Since the Great Recession, pricing personal auto is no longer the same.
Developments that took place during the Great Recession and its subsequent years have forever transformed personal auto insurance pricing cycles.
“The underwriting cycle as we knew it does not exist as strongly anymore,” said Roosevelt C. Mosley Jr., a principal and consulting actuary with Pinnacle Actuarial Resources. “Given the increased granularity of pricing and the increased speed of data analysis, I do not believe we will see severe swings in profitability that were present in the ’80s and ’90s,” he added.
The rise and expansion of predictive modeling and big data are just two of several trends that coincided with or were caused by the Great Recession that require actuarial consideration. And since auto insurance actuaries often spearhead innovative approaches later adopted by other lines, watching how they address new challenges in the current cycle’s new environment is critical. “Actuaries are wrestling with underlying trends that are definitive,” Mosley said.
There are many trends to watch. For example, some developments reduced frequency while others increased it. Manufacturers are producing safer cars while also boosting driver distraction with in-car access to mobile technology and infotainment systems. Driving under the influence of alcohol has been steadily declining while marijuana-affected driving is on the rise.
With pure premiums increasing due to a dramatic uptick in frequency, which were already fueled by rising costs per claim (severity) amounts, insurers are requesting rate increases. “In general, the number of filings being submitted by companies has surged recently as [frequency and severity] trends increased within the last 12-18 months compared to recent history,” said Paul D. Anderson, a principal and consulting actuary for Milliman.
Considering the multiplicity of trends and new unknowns that have accelerated since the Great Recession, insurers are requesting rate increases in an environment of greater scrutiny. The advent of price optimization has been met with much criticism.1 There is also the Federal Insurance Office started by the Great Recession-inspired Dodd-Frank Wall Street Reform and Consumer Protection Act, which monitors potentially unfairly discriminatory rating factors.2
In the midst of abundant considerations, personal auto insurers are also concerned about the financials. A low interest rate environment, coupled with a fiercely competitive market, continues to challenge profitability — the very incentive for selling auto insurance in the first place.
Beginning with frequency, ample evidence demonstrates the close relationship between employment numbers and accident rates, said James Lynch, chief actuary and vice president-data and information services of the Insurance Information Institute (III).
According to analysis by the III, when looking at collision coverage, claim frequency rises and falls with the job market (See Chart 1).
“As employment recovered, claim frequency had risen, almost in lock-step,” Lynch said.
By the third quarter of 2015, employment increased to 149 million workers and collision frequency had risen 5.97 per 100 vehicle-years. Both were higher than pre-Great Recession numbers.
Lynch pointed out that the number of miles driven falls during a recession, since a laid-off worker has no job to drive to. Frequency falls as well. When the recession ends, miles driven rises again; frequency does too.
As frequency has increased, so have traffic fatalities. The National Safety Council (NSC) preliminarily estimates that motor vehicle deaths jumped eight percent in 2015 from 2014, which marks the largest year-over-year percent increase in 50 years. For the year 2015, the NSC estimates 38,300 people were killed and 4.4 million were seriously injured, likely making 2015 the deadliest driving year in the United States since 2008.
Driving under the influence of alcohol has been steadily declining while marijuana-affected driving is on the rise.
Similarly, NHTSA reports that traffic deaths in 2015’s first nine months were 9.3 percent higher (26,000) than the first three quarters of 2014 (23,796).
But there is good news. In the long term, as autos and highways get safer, Americans clock a greater amount of miles but the number of crashes continues to decline, Lynch noted.
Other Frequency Factors
Other trends and new developments since the period beginning with the Great Recession are reducing crashes while others stand to increase them.
Vehicles have become safer. Front crash prevention, lane departure warning, blind spot detection, adaptive headlights, park assist and backover prevention are boosting vehicle safety, according to “Crash Avoidance Technologies,” on the Insurance Institute for Highway Safety (IIHS) website. Electronic stability control, which became standard in 2012, lowers the risk of a single fatal vehicle crash by about half and risk of fatal rollover by as much as 80 percent.3
“Another positive development is graduated driver licensing (GDL),” Anderson of Milliman said. All states and the District of Columbia have a three-stage GDL system. Adopting GDL laws will lead to “substantial” reductions of crashes for this age group — from 20 percent to 50 percent.4
Conversely, new trends that began during the period starting with the Great Recession are also contributing to higher frequency. “There are certainly more distractions than five or 10 years ago,” Lynch observed. “It is possible that automobile crashes might have decreased even more if it had not been for the increase in distracted driving,” he added.
