One size does not fit all in the life arena, says Douglas J Knowling, and life reinsurers may need to pioneer more rigorous capital requirement
Capital management has risen in importance in recent years for both direct writers and reinsurers in the US life insurance industry. Conservative regulatory reserve and capital requirements, combined with rating agency and shareholder demands, have placed some companies on a tightrope stretched between ensuring solvency through overcapitalisation and optimising returns with a thin layer of capital. Interestingly, capitalisation levels for many life reinsurers are influenced, to a significant extent, by a 'one size fits all' formula for C-2 - or insurance - risk that has not been refined or updated over the ten-plus years it has been in use.
This 'one size fits all' approach is most prevalent in a life reinsurer with a concentration of C-2 risk. Many direct writers in the US have turned into asset accumulators and sales organisations, passing off the majority of traditional insurance risks. Recent analysis of capital by risk category suggests that, across the industry, around 7-8% of capital is supporting mortality and morbidity risks. For a life reinsurer, this percentage might be more in the 70-80% range. While a life reinsurer may have a greater interest in C-2 risk, such capital levels affect reinsurance pricing and, therefore, direct writers who pass off most of their mortality risk.
Risk based capital
Risk based capital (RBC) requirements were developed in the early 1990s as an early warning system for regulators to determine if insurance companies are appropriately capitalised. The underlying premise is to ensure that capital and reserves are sufficient such that a company can meet its obligations with a high degree of certainty. It is a factor-based approach using information publicly available in the US statutory filings. If the ratio of actual capital to RBC falls below set breakpoints, the company and/or regulator must take action depending on which trigger has been hit (e.g. 'Company Action Level' versus 'Mandatory Control Level').
RBC is determined in components with a high-level adjustment to factor in interaction among risks.
- C-1 (asset default risk) capital is determined based on the relative credit risk of assets held. There is further adjustment for portfolio size and concentration among issuers.
- C-2 (insurance risk) capital is calculated based on amount of insurance in force. For life insurance, there is a schedule of factors that decrease as total net amount at risk increases.
- C-3 (interest risk) capital is based on overall riskiness of the liabilities. This can be determined on a factor basis or using complex asset/liability models.
- C-4 (business risk) capital is an add-on amount for other risks and is determined as a percentage of premiums.
The capital formula for C-1 and C-3 risks, while not perfect, has evolved over time, including adjustment of factors as well as refinement of methodology to incorporate new advancements in asset/liability analysis. For C-2 risk, however, the factors and methods have not changed.
It should be noted that RBC sits on top of regulatory reserves, which can have a high degree of conservatism built in. Such conservatism is not factored into the RBC requirements. As a case in point, starting in 2000, the so-called XXX (Triple-X) reserving method became standard for term policies with rate guarantees. A policy issued in December 1999 would have significantly lower reserves than one issued in January 2000. However, the RBC that sits on top of the reserve, presumably to ensure a high probability of solvency, is the same for both policies.
Rating agencies also have formulae to determine benchmark capital levels.
Commonly, benchmark capital levels are calculated based on formulae similar to RBC to determine a raw rating. In some cases, the formula and factors for C-2 risk are the same as those for RBC. Additional qualitative analysis supplements the benchmark capital analysis to determine the actual rating.
Many companies manage capital as a multiple of the RBC Company Action Level as a proxy for the amount of capital needed to maintain a desired rating.
Where US statutory filings are not available, rating agency capital analysis is based on the US GAAP (Generally Accepted Accounting Principles) balance sheet. Net GAAP reserves are closer to an economic 'best estimate' than regulatory, but there is still no direct recognition of inherent risk in the underlying business. To the extent a company performs additional analysis and can demonstrate understanding of its risks, such information would likely factor into the qualitative analysis portion of a rating, but 'hard' credit is not always given.
The RBC formula for C-2 life insurance was developed to cover catastrophes and volatility. It is a sliding scale that starts at $1.50 per thousand for the first dollar of net amount at risk decreasing to $0.60 per thousand of net amount at risk for any amounts in excess of $25bn. Claims volatility is driven by number of policies, face amount, and expected mortality rates.
While the sliding scale does make a provision for reduced volatility as more business comes on the books, it doesn't go far enough, and implicitly assumes that all policies are the same size across all companies.
As an example where this implicit assumption wouldn't hold true, the claims volatility for 25 million $1,000 policies would be very different than it would be for 25,000 policies each with a face amount of $1m, even though the total face amount is the same. Of course most companies don't have such extreme differences in risk profiles, but consider the different volatility that would arise from reinsuring a pro rata share of many small face amount policies versus a smaller number of policies with large face amounts.
