Silent evidence in the captive insurance landscape
As the captive insurance landscape continues to expand and mature, the types of risks that are placed into captives will grow alongside it. Even now, the risks included in captives represent a wide array of exposures and business needs. Some of these are more “traditional” in nature and have historical data that is easily capturable.
Others, however, are emerging or have limited amounts of information. Placing the latter type of risk into a captive requires both analytical support and qualitative detail, especially when the scenarios which produce a claim for these risks are infrequent, but potentially severe.
One item that may be necessary when considering risk placement for a captive is an understanding of “silent” evidence. We at SIGMA have written about this phenomenon in the past, but we think it is an increasingly important piece of long-term captive management.
What exactly is silent evidence? Per Nassim Nicholas Taleb’s 2007 book, “The Black Swan: The Impact of the Highly Improbable”, silent evidence refers to the human tendency to view history with a lens which filters out evidence differing from our preconceptions. More specifically, silent evidence is evidence that we don’t have access to due either to the type of data being collected or our own inability to recognise that it exists.
For example, consider a situation in which you are studying a database of property losses with the goal of obtaining predictors for the most severe types of claims. Using these predictors, you might then develop various criteria to determine which of your company’s properties are covered within your captive.
Ideally, your study creates a reasonable prediction model based upon claims that already meet the criteria of being filed and being considered severe. The issue that such an approach creates is that it ignores a potentially significant number of claims that have occurred but have not yet reached the severity threshold.
By accounting for this “silent” evidence, you would not immediately accept the resulting model as a fully indicative predictor. Instead, you may also pursue qualitative information on the excluded claims and decide whether your study’s results need to be amended. A similar, but unfortunately common, situation occurs when captives use loss runs which are limited in terms of the fields being captured. In some cases, these datasets are streamlined to the point of excluding potentially important evidence.
It’s clear that something must be done to alleviate the problem posed by silent evidence—that of incomplete prediction—but what? To start with, mere awareness of the issue helps a great deal in identifying silent evidence in all its potential forms. When completing or reviewing a captive-related study which utilises any form of data, consider the ways in which the data is being captured or included.
Are there any flaws in the methodology? What are some possible scenarios that would result in a data point not being included in the overall dataset? For captive owners, can any qualitative information be obtained from management regarding pertinent oversights or available industry data? How might the historical data being used differ from exposures that could be faced in the future? A great deal of insight can be gathered by simply asking these questions and considering their potential impact on the analysis.
Captive owners who are familiar with and make use of the enterprise risk framework are already at an advantage in these types of scenarios. Many times, when companies undertake enterprise risk initiatives, risks are identified that have little associated historical data and have not explicitly been monitored or analysed.
“By minimising any potential oversights in the captive risk placement process, captive owners similarly minimise the possibility of unintended, adverse loss experiences.” Enoch Starnes, SIGMA
Captives engaging in this framework might first produce a qualitative assessment of the risks faced by the captive and its parent company. Once that step is complete, key decision-makers should have a high-level understanding what the potential risks are, what types of data might be the most useful in analysing such risks, and any potential areas which the data collected might not cover.
After this information is established, it can influence the interpretation of any analytical results and help the captive owners reach a better understanding of their captive’s potential exposure to risk, both known and unknown.
The “80:20” adage may apply to these pursuits, however. Asking these questions and listing potential answers covers 80 percent of the goal and requires only about 20 percent of the necessary work. Finding possible solutions to these issues and applying them may require additional time and resources. The end result, though, is often worth the work. By minimising any potential oversights in the captive risk placement process, captive owners similarly minimise the possibility of unintended, adverse loss experiences that the captive was not designed to handle.
The methods involved in capturing and mining data are constantly improving, and the associated analytics are becoming core strategic drivers in many industries. Clearly, those involved in the captive insurance landscape should be making significant use of these potential assets, and such endeavours always begin with understanding their uses.
Having a solid understanding of data doesn’t just mean interpreting what’s shown—it also involves contemplating the data that isn’t. Captives developed as proper, long-term solutions to their parent companies’ insurance needs require decision-making that contemplates a wide array of inputs. By considering the impact of silent evidence, their owners become one step closer to achieving their risk-related goals.
Enoch Starnes is an actuarial consultant at SIGMA Actuarial Consulting Group. He can be contacted at: email@example.com