Consider mobile devices. “Technology has gotten so small and portable; the timing happened to line up with the Great Recession,” Anderson said. Before the Great Recession, mobile phones had already become ubiquitous and texting while driving was already a public safety concern.
Fifty percent of respondents to the NSC’s “Distracted Driving Public Opinion Poll,” released in March 2016, believe infotainment dashboards and hands-free technology must be safe if the auto manufacturers installed them.
Then Apple’s iPhone introduced smartphones with irresistible consumer appeal in 2007. For the first time, consumers could easily access the internet and enjoy apps at a finger touch, helping to fuel smartphone adoption. Deemed the fastest growing technology in history, 68 percent of Americans owned smartphones in July of 2015, up from 35 percent in 2011, according to Pew Research Center (PRC) numbers.
Useful apps that can direct motorists away from congestion and accidents, such as Google’s Waze — a Global Positioning System (GPS) app featuring real-time traffic navigation with gamification can inspire drivers to pay more attention to traffic patterns than the actual road ahead.
Auto manufacturers have also boosted attention-diverting features, Anderson said, including screens offering climate control, audio control, maps, directions, summaries of vehicle performance or trip information. “While these types of controls may be similar to [those found on] older generation vehicles, presenting them on a high-quality screen with multiple pieces of information visible or available seems to increase the tendency to distract the driver,” he added.
One in four car crashes involve cellphone use, according to the NSC’s 2015 edition of Injury Facts. Since automobile manufacturers have also boosted in-car access to mobile technology and infotainment systems, however, these added distractors also need consideration. As a result, the NSC has suspended calculating its estimates on cell phone crash prevalence until there is more data on the impact of in-car access to mobile technology and in-vehicle infotainment systems, explained Kelly Nantel, the NSC’s vice president of communications and advocacy. “We want to make sure there is more data to really understand the risk,” she added.
Many consumers do not realize these modern systems can be just as distracting as cell phones. Fifty percent of respondents to the NSC’s “Distracted Driving Public Opinion Poll,” released in March 2016, believe infotainment dashboards and hands-free technology must be safe if the auto manufacturers installed them.
There are also signs that frequency is going up due to the growth of states relaxing their marijuana laws since 2008 even though the federal government classifies it as a Schedule I (illegal) drug. Of the eight states with the largest increase in auto accident frequency, seven have liberalized their marijuana laws, according to research by the Property Casualty Insurance Association of America.5
NHTSA’s “Results of the 2013-2014 National Roadside Survey of Alcohol and Drug Use by Drivers” reports that the number of weekend nighttime motorists with marijuana in their systems was nearly 50 percent higher in 2014 than in its 2007 survey. Marijuana users are about 25 percent more likely to be involved in a crash compared to those not under the influence, according to NHTSA’s “Drug and Alcohol Crash Risk” study, released in 2015. The agency is aware of the growing concern with possible marijuana-related fatalities in crashes and intends to include it in the future fatality analysis reporting.
Fatalities from driving under the influence of alcohol during the weekend nighttime, meanwhile, declined by nearly one-third since 2007, according to NHTSA’s “Roadside Survey.” The federal agency’s “2014 Motor Vehicle Crashes: An Overview,” released in March 2016, reports that 31 percent of vehicular deaths in 2014 (9,967 out of 32,675) were caused by driving while under the influence of alcohol, which is lower than previous years.
Speeding continues to be a top cause of fatalities, and yet, states have continued to raise speed limits. In 2013 alone, fatalities from higher speed limits resulted in 1,900 additional deaths, essentially canceling out the number of lives saved by frontal airbags that year, according to an IIHS report.6 Today, six states have 80 mph limits, and drivers in Texas can legally drive 85 mph on some roads, according to the IIHS report.
