VFIC Paper Suggests Duration Matching Is "Just One of Many" Risk Minimizing Strategies

The CAS Valuation, Finance, and Investments Committee (VFIC) recently completed a paper entitled "Interest Rate Risk: An Evaluation of Duration Matching as a Risk-Minimizing Strategy for Property/Casualty Insurers." This paper will be published in an upcoming issue of the Forum, and will be presented at several CAS and other meetings throughout 2002. The paper applies modern dynamic financial analysis (DFA) techniques to the evaluation of alternative investment strategies available to insurers.

VFIC tested the hypothesis that, compared to any other investment strategy, matching a company's asset and liability duration will optimize the risk profile of the company. Although the results varied by scenario, the overall conclusion was that duration matching does not stand out as a clearly optimal strategy for property/casualty insurers. Duration matching, in fact, was frequently just one of many optimal strategies from which the company had to choose based on its desire for return and appetite for risk.

VFIC decided to address this topic in order to shed more light on an often debated but sometimes misunderstood topic. In doing this, VFIC is building upon earlier work done by other committees. These other studies were limited by the fact that their committees did not have access to modern tools designed to support DFA. In particular, the former Financial Analysis Committee of the CAS did work on this subject that was never published in detail because of difficulty in validating the results.

Duration matching of the asset and liability portfolios has been advocated by many as the preferred investment strategy for property/casualty insurers. Duration measures the weighted average time to maturity of a particular investment portfolio, usually a group of bonds. If liabilities are discounted by current interest rates, then (all else equal) the value of the duration-matched liabilities and assets will both decrease when interest rates increase; as a result, surplus is theoretically insulated.

A duration-matched strategy, however, can reduce the insurer's income. This arises from the short duration of the liabilities of most property/casualty insurers, and the lower investment income that would normally result. Thus duration matching has a cost. Analysts find that most property/casualty insurers invest in portfolios with durations longer than those of their liabilities, suggesting that those insurers have concluded that the duration hedge's value is not worth the cost.

Life insurers have often used a matching strategy as a benchmark, but the duration of a typical life insurer's liability portfolio is much longer than a property/casualty insurer's, which greatly reduces the cost of duration matching. Life insurers also discount many of their liabilities, increasing the hedge's value. Finally, because most life contracts provide fixed amounts of benefits, the only risk (other than mortality) is interest rate. So for life contracts, duration matching is more likely to be optimal. But some regulators and other analysts have assumed the strategy should apply to property/casualty insurers as well. Accordingly, regulators have at times developed proposals that would penalize property/casualty insurers who do not use a matching strategy. Thus VFIC's interest in this topic arose to some degree from a corresponding interest among regulators.

VFIC ran its DFA model for a workers compensation carrier (with long-tailed liabilities) and for a homeowners carrier (with short-tailed liabilities, but with catastrophe exposure causing an increased risk that the invested assets would be liquidated before maturity). Although different in some details, the overall results from both companies were quite consistent.

VFIC noted that the results of the analysis were strongly influenced by the accounting convention selected. For example, statutory results showed greatly reduced asset risk because of amortized cost accounting. Therefore, on a statutory basis the longer investment strategies often yielded higher return with lower risk.

Furthermore, the risks were distorted under both statutory and GAAP accounting because the liabilities are not discounted. VFIC is considering a future research project in which "economic value" (including discounted losses) will be used as the accounting convention; this final test of whether duration matching holds any benefit for property/casualty insurers would be the subject of a subsequent paper.

One fact was undeniable: On average, long duration strategies yielded higher returns than duration-matched strategies. So, although duration-matched strategies may in some cases be less risky, they are also less profitable. Under traditional risk/return analysis, this supports the argument that both are optimal strategies that differ primarily in their position on the risk/return curve. As such, VFIC found that matching durations is reasonable, but not "better" than longer strategies.

A copy of this paper can be found in the Research Section under Committee/Task Force Projects. VFIC welcomes comments from the membership and the general public on its paper, as well as any comments on this overall topic. Please direct any remarks or inquiries to the author of this article at (952) 897-5300 (e-mail: ken.quintilian@milliman.com), or contact the 2002 VFIC Chair, Paul Brehm, who can be reached at (651) 310-4800 (e-mail: paul.brehm@stpaul.com).