Non-Formulaic Reserves: Where Are We Going?

Non-Formulaic Reserves: Where Are We Going?

Two experts discuss the future of Non-Formulaic Reserves.

The April 2004 edition of The Actuary newsletter contains an excellent article titled, "Principles-Based Regulation Of Variable Annuities Proposals Now At Pivotal Stage," written by Max J. Rudolph. This material is equally applicable today as research and debate continue to revolve around the evolution of non-formulaic techniques and applications. The purpose of the article you are reading now is to discuss the emerging issue, in general, that surrounds principles-based reserving. This very controversial subject has the potential to provide new opportunities for our customers and our employers. Our expertise is certainly needed to appropriately address this emerging issue.

Talk About It

The year 2005 has seen considerable discussion between life insurance companies, regulators and accounting groups regarding non-formulaic reserves. These explicit discussions around formulaic calculations (calculated using pre-defined deterministic formulas and assumptions) and non-formulaic calculations (calculated using company-specified stochastic formulas and assumptions) have implicit implications on a much larger scale. Our everyday usage of non-formulaic techniques is rapidly growing and currently includes such examples as X-factor validation, asset adequacy testing, FAS 133, C-3 RBC and SOP 03-1 compliance. It is important for us to understand where we are and where we want to be as we work toward the optimum solutions for regulators, accounting bodies and the actuarial profession for tasks such as product development, compliance, reserving and taxation. Without a doubt, we are all affected by this emerging issue.

Two of our SOA members, Tom Campbell of Hartford Life and Ed Robbins of Allstate Life have been interviewed to hear their thoughts regarding non-formulaic principles.

Tom Campbell

M.B.: What approaches are now being used to determine whether non-formulaic reserves are appropriate?

T.C.: Two approaches are being used at this time. One is the cash flow (i.e. asset adequacy) testing that is performed now by the valuation actuaries at the end of each calendar year. The other is gross premium valuation. New tools are clearly needed to assure we will be able to do this correctly and with a reasonable amount of effort.

M.B.: Would flexible experience studies be one of these tools that could help this process?

T.C.: Yes, absolutely. Prudent best estimates (which is one of the goals of non-formulaic reserves) require the use of data that fits the need of each insurance company. The "one-size-fits-all strategy" that we have been using in preparing previous experience studies does not address the real life needs of new products, new markets, new underwriting practices, new channels, etc. In fact, several of these criteria have not been explicit factors in preparing previous experience studies. By allowing the use of pivotal tables to select the criteria desired for the experience tables the corresponding non-formulaic reserves will be considerably improved.

M.B.: Do

you have a wish list for other enhancements you would like to see in experience studies?

T.C: Yes, some new variations (or types) of studies I would like to see include withdrawal patterns, expenses, annuitization rates, free partial withdrawal rates, UL secondary guarantee funding levels, and annuity lapse rates associated with death benefit guarantees, living benefit guarantees and guaranteed minimum income benefits. This is definitely not a complete list, but it should serve the purpose of suggesting that a broader depth is needed when we consider preparing experience studies in the future. One additional request is for a mortality study for deferred annuities.

M.B.: Any downsides you foresee in the goal of trying to prepare more useful experience studies (especially for helping the process of calculating non-formulaic reserves)?

T.C.: Yes, credibility needs to be a key factor. In other words, we need to know if the data we are looking at (ex. preferred underwriting mortality rates) is coming from comparable companies and are statistically significant. It would be great if we could choose the companies whose experience we would use when running these dynamic experience studies. This may not be possible but at least you can see the issue that needs to be addressed.

M.B.: What is the regulatory environment toward non-formulaic reserves?

