded by deteriorating debt, fiscal and trade issues."
  • "Economic collapse of the United States based on debt and deficits."
  • "Significant corporate defaults (tripling over current rates)."
  • "Economic depression for a seven–year period."
  • "Significant bear market returns for a 10–year period."

At this point I concede that many of you won't yet be satisfied without some of the quantitative results being listed, so here's a thumbnail of plausible values from the many listed in the report. Following are the 80th percentile confidence interval results from the Trend Impact Analysis (TIA) method. The TIA method utilized the plausible future developments identified in round one along with round–two–obtained associated probabilities and impacts to produce median estimates and these confidence intervals. In essence, the opinions, rationales and judgments from the expert panel were widely separated which led to the wide ranges below. The study's conclusion that the variables are intrinsically (that) uncertain was perhaps not a bad conclusion.

Given the recent one–in–many–years or many standard deviation economic and financial results, these wide confidence interval ranges, even at the 80th percentile, do not seem as implausibly wide to me as they did in 2005.

  • CPI: 0.6 to 9.9 percent.
  • 10–year Treasury Spot Yield: 3.3 to 11.4 percent.
  • S&P 500 Total Rate of Return: –20.2 to 23.1 percent.
  • Corporate Baa Spot Yield: 3.8 to 14.8 percent.

I hope that this helps you to better understand the goals, objectives and benefits of utilizing futures research techniques and increases your motivation to incorporate them into your professional lives when setting both business and risk management strategies and tactics, and for those of you who have yet to do so, to get involved with the Futurism Section.

Steven W. Easson, FSA, FCIA, CFA