Thursday, October 30, 2014

Uncertainty

I remember being taught, a long time ago, that if you can't describe something, you can't measure it, and if you can't measure it, you can't manage it. I have been fascinated by management metrics since then, and I still believe that if performance metrics are designed correctly and aligned with an organization's strategic objectives they will naturally encourage employees to behave such that those objectives are achieved (for more on this, see my 2005 article Keeping Score .)

So when I came across the book Managing Uncertainty by Michel Syrett and Marion Devine, I had to look twice because the title troubled me. Uncertainty, of course, is the opposite of certainty--so being able to describe it well enough to measure it would seem to present an enormous challenge. According to the authors, while there is no precise, widely accepted definition of uncertainty, there are definite degrees of uncertainty. The book opens with a 2002 quote from Donald Rumsfeld:
There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are things we do not know. But there are also unknown unknowns--the ones we don't know we don't know.
Clear? One problem with planning (projects, strategy, risk or change management, financial, estate, etc.) is that we often use planning models that assume we can collect enough information about strengths, weaknesses, threats and opportunities to predict with acceptable certainty what challenges we will face during the plan period. Once the challenges are identified, we simply develop action items to address them, craft metrics to track performance against those items, assign accountability and put the plan in motion. But early in Managing Uncertainty the authors credit Arnoud De Meyer, President of Singapore Management University, with identifying four types of uncertainty:
  • Variation--where small influences cannot be easily anticipated individually, but the resulting total impact can be identified and managed
  • Foreseen--where identifiable and understood variances may or may not occur
  • Unforeseen-- where an event's possibility is not recognized or its likelihood/probability is sharply discounted (unk unks)
  • Chaos--where unforeseen events invalidate the strategy
In their paper On Uncertainty, Ambiguity, and Complexity in Project Management , Professor De Meyer and colleagues Michael T. Pich and Christoph H. Loch suggest that projects (and other plans) can be expressed as equations or "payoff functions" that are dependent on two things:
  • The state of the world
  • The choice of a sequence of actions
In other words, success in planning requires knowledge of the challenges presented by an operating environment, and the perspective to come up with appropriate action steps to deal with those challenges--although "adequacy of the available information" is critical to that success.  In their words: “Inadequacy of information is caused either by events or causality being unknown (ambiguity), or by an inability to evaluate the effects of actions because too many variables interact (complexity).” The more inadequate the information, the more difficult it is to plan successfully. But since information inadequacy arises from a lack of awareness or a lack of understanding, there are techniques (centered around learning and selectionism) that can be used to improve planning in unforeseen uncertainty situations. For a few practical examples, take a look at Three Tools to Manage Uncertainty by Kim Girard.

All of this struck me as interesting and relevant because I could see an obvious parallel within the insurance industry, particularly with the processes and procedures insurance claims managers use to set loss cost reserves. Anyone responsible for setting loss cost reserves on claims deals with De Meyer's four types of uncertainty, and they have no choice but to grapple with the adequacy of information available to them.

For example, when setting loss reserves on complex litigated claims that we believed were going to be tried to a verdict, we used to consider the likely verdict value (tied to damages, liability, venue, etc.) and modify it using a multiplier we called the "percentage chance of losing." So if the case had a verdict value of $500,000 and we had a 60% chance of losing, the reserve would be $300.000.

Ambiguity and complexity routinely invalidated that approach, however, because even if we were confident in our predicted verdict value and the % chance of losing, the actual verdict would rarely match our loss reserve. But when we did things to reduce ambiguity and complexity, when we developed better information about the likely trial environment (the state of the world--jury research, verdict history, attorney qualifications, etc.) and when our resolution plans included the most appropriate and impactful sequence of actions prior to and during trial, we usually got better outcomes.

I smiled when, in the course of thinking about this topic, I remembered situations during my career when I encountered individuals who were trying to manage uncertainty involving case or portfolio loss reserve adequacy in rather unorthodox ways. One claims officer "managed" reserve development by insisting on personally approving all reserve increases over $10,000, and he then moved very slowly on those requests. Another refused to improve any proposed reserve increase unless a corresponding and counterbalancing reserve decrease on another file was also submitted. A third incented his employees to minimize claim payouts each month, knowing that reserve adequacy for his environmental claim portfolio was determined based upon burn rate (number of years the reserve would last at current payment rates). A business unit leader, who was new to long tail lines and uncomfortable with the concept of prior year reserve development, insisted on sending teams of consultants out into the field offices to "fix" long tail line reserves. In the process he spooked the claims managers, artificially inflated case reserves and destroyed the reserve consistency that the actuaries relied upon to calculate ultimate loss exposures.

But my most enduring memory involved the business unit leader who told his assembled claims management group in September one year that he was tired of prior year reserve development, and that he didn't want to see any more of it after the first of the year. I don't need to tell you what happened with loss reserves in the fourth quarter that year.

All of which goes to show, I suppose, that some efforts to manage uncertainty end up creating even more uncertainty.

Dean K. Harring, CPCU, CIC is a retired Chief Claims Officer and an expert and advisor on property casualty insurance claims and operations.  He can be reached at dean.harring@theclm.org or through  LinkedIn or Twitter.