T.C.: Today the insurance departments are very receptive to this new approach to reserving. Of course there are state variations, such as New York that still wants to see a formulaic floor on reserves. It is too early to predict what the states will finally adopt, but the current direction seems to be to apply stochastic testing to 1,000–10,000 random scenarios and then rank the reserve results. The final reserve needed comes from averaging the reserves needed from the worst 35 percent of these stochastic scenarios. Transparency and accuracy are two key criteria for this to be successful. To capture where we are today, it is far better to try something than to do nothing. Waiting helps no one. Two very hot areas that illustrate this are variable annuities with guarantees and AG 38 (UL with secondary guarantees). Two other hot areas that also illustrate the need for non-formulaic reserving are the whole risk management framework and dynamic solvency testing.

M.B.: What can we do to help the regulators to accept change?

T.C.: We need to do a great sales job. We need to be convincing. The sad truth is that each regulator is different and each will likely need a different, tailored presentation. To maximize our strategy we need to focus on one product type at a time. Two particular topics the regulators will address early in this process will be reinsurance and policyholder funding levels.

M.B.: Any comments about the international scene?

T.C.: Yes, Canada has had non-formulaic reserving in place for several years. Japan is currently working on this adoption even as we speak. This is an example where federal regulation of our financials would be advantageous. It would be so much easier to sell this once (to the federal government) than 50 different times to 50 different jurisdictions.

M.B.: Do you believe we will be able to apply these new ideas to GAAP?

T.C.: GAAP is evolving even as we speak. SOP 03-1 and FAS 133 already embrace non-formulaic principles. This trend is expected to continue as we also move closer to fair value accounting.

M.B.: Any additional thoughts before we end this conference call?

T.C.: Yes, the SOA could definitely help with developing new methods for generating and calibrating the stochastic scenarios. Improved modeling techniques are greatly needed, too. Try to appreciate the difficulty in running 1,000+ scenarios where each scenario has at lease one million contracts and each scenario projects at least 30 years into the future. Tax issues will potentially play havoc with the success of non-formulaic reserves.

Ed Robbins

M.B.: Are you aware of any other approaches that are now being used to obtain non-formulaic reserves?

E.R.: I would first like to elaborate on the definition of "non-formulaic reserve" methodology. By non-formulaic reserve methodology, in addition to the absence of a closed-form reserve formula, I will assume that such methodologies virtually always determine reserves for a block of business in aggregate, as opposed to a seriatim approach.

Other approaches to obtain non-formulaic reserves have been used for many years in actuarial work, primarily in the health and property/casualty fields. For instance, major medical claim reserves are typically set by aggregate development (lag) methodologies. Under these methodologies, approaches are determined and set uniquely by each insurance company and vary from period to period. On a less significant basis, smaller reserves have commonly been set using average historical trending approaches, often as factors applied to convenient base units.

M.B.: Do you believe flexible experience studies would be a tool that would help this process?

E.R.: I do see this as a potentially powerful tool. However, each company must also consider its own policy provisions and markets. It is generally best to use our own experience to the extent it is statistically credible. Where full credibility does not exist, a blend of company experience and industry flexible experience studies would appear to make the most sense.

M.B.: What potential concerns do you see if the insurance industry moves further toward non-formulaic reserves?

E.R.: My list of concerns is as follows:

  • Non-formulaic reserve calculations at the individual product line level are not as meaningful as they might appear. The best test of reserve adequacy comes from aggregating company-wide. For example, equity-indexed annuities can form an effective natural hedge of a variable annuity block that contains MGDB's and VAGLB's.
  • To focus this significant expenditure of effort on reserves and not on "total asset requirements" including required capital, where stochastic, aggregate approaches to risk measurement have the most value, is to focus on the wrong metric.
  • To expend scarce actuarial resources on many different standalone, product-by-product, non-formulaic reserve-setting exercises would be very time consuming, and especially burdensome on smaller companies.
  • Formulaic reserves are relatively easy to check. Non-formulaic reserves would be far more difficult to audit.
  • There is a concern that constantly changing current assumptions and market conditions going into the non-formulaic reserve calculations would cause undue volatility of statutory earnings.
  • Tax concerns. We can discuss that later.