Demographic changes might also contribute to rising frequency. During the period starting with the Great Recession, baby boomers were still in the safest driving age, the 50s and 60s, but around 2014, they started entering older ages when driving becomes more risky, Anderson said. Meanwhile, research by Highway Loss Data Institute (HLDI), reveals that from 2012 to 2014, more teenagers — the least experienced and most risky drivers — found jobs and got on the road after a period of decline from 2006 to 2012.7
Severity also looked different in the period starting with the Great Recession, Lynch noted. “Severity normally rises faster than the inflation rate. In the years immediately following the Great Recession, the growth in severity was lower than normal. More recently, it has returned to the norm and is rising faster than inflation,” he added. (Chart 2.)
Several factors explain the increase of costs per claim. The average cost for repairing an automobile has become far more expensive due to the rising costs of parts. According to the U.S. Bureau of Labor Statistics consumer price index data, the cost of auto body repairs rose 20.9 percent from 2007 to 2015, Lynch said.
“There is so much technology embedded in the car that it is not just as simple as buying a part,” Pinnacle’s Mosley said. “That part may now have some technology imbedded in it or placed near it that will now be impacted by the repair.”
Labor cost is also more expensive. “The way cars are made now, it takes more time to get the necessary parts — that grows the labor charge,” Lynch said. “Open the hood of a car now, it is much smaller and parts are harder to get at, so if you have to replace a part it takes more time.”
The Insurance Research Council (IRC) identified other new developments actuaries need to watch, such as the growth in claim severity driven by increased utilization of medical services, said David Corum, the organization’s vice president.
The increased use of medical services goes hand in hand with growth of attorney involvement, according to IRC’s 2014 study, “Attorney Involvement in Auto Injury Claims.” After examining 35,000 closed claims for claimants with neck or back strains with fewer than 10 days of restricted activity, the study found greater utilization of chiropractic treatment, MRIs and pain clinics for personal injury protection (PIP) claims (Chart 3) and BI claims (Chart 4).
The study also found that the percentage of lawyer-represented claimants rose to 36 percent of PIP claims in 2012, up from 31 percent in 2007. For BI claims, the rate of attorney involvement has been flat, with 49 percent involvement in 2007 and 50 percent in 2012.
In 2012, PIP attorney-associated claims cost an average of $8,457 compared to $3,297 without an attorney (Chart 3). For BI claims, average cost per claim was nearly two-thirds more expensive at $9,619 with an attorney compared to $3,365 without one (Chart 4). “Since attorneys may get about one-third of claimant’s financial compensation, claimants sometimes receive a smaller net claim payment,” Corum said.
There is also a steady increase in the appearance of claim abuse, according to IRC’s study, “Fraud and Buildup in Auto Injury Insurance Claims, 2015 Edition,” released January 2015. Claim abuse is fraud, defined as material misrepresentation of an accident’s facts or loss while buildup — or “soft fraud” — is the inflation of an otherwise legitimate claim.
The appearance of abuse for paid PIP claims rose from 13 percent of claims in 2002 to 15 percent in 2007 and to 18 percent in 2012. “BI claims have had a higher level of abuse but there is not as much of a clear trend,” Corum said. Nineteen percent had an appearance of abuse in 2002; this increased to 24 percent in 2007 and lowered to 21 percent in 2012.
In the period beginning with the Great Recession, profitability for auto insurers began to significantly reduce, according to the National Association of Insurance Commissioners (NAIC).
In 2007, the combined ratio for liability and collision coverage was 101.8 and 93.4, respectively. For 2014, the combined ratio deteriorated to 103.8 for liability and 100.2 for collision, according to SNL Financial LC figures posted on III’s website.
Return on net worth for personal auto insurance also declined since the Great Recession began, according to data from the NAIC. In 2014, the return on net worth was 3.6 percent, a substantial decline from 8.3 percent in 2007, according to the NAIC’s “Profitability by Line by State 2014,”10 released in December 2015.
One factor directly related to the Great Recession is eroding interest rates by the Federal Reserve to spur economic growth. Yields for 10-year U.S. Treasury notes in 2007 were about four percent, dropping to 1.75 percent in 2008 when the severe impact of the Great Recession was getting started and even below one percent in 2009. “Yields have been essentially down five percent for a full decade,” Lynch said.