An upper limit of $25bn of face is well short of the true upper limit for meaningful reductions in claims volatility as determined by statistical analysis. A face amount of $25bn might be generated by 250,000 policies each with a face amount of $100,000. While this would be a fairly large block for many direct writers, it is a fraction of what a life reinsurer might have. As a point of reference, a block of 2.5 million policies would have a standard deviation, as a percentage of expected claims, approximately one-third of the block of 250,000 policies.
Beyond the claims volatility shortfalls of the RBC C-2 approach, using the same factors for all companies and blocks of business is also problematic.
There is no differentiation by exposure to various catastrophes or other drivers of risk. A block of high face amount COLI (Corporate-Owned Life Insurance) business concentrated in a metropolitan area would have the same capital factor as small face amount policies spread over a larger geographical area.
In addition, once the in force has surpassed $25bn, this type of capital formula generates the same capital requirement for new business regardless of inherent risk in that business. There is no recognition of variation in profit margins, quality of underlying assumptions or relative conservatism built into assumptions. For a reinsurer, such inherent risk variations are further amplified by differences in reinsurance structures and risk sharing arrangements.
Economic capital approach
Many companies are developing an alternative approach to determine the appropriate capital level based on the underlying economics and associated risks specific to their business. While there are several approaches to economic capital, a common method is to determine the total amount of reserves and capital such that obligations can be met with a high degree of certainty (i.e. the 95th percentile). Recognising that setting the total amount of assets to be held in support of the business matters more than the breakdown of reserves and capital, economic capital can be determined as the amount needed in excess of net GAAP reserves. In fact, GAAP reserves might represent something along the lines of 60-70% certainty of meeting obligations (best estimate plus provision for adverse deviations), so the amount of economic capital will depend on the level of certainty desired in total.
In a perfect world, one could develop a multivariate model based on statistically credible data that could stochastically measure the economic capital needed to support a block of life business. In reality, we are a long way from having such models and data available for all drivers of risk. However, we can develop models that make use of stochastic analysis where meaningful, statistical analysis where credible, and extreme scenario deterministic projections where necessary.
Risk's main drivers
Insurance risk can be broken down into three main drivers: claims volatility, catastrophic events, and misestimation. Shocking best-estimate assumptions across each of these drivers can develop a set of extreme scenario assumptions.
Claims volatility could be measured using a stochastic model. Alternatively, statistical analysis can be performed to determine variance factoring in number of lives, variations in face amount and expected claims. The variance can then be used to determine a multiple of expected claims that represents the desired degree of certainty.
There are various data points available on prior catastrophes including natural disasters, epidemics and man-made disasters such as war and terrorism.
While these data points are useful in providing a point of reference, they may or may not be predictive of future catastrophes as medical treatments, global communications and travel develop. Other environmental factors may also have changed significantly. Ultimately, a level of catastrophic shock that fits within management's risk appetite - and meets the approval of other interested parties - must be determined.
Misestimating best-estimate assumptions may represent the biggest risk life reinsurers face. Such misestimation includes not only getting current experience assumptions wrong, but also missing the long-term trend in experience. Projecting erroneous current experience might be due to lack of quality data or misinterpretation of data that is available. Missing the trend of experience may also be the result of similar data problems, but could also be due to a shift in environmental factors such as trends in medical care. As with catastrophic events, a shock to best-estimate assumptions to factor in misestimation must be made that fits within interested parties' comfort levels.
A gross premium valuation can be performed based on the extreme scenario assumptions developed for each risk driver. The result is the total economic assets needed to support the business with the desired degree of certainty.
While such an approach does not allow for absolute measurement of the 95th percentile, such a level can be targeted when setting assumptions.
Better to make explicit assumptions and know what is being covered than to employ false accuracy of a complex model based on leaps of faith. In any case a capital level has been set that can be described and demonstrated.
While it is possible to price and manage business on an economic capital basis, the realities of regulatory and rating agency capital requirements cannot be ignored. Life reinsurers have been able to use various reinsurance solutions such as retrocession to offshore captives with reserve credits supported by letters of credit. However, there are limits to the volume of business that can be supported, with demand on the rise. Recently, capital market solutions have surfaced that fund redundant reserves by providing assets to be held in trust. Capital providers and rating agencies have become comfortable with the concept of economically necessary reserves and capital (best estimate plus additional provision to ensure obligations can be met with a high degree of certainty).
Life reinsurers now hold the majority of the insurance risk on new business issued in the US. Economically and statistically, there are efficiencies in pooling risks at this level of magnitude. However, capital requirements haven't kept pace with changes in the industry. Just as C-3 requirements have changed as asset/liability modeling and analysis improved, so might C-2 requirements, with more rigorous analysis and demonstration of a more in-depth understanding of the drivers of insurance risk. Because they hold the majority of the risk and have a high level of expertise, life reinsurers will likely need to lead the way.