That said, for guarantees on variable products, there may be no reasonable formulaic approach to gain a sense of reserve adequacy, and we might have no choice but to go to non-formulaic reserves on those products.

M.B.: It sounds as if you may not be in favor of moving rapidly toward a reserve standard very different from what we use today?

E.R.: I wouldn't quite say that. There is room for both formulaic and non-formulaic approaches. In fact, I believe formulaic and non-formulaic reserves are quite complementary and can work in tandem. In the work of the Variable Annuity Reserve Working Group, aggregate reserves are arrived at by non-formulaic, stochastic means, but a formulaic standard scenario reserve attempts to provide a minimum floor on a seriatim basis. Its purpose is to serve as a safety net to protect us from ourselves in case we become too optimistic about the future. In general, I can easily see such a floor being adjusted from time to time if we are also performing additional (non-formulaic) reserve and capital testing to find an appropriate range of reasonable reserve levels. I am not a believer that a risk metric like a reserve can be well represented by a single number, though our financial reporting environment requires that. I support the view that reserves that are neither unduly redundant nor overly conservative should be a goal, and one way to get there is by examining non-formulaic risk profiles of the business to adjust formulaic reserves from time to time when necessary.

M.B.: Are you pleased with the phrase "non-formulaic reserves"?

E.R.: The phrase as such is not the issue, as some types of business probably require that. Perhaps more pertinent and useful is the phrase "non-formulaic approaches to measures of company total asset requirements." The proper level of total asset requirements is the relevant metric, and where you draw the line between how much of total asset requirements reflects reserves and how much reflects capital is not as important pre-tax. I would also stress that non-formulaic risk analysis approaches have wide application beyond total asset requirement determination–for example, product pricing and appraisals of companies and portfolios.

M.B.: Any final thoughts you wish to add?

E.R.: Two thoughts:

  • This is as good a time as any to talk about taxes. I am concerned about tax reserve deductibility. In particular, the structure of Section 807 of the Internal Revenue Code suggests that a seriatim formulaic reserve stands the best chance of providing the industry with a fair level of deductible tax reserves. That seriatim reserve would do well to be an actuarial calculation based on a recognized mortality table and assumed interest rate. Because an equitable level of deductible reserves is crucial to fair taxation, there is no question that fair taxation is a professional actuarial issue, and we should feel some responsibility to see that the authoritative statutory reserve guidance that we develop results in relatively fair taxation.
  • As a summary comment, I agree with Tom that we need to think beyond the limits of single-point, individual policy, closed-form reserve calculations as a metric to measure company solidity. First of all, stochastic approaches to determination of the risk profile of in force blocks can be extremely helpful in setting levels of reserves, including recalibration of formulaic reserves. Second, reserves as such are not nearly as important as total asset requirements, of which reserves form only a part.


Numerous conclusions can be drawn from this material. First, we need to move carefully so that we have the full support of the regulators, IRS and accounting bodies. Second, change is occurring and we need to be very involved so that actuaries are participating in (and not victims of) these discussions. A greater body of experience data is an excellent tool for non-formulaic calculations. Improved modeling techniques are needed, too, so that the calculations can be manageable even for the smallest life insurer. International knowledge and experience is priceless in the direction and insight that can be obtained. Formulaic techniques are not going to disappear, rather the choices and the need for non-formulaic techniques and the role of the actuary are only going to expand. Solutions incorporating both formulaic and non-formulaic techniques need our support. Success in evolving toward successful growth of non-formulaic techniques depends on our unity, knowledge and willingness to change. These are a few of the leadership tasks that can be met by the SOA.

Mike Brown is life practicing actuary for Principal Financial Group.

Tom Campbell is vice president and corporate actuary for The Hartford Life Insurance Companies.

Ed Robbins is senior actuary for Allstate Life Insurance Co.