Another significant reason for lower profitability is the intense competition among personal auto insurers, which had to contain rates to maintain market share despite upturns in severity. According to the NAIC’s “Auto Insurance Database Report 2012/2013,”8 combined average premium was $954.30 for 2013 and close to $959.76 in 2004.9
Much of the rise in competition was made possible through the growth in predictive modeling. “There has been a revolution in rating variables,” Lynch said. “Credit scoring allows policies to be priced more accurately and most people benefit. Actuaries have gotten better and better at pricing and identifying who are at the greatest risk of being in an accident.”
The advent of predictive modeling occurred more than 20 years ago, but during the period beginning with the Great Recession, it transitioned from a competitive strategy to a business necessity. “[There is a] large percentage of auto premium set with predictive modeling,” Anderson said.
The growth of predictive modeling during the Great Recession and subsequent years is apparent. In 2009, when Towers Watson began its annual predictive modeling survey, 76 percent of personal auto carriers were using predictive modeling for underwriting/risk selection and/or rating/pricing. By 2015, virtually all — 97 percent — reported the same, according to Willis Towers Watson’s “2015 Property & Casualty Insurance Predictive Modeling Survey,” released in February 2016.11
Predictive modeling, Lynch said, does tend to keep rates lower because insurers can more accurately understand the risk they are taking. “That means the insurer can allocate less capital to the line, which in turn means fewer dollars of profit will get them to the needed return on capital,” he added. Predictive modeling also boosts profitability, affirmed 83 percent of the total property-casualty insurers in the Willis Towers Watson study.
Predictive modeling, Lynch said, cannot “overwhelm the spike in frequency.” However, what it can do, Mosley said, is respond more quickly and tactically to frequency and severity trends compared to prior pricing cycles.
“Actuaries have gotten better and better at pricing and identifying who are at the greatest risk of being in an accident.”
In the early 2000s, Mosley explained, the industry had to react strongly to frequency and severity triggers. “Insurers would see trends shooting up, prices would go up and they would shut off the underwriting valve and be more careful about customers,” he added.
“We are past those fluctuations. We get data quicker and we can react to things that we could not do historically,” Mosley said. “Companies are smarter and more comfortable with how they increase or decrease prices due to specific trends in the marketplace,” he added.
If anything, because insurers are “segmentation smart” with pricing and underwriting, “they cannot pull the price trigger as hard anymore due to the risk of anti-selection issues,” he explained.
While insurers brace themselves from the effect of rising premiums, there are also wildcards — future innovations becoming available that could affect future frequency and severity and potential public policy changes.
The number one wildcard is technology itself, sources agreed, and it is difficult to anticipate.
The potential growth in telematics for improving driving behavior and providing insurers with richer risk data is one wildcard that could affect personal auto insurance pricing in the future.
From a regulatory standpoint, the stage is set for telematics. All but five U.S. jurisdictions allow these devices, according to an NAIC Center for Insurance Policy and Research (CIPR) study12 released last year. Household-name insurers are offering varying telematics programs to their customers.
The results of telematics are also favorable. “Telematics can change driving behavior and that is a good thing,” Corum said. More than half (56 percent) of motorists participating in an IRC public opinion survey made changes to their driving behavior since installing a telematics device, according to the report, “Auto Insurance Telematics: Consumer Attitudes and Opinions,” released in November 2015.13
Eighty-two percent of device users said they receive information from their insurance company about their driving and of those, 81 percent said they reviewed the information and 88 percent of those who reviewed the information said they found it helpful.
But there are also hurdles to overcome. Nearly half (47 percent) of the 1,135 respondents said they were unlikely to accept a telematics device due to privacy concerns. “It means the industry cannot assume that everyone is going to allow (the devices) to be put in their cars,” Corum said. This “strong opposition” may delay the impact of telematics devices until a positive track record has been established, he added.
Meanwhile, telematics has been reflected in the prices of most large insurers, but Anderson said there is uncertainty around how long it will take for medium-sized companies to implement it, he said. “Telematics provides a large source of data, but unlike personal credit information, you can’t buy telematics data for individual policyholders,” he explained, “Insurance companies have to collect their own telematics data and copying other carriers is difficult because most of the large carriers use their own approach.”
