Managing Catastrophe Risk

May 1996
Executive Summary

Nineteen tropical storms occurred in 1995, more than in any year since 1993. Fortunately, most of 1995's storms did not make landfall. But three that did — Hurricanes Erin, Marilyn, and Opal — contributed to the year's $8.3 billion in total insured catastrophe losses and helped make 1995 the fourth costliest year ever for catastrophes in inflation-adjusted dollars.

The recent surge in catastrophe losses has caused insurers, regulators, legislators, and others to question whether the property/casualty insurance industry has the financial capacity to handle the growing catastrophe risk. In an attempt to study this potential problem, many insurers are using computerized catastrophe models to estimate possible catastrophe losses.

However, "the industry" does not provide insurance. Individual insurers do. To analyze the insurance industry's financial capacity to handle catastrophe risk, one must study each insurer. This study [1]1. 1. The study quantifies an insurer's overall catastrophe exposure using the ratio of its annual expected catastrophe losses to its surplus.takes a first step in that direction.

The study uses a computerized catastrophe model to analyze insurers' exposures and determine the potential effect of earthquakes and hurricanes on the financial strength of 80 insurer groups, representing 28% of the industry's property insurance premium.

The model includes thousands of simulated earthquakes and hurricanes, each with an associated probability of occurring in any given year. ISO estimated the losses for each simulated catastrophe and combined the results to get the distribution of annual catastrophe losses for each insurer group. This study compares the model-generated losses with each insurer group's surplus.

Since many property reinsurance programs are specifically designed to protect insurers from catastrophic loss events, this study also analyzes the effect of reinsurance. ISO adjusted the model's estimates of direct property losses to take into account the effects of reinsurance programs.

The study examined the probability that annual losses will exceed specific percentages of surplus. (See Table 1.)

• Net of reinsurance, 73 of the 80 insurer groups studied manage their exposures well enough that the chance of insolvency because of catastrophes in any given year is less than 1%.
• Net of reinsurance, for the remaining seven groups, the program of insolvency from catastrophe in any given year is greater than 1% but less than 5%.

In addition to analyzing the probabilities that each insurer group's annual catastrophe losses exceed specific levels of surplus, the study also examined the expected percentage of each group's annual catastrophe losses to exceed specific levels of surplus. These percentages, referred to as excess pure premium ratios (EPPRs), provide a better measure of the insurer group's loss volatility and the threat to surplus. The EPPR reflects the combination of the probability of large events as well as the amounts by which losses could exceed a specified amount. according to the study's findings:

• Catastrophe loss volatility, measured by the EPPR, differs markedly by insurer group, even among groups with similar catastrophe exposure.
• Potential events could cause a catastrophe of $50 billion or more for the industry. Based on current reinsurance levels, and assuming that all reinsurance is collectible, these events could result in insolvency for up to 36% of all insurers, depending on the geographic location of the event. Unfunded claims for these insurers could be as much as $56 billion.

The computer model showed that some of the largest catastrophe losses to the industry resulted from simulated hurricanes hitting a broad expanse of the eastern seaboard — including, in particular, states in the Northeast. Other large catastrophe losses resulted from simulated earthquakes centered in the New Madrid seismic zone.

State and national catastrophe programs could provide some relief. The study concludes that pooling mechanisms are significantly more effective when the geographic spread of participating insurers is broader and when the pool covers more perils. For example, a national multiperil plan is more effective than single-state or single-peril plans. ISO constructed sample plans based on the 80 insurer groups.

Based on these hypothetical plans, the study found that over the long term, under a national earthquake and hurricane pooling plan, the percentage of annual catastrophe losses exceeding four times the expected annual catastrophe losses would be half as much as under a national single-peril plan, and 35% as much as under a single-state, single-peril plan. (See Table 2).

The infrequency and high severity of catastrophes contribute to insufficient capital in the property/ casualty industry to absorb losses from megacatastrophes. The traditional methods of dealing with large losses from catastrophes, such as reinsurance and guaranty funds, are also inadequate. Individual insurer actions to limit their exposure to catastrophe losses have led to availability problems for insureds in high-risk areas. Solutions to the shortage of surplus to manage catastrophe risk — and to availability problems — will require access to capital from outside the industry.

• Geographic concentration and diversification account for most of the differences in EPPRs — and in the probability that losses will exceed a given percentage of surplus — among insurers with similar overall catastrophe exposure. Figure 1 shows that an insurer group with a geographically concentrated book of business can have one-third the probability of suffering catastrophe losses greater than 10% of surplus, compared with an insurer group having a more geographically diversified portfolio. [2]2. Both groups have similar overall catastrophe exposure, relative to their surplus. And, for the geographically concentrated insurer group, the percentage of expected catastrophe losses in excess of 10% of surplus is two and a half times greater than for the geographically diversified group. The geographically concentrated insurers are exposed to a lower number of catastrophes, each with high severity, whereas the geographically diversified insurers are exposed to a higher number of catastrophes, each of low severity.
• Many insurers could, a least theoretically, reduce the volatility of their catastrophe losses. The study found that 13 of the 80 insurer groups could reduce their EPPRs at least 50 percentage points by mirroring the geographic distribution of catastrophe exposures for the 80 groups. Similarly, an additional 18 insurer groups could reduce EPPRs between 25 and 50 percentage points.

 Figure 1:
A Geographically Concentrated Insurer Group Compared with a Geographically Diversified Insurer Group

Pooling is one strategy for an insurer to mirror, at least partially, a more dispersed geographic distribution.[3]3. Insurance pooling is the practice of sharing and distributing risk among (primary) insurers. Normally, two or more insurers will divide the premium for a policy or a set of policies into pre-agreed-upon portions and then contribute to paying any losses on these policies in the same proportion. However, even under a pooling arrangement, catastrophes can still inflict unacceptable losses on individual insurers.

Other study conclusions are as follows:

• Catastrophe reinsurance is not a significant factor in guarding against insolvency for larger insurers. The average ratio of catastrophe limits to surplus is 14% for the 20 insurer groups with the most surplus, but the ratio is over 100% for the 20 insurer groups with the least surplus. Here are some reasons for these catastrophe limits:
1. Reinsurers provide catastrophe insurance in fairly standardized layers, with the upper limits of these layers representing higher percentages of surplus for smaller insurers than for larger insurers.
2. Large catastrophes are rare; therefore, an insurer that does not have high catastrophe limits is likely to survive.
3. Worldwide reinsurance capacity is also low. Nonproportional reinsurance coverage against natural disasters in key markets was $46.8 billion in 1995, with the United States accounting for a 36% share.
4. An insurer willing to pay the price of sufficient catastrophe insurance could have trouble competing for business.
• The volatility of catastrophe-prone lines limits the amount of exposure an insurer can safely write. In the wake of recent catastrophes, insurers are reexamining the extent of their exposure in catastrophe-prone areas. Several insurers have attempted to limit the risk to their surplus, leading to local availability problems for insureds.

1. The study quantifies an insurer's overall catastrophe exposure using the ratio of its annual expected catastrophe losses to its surplus.

2. Both groups have similar overall catastrophe exposure, relative to their surplus.

3 Insurance pooling is the practice of sharing and distributing risk among (primary) insurers. Normally, two or more insurers will divide the premium for a policy or a set of policies into pre-agreed-upon portions and then contribute to paying any losses on these policies in the same proportion.