I have limited reserving experience, from much earlier in my career. The bulk of my time since 2000 has been in risk. In March 2020, I became group chief risk officer and chief actuary of Everest Re Group. I have spent a lot of time since then “learning actual reserving.” This article presents my perspectives and questions about reserving, viewed through a risk lens.
For a summary of my risk lens, let’s rewind the clock to 2004 (see Figure 1).
Figure 1 comes from “Benchmarking Enterprise Risk Management Practices in a Derivatives Firm,” a 2004 article originally published in The Actuary magazine. The message to actuaries then (and now, and forever) is: Traditional actuarial roles are risk functions.
The next component of the risk lens is the well-known “Three Lines of Defence Model” (see Figure 2).
I adapted this to the “Three Lines of Reserving Defense Model,” as shown in Figure 3.
This model frames reserving department standard practices in a three-layer escalation process. Quarterly or monthly monitoring can take many forms. In riskspeak, monitoring looks for key risk indicators (KRIs), which have thresholds that trigger responses when tripped. One well-known monitor in use at many companies is Actual vs. Expected (A vs. E). Companies using A vs. E have developed “KRI-type” thresholds for what constitutes “actual is sufficiently worse than expected to warrant investigation.”
Advances in computing power open possibilities for companies to employ more advanced monitoring, including what is known as “actuary in a box.” This is a risk term from capital modeling in the Solvency II world. It refers to the modeling of reserving risk, which is the volatility of “actuarial best estimate” (ABE) in future modeled calendar-year scenarios. Think of it as follows: The capital model simulates reported loss in a future time period and a mechanical approach (actuary in a box) calculates the ABE for that scenario. A risk monitor could be based on change in mechanical ABE.
The next level response is investigation, aimed at first rooting out anomalies (or errors) versus true emergence issues. Anomalies are sent back to the operations of the company (e.g., claims, underwriting processing) for correction. True adverse emergence triggers the third level, formal adequacy assessment — what is commonly called “reserve studies.”
This distinction among the lines of reserving defense helps frame the difference between monitoring, investigation and actuarial reserving work product — assessment of reserve adequacy. It also sets the stage for introducing advanced techniques, such as machine learning. Innovative approaches can be positioned in a first line capacity, as risk monitors, with no misunderstanding that they represent a “second opinion” of reserve adequacy. They are simply there to highlight areas for we humans to focus.
Actuarial challenge
This is a formal Solvency II requirement, serving as a check-and-challenge of all key actuarial assumptions. Think of it as an “internal external view.” An actuarial challenge analysis is performed independently of both pricing and reserving. The scope spans planning, pricing, results monitoring and reserving. Its natural home is therefore risk.
IBNR allocation Is capital allocation (and not in a good way)
This is because IBNR is deep-in-the-money earmarked risk capital — and everyone knows what I think about allocating capital! The main problem with allocated capital is its meaning. The reality is, regardless of the allocated capital, the company is responsible for the full value of any claim. Looking at allocated IBNR, what is the meaning of allocating IBNR to a segment in an amount that is less than the typical limits of the policies that segment sells? If we are telling that segment, “Too bad — you need to fund your own losses with your IBNR allocation,” then what are they supposed to do about the highly likely deficit situation when a limit loss exceeds the funds available? (A “limit loss” is a loss maxed out at either a policy or contract limit.) And if we are not saying you must fund your own losses, then what does the IBNR allocation mean?
In response to this unfortunately common problem, some reinsurance finance leaders (where the issue is often more acute than in insurance) have floated the idea of a central “large loss” IBNR account — no allocation of IBNR at all. That is my kind of answer! When large losses arrive, regardless of the business segment, fund them from the central large loss IBNR account. Reserving regularly assesses the adequacy of the large loss IBNR account in total and tracks marginal changes over time.
One more comment: I understand some IBNR allocation is mandated by regulatory reporting requirements. In that case, we should strive to have it be as formulaic (inexpensive to produce) as possible.
Net reserving
Looks like what we do is:
(i) Assemble a triangle of losses net of whatever reinsurance programs were in place over time.
(ii) Apply standard methods.
(iii) Hope “homogeneity” applies.
Sounds fine if we have had the same reinsurance structure in place for many years or if we only had quota shares (which we can easily adjust for in typical reserving software). More commonly, though, companies have changing reinsurance programs over time, including excess of loss treaties (and let’s not even go there on facultative). What then is the most accurate approach for calculating net reserves?
We can look for assistance in standard practices of reinsurance broker actuaries. They provide the analytics to support cost-benefit comparison among alternative proposed reinsurance programs. They do this with proprietary software that calculates granular net loss distributions from input gross loss distributions. If we assume the worst case, a very different reinsurance program every accident year, then our proposed new method would perform, by accident year, gross-to-net modeling like the broker actuaries. This requires individual claim level reserve analysis, which is growing more popular around the world. (See “Beyond Triangles: Capturing Insights from New Analytic Technology,” Actuarial Review, September/October 2021.)
The central question
Most companies perform periodic formal adequacy assessments — a common schedule among reinsurers is that every segment is formally studied no less frequently than annually. In between the formal assessments, companies have some kind of “system” that responds to actual emergence, with the response involving the movement of funds from the IBNR account to the case reserve account.
The “central question” is: How should the system respond? A risk person would focus on the system performance — what do we want these interim, system-based indications to tell the business leaders? Here’s a simple example:
- Booked IBNR = $10M
- New claim = $5M
- Three main “System Response” Options:
- Booked IBNR $10M → $5M
Hold Ultimate, erode IBNR - Booked IBNR $10M → $10M
Hold IBNR, increase Ultimate - Booked IBNR $10M → More than $10M
Increase IBNR, increase Ultimate even more
- Booked IBNR $10M → $5M
Remember, this is our system’s response in between formal reserve studies. Which interim signals best serve management? How about that central IBNR account idea?
Hopefully, you all found these explorations and musings helpful. Send us any thoughts you have on some of these open questions to ar@casact.org.
Donald F. Mango, FCAS, is group chief actuary and chief risk officer at Everest Reinsurance Company in Warren, New Jersey.
