All financial institutions are faced with a basic business question: What is the best use of available funds? Historically, within the insurance industry, this issue of capital allocation has been addressed using some very basic projections of future balance sheets and profit-and-loss statements, with capital projections based on regulatory capital requirements.Global insurers-as well as regulators and rating agencies-are warming to the principle that capital requirements should be based on an assessment of the true underlying risks to the business. Only then can a proper decision be made as to whether one strategic option may be better than another. Without such an analysis, a company might adopt a particular strategy that, over time, eats away at capital and results in a very inefficient use of available funds.
One suggested solution is to model and gauge the risk/return profile of all the strategies available to an insurer. This way, the decision-making process is not one that seeks only to maximize return, but one that maximizes return within an acceptable level of risk.
When "return" is defined as return on the true risk-based capital for a strategy, an insurer is capturing all the elements of risk relating to the underlying business, and the definition of "risk" in the risk/return spectrum can focus on more operational objectives, such as keeping to an acceptable probability that earnings will be at a certain level. In this way, management can be sure it is making strategic decisions that fully recognize the underlying risk exposure of any particular strategy it chooses.
On the surface, the calculations required to implement this kind of process should be fairly straightforward. But in practice they are very complex.
In establishing true risk-based capital, a company must accurately project future cash flows, making assumptions about all the moving parts. Such a model may be "stochastic," incorporating the probability distribution of complex variables.
This involves generating many potential scenarios, which often is the only way to accurately model complex variables. For example, when modeling cash flows that are a function of equity market returns, the markets must be simulated using a stochastic model to simulate cash flows under many different scenarios.
To determine capital requirements, an insurer must decide how much capital to put aside to guarantee that, at least X% of the time, the company will have sufficient funds to achieve a certain financial objective.
That decision might be: Set aside enough capital so that 95% of the time sufficient funds are available to ensure there will be at least a zero surplus position after 20 years. Or, put another way, in 5% of the scenarios there will be a surplus shortfall, which is an acceptable risk to senior management.
Ultimately, this is the most important-and most difficult-of the decisions to make about capital allocation.
A good model also recognizes interactions between risks and considers the validity of assumptions made to quantify those relationships. For risks with an abundance of data, insurers can evaluate and test correlations between risks over time and various scenarios. For risks with sparse data, insurers must use other techniques to capture the business dynamics. Relationships between such risks across scenarios provide greater insight into interactions than conventional correlation factors.
A further twist on the complexity of developing a good capital model for risk management is that operational risks-such as key-person risk and tax regulation change risk-should be addressed, in addition to financial risks.
To perform these types of calculations, a highly sophisticated financial modeling system is required. Such a system should have the capability to model highly complex assets and liabilities, and project cash flows under multiple scenarios-and do that at a high speed.
The starting point for the liability model is to examine the structure of the product or products involved, and then build this same product structure on the modeling platform. All the detailed features of liabilities should be captured, such as benefits provided and options available to the policyholder.
The next step is to consider what policyholder profile to model. That might be an actual in-force file or-for pricing purposes-a series of pricing categories, such as male non-smoker, or male smoker.
Finally, an insurer should consider the items they need to make assumptions about, such as mortality and lapse rates. This is where both art and science come together in the modeling process. Considerable judgment is typically involved, backed by the rigors of sound statistical and mathematical techniques.
On the assets side, the approach is directly analogous to that of the liability model. Asset features are captured, decisions are made about the portfolio of assets-an assumed blend of assets or a portfolio of actual assets, for instance-as are projection assumptions, such as investment return and defaults or downgrades.
At this point, the liabilities and assets are brought together in the modeling platform engine, which projects liability and asset cash flows for the modeled business for the various scenarios.
A full-blown stochastic model involves 100, 1,000 or 10,000 scenarios. The number and the detail of the model is a function of how many scenarios are considered necessary to capture the tail end of the risk distributions involved.
Such processes obviously can involve huge computational power. In these cases, many companies use a distributed processing framework, such as grid computing. With distributed processing, large jobs are split into smaller, separate processes, which can then be run on separate computers simultaneously.
The results from these distributed processes are automatically combined to deliver the final results in a fraction of the time it would otherwise require. By sharing the workload among multiple computers, large-scale enhanced financial models suddenly become more practical.
In addition to technology requirements, insurers should consider how to measure whether investing in developing and using such a framework is genuinely adding value to the organization.
One way to do this is to set up comparison benchmarks. For example, an insurer can compare actual results against results that would have been achieved using rejected strategies, after allowing for the cost of capital.
The ultimate aim of rigorous risk and capital management is to create value. Management can do this by using sophisticated actuarial modeling techniques to accurately measure the amount of capital it needs to support its portfolio of risks, and then construct an optimal capital structure to minimize the cost of capital-not just in absolute terms, but relative to the price of risk it bears. In short, value is created when the return on risk exceeds the cost of capital.
Much of the value of doing true risk-based capital assessment is not be easily quantified. But having this kind of information available to senior management can help in the decision-making process-even if it's not the primary driver behind a decision.
For this reason, the benefits of risk-based capital assessment aren't only financial. Risk-based capital management also reflects good corporate governance and sound management.
Tony Dardis is the North American practice leader for the Tillinghast Software Solutions practice within the Tillinghast business of Towers Perrin.
Register or login for access to this item and much more
All Digital Insurance content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access