For the industry, the overall net private passenger auto insurance loss ratio was 58.0 in 2004, down from a high of nearly 71 in both 2000 and 2001. The individual components of the line--liability and physical damage--showed similar improvements. (See bar chart, page 21).

For the top 25 insurers (ranked by direct written private passenger premium volume in 2004), only one--Zurich Insurance--reported worse overall direct and net (of reinsurance) private passenger loss ratios in 2004 than in 2003. And every single one of the top 25 insurers reported better combined ratios in 2004, with more than half posting combined ratios of 95 or less.

Across the industry, the aggregate net combined ratio--which reached a high of 111.4 in 2000 during the six-year period analyzed by NU--fell under 100 for the first time in 2003 (coming in at 98.6), and dropped another 4.5 points to 94.1 in 2004.

While the 13-point loss ratio drop since 2000 drove most of the combined ratio improvement, the loss adjustment expense ratio also improved by one point. In 2004, the ratio of LAE to net earned premiums for private passenger auto was just over 12, compared to 13 in 2000 and 2001.

While there is a two-part explanation for the lower loss ratios in recent years--higher prices and lower loss levels--the meaningful improvement in loss trends has been the focus of analysts and market participants. As illustrated on the graph on page 21, industrywide direct losses for personal auto, which grew 11 percent in 2000 and 9 percent in 2001, began to flatten in 2002--growing only 1 percent in that year. The low-growth trend continued in 2003, and actually reversed in 2004, with losses falling 2 percent compared to 2003.

Frequency Trends Start To Flatten

Experts point to declines in claims frequency (the number of claims per car) to explain the favorable loss trends. And there are several factors behind the falling frequencies, Michael LaRocco, president of product, underwriting and claims for Seattle-based Safeco Insurance Companies, told NU during an interview in New York at the Independent Insurance Agents and Brokers of America annual conference last week:

o The aging of the population

o Smart cars

o Drunk-driving laws

o Teenaged-driver restrictions

o Better highways

o Increased deductibles

With the baby boomer generation coming of age over the past 10 years, there are more people aged 40 to 60 who are better drivers, Mr. LaRocco said. Meanwhile, "global positioning systems and electronic stability controls are clearly making a difference," he added.

Mr. LaRocco said graduated driving restrictions have "been a huge factor" pushing down claims. He explained that these restrictions limit the amount of time that young people can drive at certain times of day and require adult passengers.

Commenting on the impact of increasing deductibles, he said: "In my father's day, people had $0 or $100 deductibles. When I came to Safeco [in 2001], over 80 percent of the customers had $250 or less. Now over 80 percent have $500 deductibles, eliminating the lower impact-type claims."

Although he said that a negative frequency trend remains evident, he doesn't see this continuing: "It has really started to flatten, and that's a clear indication [that] it will start turning slightly positive."

As for the notion that higher gas prices might further improve trends, Mr. LaRocco said, "Historically, we've never seen that."

"Having said that, this is a unique spike," he added, noting that thresholds like $3 per gallon for gas, and $50 to fill a tank, are "outrageous numbers" that haven't been studied. Intuitively, "I can see this affecting seniors who can choose not to drive, or young people who can take a train," he said.

However, he added, "the families are going to still have to take the kids to school," predicting a "short-term blip" that could cause a temporary slowdown in claims. "But they'll ultimately be moving slightly positive."

Severity: The Next Story?

More than a month before Hurricane Katrina sparked the latest hike in gas prices, Glenn Renwick, president and chief executive officer of Progressive Corp., said the personal auto trend to keep an eye on was severity. While he didn't predict any dramatic upsurge in auto insurance prices, Mr. Renwick did say that pricing would likely respond to shifting loss trends, which show increasing severity outstripping declines in frequency.

"I think we're probably going to enter into a period, in some number of months--it's certainly not on the immediate horizon, but I don't think it's more than a year-and-a-half [away]--where we will see a little bit more positive-type rate action in the market to respond to some of the trends," he said.

The remarks came during a conference call held a day after the Mayfield Village, Ohio-based insurer filed its second-quarter 2005 report with the Securities and Exchange Commission, revealing a 2 percent increase in income and a 7 percent jump in net premiums for the quarter.