Anderson noted that there is also the possibility that real-time feedback intended to change driving behavior has the potential to be distracting.
Another wildcard is how well insurers will be able to use predictive modeling and big data to improve pricing and insurance operations. The Willis Towers Watson survey indicates that many of the property-casualty carriers surveyed are already using big data to reduce costs from litigation, fraud and claim management or plan to do so in the next two years.
Specifically, 17 percent of respondents are currently using predictive modeling for claim triage, but more than half (52 percent) intend to do so in the next two years. While 10 percent already use modeling for evaluating litigation potential, half (51 percent) plan to do so in the same time period. “Carriers clearly see predictive models adding value across all lines of business, with personal lines continuing to lead the way,” the survey said.
There is conjecture that transportation network companies such as Uber and Sidebar could affect losses, but neither Anderson nor Mosley see a real effect. Mosley wondered if personal automobile insurers unknowingly pick up claims for drivers who do not have insurance riders or commercial auto coverage. “We now have Uber drivers behaving like taxi drivers and they do not have taxi training,” Mosley said.
Experts agree it will also take time for the American vehicle fleet to turn over once driverless cars become a commercial reality. It took 20 years for 95 percent of the cars on the road to have anti-lock brakes, Anderson said.
Driverless cars are another wildcard destined to affect the industry, but its near-term effect is difficult to predict. Technology can move fast, but how automated vehicles will be regulated and their commercial availability for the average American is uncertain.14 Experts agree it will also take time for the American vehicle fleet to turn over once driverless cars become a commercial reality. It took 20 years for 95 percent of the cars on the road to have anti-lock brakes, Anderson said.
Public policy changes are another wildcard. From the growing role of the Federal Insurance Office to the coming presidential election, even more unfathomable changes could be ahead.
The impact of the Great Recession and developments during its subsequent years forever changed the personal auto insurance landscape. Technological advancements have improved pricing and expanded data and have challenged driver behavior for the good — with telematics — and the bad — by adding more driver distractions.
During this period, public policy changes also challenged the status quo. On the federal level, the Great Recession-inspired Dodd-Frank Act introduced additional regulation and scrutiny. Economic recovery-motivated low interest rates and tight competition continue to challenge profitability.
Keeping an eye toward evolving developments while determining and focusing on relevant trends are perhaps the greatest challenges personal auto actuaries face. Since personal auto insurance is often the petri dish for actuarial innovation, how actuaries address these challenges will influence the course for other property-casualty insurance lines in the future.
Annmarie Geddes Baribeau has been covering actuarial topics for more than 25 years. Her blog can be found at http://annmariecommunicatesinsurance.com
1 “Price Optimization and the Descending Confusion,” AR, September/October 2015, http://bit.ly/1VBEG1H.
2 “Demystifying the Regulatory Web,” AR, March/April 2016, http://bit.ly/1X2Iv0W.
3 “Crash Avoidance Technologies,” IIHS, http://bit.ly/1Uftjr3 (viewed April 6, 2016).
4 “Teen Drivers — Graduated Driver Licensing,” NHTSA website, http://1.usa.gov/1spmDjY (viewed April 20, 2016).
5 “From El Niño to Legalized Marijuana, New Answers Behind the Rise in Car Crashes,” http://bit.ly/1UxfAPl, (viewed April 17, 2016).
6 “Speed Limit Increases Cause 33,000 Deaths in 20 Years,” IIHS, April 12, 2016, http://bit.ly/1pFfUAw.
7 “Teens Get Back in Driver’s Seat as Economy Picks Up,” PR Newswire, February 4, 2016, http://prn.to/1ToXy3m.
9 “Auto Insurance Database Report 2003/2004,” NAIC, http://bit.ly/1PrdKkb.
11 “P&C Insurers’ Big Data Aspirations for Advanced Predictive Analytics,” http://bit.ly/1UDLcD1
12 “Usage-Based Insurance and Vehicle Telematics: Insurance Market and Regulatory Implications,” March 2015, http://bit.ly/1P5yDOE.
14 “Destination Driverless,” AR, November/December 2015, http://bit.ly/25zxyFY.