Favorable Loss Development Boosted 2006 Underwriting Profits

In 2006, U.S. property-casualty insurers did something that seemed next to impossible nearly a decade ago, collectively posting a combined ratio more than seven-and-a-half points under breakeven--or 92.4 overall. An NU analysis found that a combination of factors converged in 2006 to produce a combined ratio that experts say was the best since the 1930s.

A 92.4 combined ratio seemed well out of reach back in 1998, when experts were questioning the sustainability of a 1997 combined ratio that was nine points higher. In fact, it wasn't even on anyone's radar screen at the end of 2005, when analysts polled by the Insurance Information Institute predicted that the 2006 combined ratio would come in at 98.

Since posting a 101.6 combined ratio in 1997--back then touted as the best in two decades--the industry has had some good years, coming close to an underwriting profit in 2003 and posting a 98.6 combined ratio in 2004.

The overall industry combined ratio, however, didn't dip below the high-nineties until last year's remarkable 92.4 hit the books--and even the impact of four hurricanes in 2004 and Katrina in 2005 don't fully explain the stark contrast between 2006 and the immediate prior years.

Removing roughly eight points of catastrophe-related losses in 2005, or four points from 2004, and adding back a more normal catastrophe load of three points would put the industry combined ratio at 95.9 in 2005 and at 97.6 in 2004--still more than 3.5 points worse than the result for 2006 (which had a comparable three points of catastrophe losses).

The assumed combined-ratio impacts of catastrophes are based on dollar figures released by the Insurance Services Office and the Property Casualty Insurers Association of America in reports of year-end net financial results for U.S. private insurers for each of the last three years.

These reports indicate the following net catastrophe loss totals: $12.9 billion in 2003, $16.5 billion in 2004, $33.6 billion in 2005 and $11.7 billion in 2006.

While catastrophe loss totals for 2003 and 2006 differ by just $1.2 billion, pre-tax underwriting profits have climbed $37.5 billion since 2003--to a $31.9 billion profit in 2006, compared with a $5.6 billion underwriting loss in 2003.

So, what propelled industry underwriting profits to new heights in 2006 that weren't achievable in 2003, even with the push of hard-market price hikes and a similar level of catastrophe losses?

In conjunction with NU's annual ranking of p-c insurance groups and insurance companies--attached to this story--we undertook an analysis of the surprising result, and found that no single trend or event explains how the industry managed to achieve a 92.4 combined ratio.

Among the factors cited by industry experts:

o Loss reserve takedowns for prior years.

o The absence of mega-storm losses.

o The impact of exposure management and loss mitigation activities.

o Favorable loss-cost trends in most lines of business.

o The residual impact of price increases from a recent hard market.

One such expert is Robert P. Hartwig, president of the Insurance Information Institute, who cited 1935 as the last comparable year in a written commentary of year-end 2006 results published in April.

No single line or category of business (property or casualty, personal or commercial) by itself fully explains the result, either. As illustrated on the graphs accompanying this article, aggregate combined ratios for the industry improved for every major line of business since 2003.

In addition, our premium rankings--which include combined ratio results for the top 100 groups and top 100 companies for the last two years, and for the top 20 groups dating back to 2003--reveal nearly universal combined ratio improvements for individual market participants in recent years.

(One notable exception stands out among the top-20--Madison, Wis.-based American Family, whose results were severely damaged by Midwest storms.)

While experts have been discussing the positive impacts of post-9/11 price hikes, favorable auto claims frequency trends, and the reforms in federal class-action and state tort laws for several years, only recently have reserve takedowns officially joined the benevolent conspiracy--emerging as a leading factor to explain the industry's stellar combined ratio for 2006.

A combined ratio is essentially the sum of a loss ratio (the ratio of incurred losses--including loss-adjusting expenses--to earned premiums) and an expense ratio (which compares underwriting expenses such as commissions to written premiums).

Incurred losses used in the combined ratio calculation for any given calendar year consist of losses paid and reserves set up for claims that occurred during that year (current accident-year incurred losses) and changes in loss reserves for claims that occurred in prior years.

Nearly every chart and graph in this issue hints at the dramatic impact of this second component on calendar-year incurred losses.