Referring to "positive" and "plus" trends throughout the call, what he was describing was an environment that will initially be less favorable for auto insurers, which will suffer compressed profit margins as loss costs tick upward. Ultimately, however, it will be less favorable for consumers, now enjoying the benefits of competitive pricing.

"The real story is severity," he said, putting average severity jumps in the 5-to-6 percent range. He said that such percentages are "not extreme by any stretch" and also "not unexpected in our industry." But they are notable in that they come after a period of declines in overall loss costs (the product of frequency and severity trends together).

At Safeco, Mr. LaRocco is reading similar tea leaves. "We've seen severities generally in that mid-single-digit positive area, and generally it's been offset by frequencies being down. But there's no question that we see a continuing trend in severity going up," he said. "The main reason is medical inflation. The biggest severity jumps to date have been in bodily injury claims, and medical inflation is clearly the number-one driver there."

While some analysts worry that worsening loss trends--together with price competition--could spell trouble for auto insurers, Mr. LaRocco said he hasn't seen evidence of irrational market behavior. "We are seeing rate decreases. I think there is clearly a trend to take some rates down. [But] we're not seeing any across-the-board significant rate declines like the industry has been prone to do in the past," he said.

He said aggressive behavior is confined to three areas, with advertising being the most pronounced, followed by incentives to agents and a loosening of underwriting guidelines. Insurers might be paying on a per-application basis, or giving away trips to encourage agencies to choose them, he said.

On the underwriting front, he said, a risk previously thought of as standard by an insurer might now be classified as preferred--or a risk that used to be thought of as non-standard would now be considered standard.

Morgan Keegan, a Memphis-based equity research firm, also doesn't see a return to unsound price competition of the late 1990s. "The adoption of risk segmentation/price-tiering by some of the laggard giants in the industry--particularly State Farm--argues for an incremental measure of rational competitive behavior going forward," the firm said in a report published last week.

Sophisticated Models

Mr. LaRocco, who touts the use of "automated multivariate underwriting and pricing models" at Safeco, said there's a wide range of sophistication around the concept of tiering.

"You can have a simplistic underwriting model that takes several factors, adds them, scores people from zero to 100, splits that into 10 tiers and say you have tiering. But it's very simplistic," he said, contrasting multivariate models that examine how factors correlate with one another.

Giving an example, he said insurers historically surcharge a driver that has an accident, but "if you've had an accident and you're in your 20s, your future behavior is different than if you're in your 50s. That's multivariate--taking those two factors, correlating them together and charging the 50-year-old driver differently."

Putting Progressive and Allstate among the best competitors using such models, he said multivariate models can rely on traditional factors--violations, age, gender, marital status--and new ones, such as lapse in insurance, education and occupation.

While many models consider credit scores, he said they aren't broken when insurers lose credit-scoring battles in various states. "We have models that include credit, and models that don't," he said, noting that there is no issue in states where there's a level playing field.

In some states, however, direct writers that don't use credit for underwriting might only solicit to risks with good credit scores. "They gain a competitive advantage, and to independent agents, that's just unfair," he said.

Page 14==Top 25 Chart

Flag: Second Straight Industry Profit

Head: Private Passenger Automobile Results--Top 25 Writers ($000)

With an overall industry combined ratio of 94.1, 21 of the 24 largest private passenger auto insurers recorded underwriting profits in 2004, and the industry posted its second-straight full-year underwriting profit. Although the largest auto writer, State Farm, did not grow last year, double-digit growth at Progressive and GEICO (Berkshire) helped push industrywide growth to about 4.0 percent.

Page 20==Bar graph "Net Pure Loss Ratios"

Flag: Results Improve

Head: Private Passenger Net Pure Loss Ratios

The industry's net pure loss ratio for private passenger auto has fallen nearly 13 points since 2000 and 2001. Both the liability and physical damage components improved over the period.

Page 21==Bar graph "Private Passenger Direct Industry Data"

Flag: Losses Drop

Head: Private Passenger Direct Industry Data

The line plotting private passenger loss ratios has gone in one direction--down--since 2001. Even as premium growth began to subside in 2004 (falling to 4.7 percent from 9.0 percent a year earlier), lower claims frequencies meant continuing favorable loss trends. The level of incurred losses actually fell 2.4 percent in 2004, compared to 2003.