Insurer loss-reserving practices produced $7 billion of reserve releases for prior accident years during calendar-year 2006, shaving 1.6 points off the industry combined ratio. Among the charts providing a further glimpse of this impact are:

o A schedule of combined ratios for the last four years for the top-20 insurance groups (based on 2006 net written premium volume), revealing abominable combined ratios for large insurers like Hartford and CNA in 2003, and their much improved results in 2006. The 2003 ratios were the products of more than $2 billion of reserve hikes for each group relating to losses incurred in prior years.

o A year-by-year comparison of property and casualty pre-tax underwriting results, revealing that while property underwriting profits made up two-thirds of the industry's $31.9 billion pre-tax underwriting profit in 2006, more than 70 percent of the $37.5 billion improvement in underwriting results since 2003 came from casualty lines, which were heavily impacted by reserve changes.

o A direct analysis of Schedule P--the annual statement exhibit that displays incurred loss-development histories--indicating that while p-c insurers collectively added almost $14.1 billion to reserves for losses incurred in prior years during 2003 (adding 3.6 points to the 2003 combined ratio), during 2006 insurers took down prior-year reserves by roughly $7.1 billion in total.

The difference--more than $21 billion--explains 5.2 points of the 7.7-point difference between the 2003 industry combined ratio of 100.3 and the 2006 ratio of 92.6.

A similar analysis for 2004 reveals that reserve changes for losses incurred in prior periods accounted for 4.1 points of the six-point combined ratio difference between the 2004 industry combined ratio (98.6) and the 2006 combined ratio.

During 2006, while insurers took down reserves for losses incurred in 2003-2006 by more than $16.5 billion, they continued to add to reserves for problematic accident years 1997-to-2001. The needed additions for these problem years, however, have tumbled from $11.9 billion to only $4.4 billion in 2006.

But reserve takedowns don't always produce underwriting profits, as an analysis of Schedule P developments during 1997 makes clear.

During 1997, p-c insurers collectively lowered their loss reserves by $9.9 billion--even more than the $7.1 billion takedown recorded for 2006--taking 3.5 points off the 1997 combined ratio. But the combined ratio still came in nearly two points above breakeven--at 101.6. In dollars, the industry suffered a $5.9 billion pre-tax underwriting loss in 1997.

At the time, analysts acclaimed the result as the best in decades. But they also declared it an anomaly, citing a low level of catastrophe losses compared to recent prior years. In addition, they worried that reserve redundancies would soon run out and a softening market was putting premium growth out of step with loss cost inflation (see NU, April 6, 1998; Jan. 5, 1998; and Dec. 22, 1997).

While levels of reserve changes and catastrophe losses, as well as soft-market conditions, are common threads between 1997 and 2006, experts in 2007 believe there's more room to take profits from conservative reserve levels carried for recent accident years (see related sidebar, "Still Fat?") and that prices are adequate (see "Fitch Analysis," page 12).

The dual impacts of premium hikes to adequate levels in the intervening years and favorable trends in loss costs help to explain why combined ratios in every major line of business were better (lower) in 2006 than they were in 1997.

For a line that had one of the worst combined ratios in 1997--other liability--the accident-year 1997 loss ratio was about 80. With competitive pricing persisting and unexpected tort issues (like construction defect claims) emerging, initial loss ratios in this line continued to hover in the high-70s and low-80s--and ultimately settled out over 100--for three of the next four accident years.

Since accident-year 2001, however, earned premiums have more than doubled in this line, while losses have risen only 36 percent over the same period, pushing accident-year loss ratios down near 60 for the last three accident years.

(The 36 percent increase was calculated using losses that include all developments through 2006 for each accident year. The increase is 64 percent if year-end losses--evaluated 12 months after the start of each accident year--are used instead.)

Analyses of the most profitable lines, such as auto physical damage, reveal similar trends.

After a period of price competition early in the decade, earned premiums in the private-passenger auto physical damage line have grown 24 percent since 2001, while losses have risen less than 3.0 percent.

The end result is a combined ratio in the low-90s, compared with 99.1 in 1997, and a high of 108.8 in 2000--when aggressive price cuts were described on the pages of NU (see Aug. 21, 2000, page 3).