The Rise And Fall Of WC Accident FrequencyClaims down nationally, but certain states buck the trend,complicating pooling, pricing of coverage

The decline in the rate of occupational accidents has been awell-known phenomenon of the workers' compensation marketplace ofthe past decade and has significantly colored the rates andperformance expectations for the line. In reality, however,state-specific accident frequency rates continue to divergemarkedly from the improving national trends in terms of level,stability and direction of movement.

This unstable, cyclical behavior has greatly complicated thepooling of risk and pricing of insurance coverage.

While driven in part by the shifting mix of U.S. employment, thenationwide decline in accident frequency primarily reflectssignificant improvements in workplace safety through the combinedefforts of employers, insurers and the Occupational Safety andHealth Administration. Over the past six years, in fact, thenational accident rate for all private employers has declined byalmost 18 percent, in 2002 reaching a rate of just 2.8 injuries andillnesses per 100 full-time workers per year.

If workers' compensation insurance were priced and insured atthe national level, the continuity and stability of such long-termdeclines in accident frequency would exert a profound, stabilizingimpact on the workers' compensation insurance system. But, ofcourse, it is not.

Accident frequency rates in states such as Arizona and NewJersey have declined appreciably over the past six years, withArizona recording an impressive 27 percent decline in its frequencyrate for all private employers. At the same time, appreciableaccident frequency improvement has been elusive in states such asCalifornia and Vermont. Indeed,Vermont's accident rate actually increased some 13 percent, onbalance, between 1996 and 2002 (latest data available).

At the industry level, the trajectory of accident frequency canbe especially pronounced, even within those states recordingoverall improvement in frequency.

For example, New Jersey experienced considerable deteriorationin construction industry accident frequency over the 1996-2002period, even though the corresponding all-industries' accidentfrequency measure declined.

In Vermont, the increase in aggregate accident frequency overthe 1996-2002 period was 0.4 injuries per 100 full-time workers,but the surge in the accident frequency rate for manufacturingworkers was five times that, roughly 2.0 injuries per 100 full-timeworkers. Due to the backward-looking, aggregate process used inworkers' comp rate making, pronounced upswings in a majorindustry's accident frequency such as that recorded for Vermontmanufacturing can adversely affect underwriting experience for aset of industry-related class codes for a number of years beforethe higher accident frequency is fully manifest in the class-levelrate structure.

While many factors can influence the observed accident frequencyrate, one of the most important and most cyclically variable is thepace of business activity in a given industry. Construction andmanufacturing, due both to their impact on system-wide ratemakingand their volatility, are especially relevant examples.

The extent of state-to-state and period-to-period variability inconstruction activity is illustrated in Chart 3, using the growthof construction employment, which is a good proxy for the pace ofeconomic activity in the sector.

Over 1999-2000 (now the central part of the loss experience timespan for states' workers' comp ratemaking), California'sconstruction employment expanded at an exceptionally strong 19percent annual rate almost twice the national average growth. Whilethis growth subsequently slowed, California has maintained positivegrowth over the entire 2001-2004 period, while nationally,construction employment has been flat.

In contrast with California, construction employment growth inNew Jersey never reached such lofty heights in the late 90s.However, it has continued to grow through the recent recession andis forecast to expand at a 2 percent rate over the next threeyears. Vermont, on the other hand, has paralleled the nationalaverage by dropping to a negative growth status in 2001 and isprojected to remain soft through 2006.

In construction, as in other sectors, rapid production growth oftenexerts a detrimental influence on occupational accident frequency,and through it, on incurred workers' compensation loss experience.This negative impact flows through a number of employeecharacteristics and working condition factors.

Consider, for example, that to meet the increased demand inperiods of strong growth, employers will hire from a more shallowpool of workers who often lack experience, or rely more heavily onovertime both known contributors to workplace accidents.

In Vermont, for example, the fraction of construction employeeson the job for less than a year jumped markedly in 1997 from lessthan 10 percent to more than 25 percent as construction activitysurged, and it remained quite high through 2000. The constructionindustry in New Jersey, however, experienced a modest decline inthe fraction of new job entrants falling from around 19 percent toroughly 13 percent despite continued strong constructionactivity.

Periods of strong construction activity can also produce anincreased use of overtime hours. Vermont's construction industryrecorded much higher than normal overtime hours in the late 1990s,(more than 35 percent in 1996, compared to 20 percent in NewJersey) perhaps in part due to the state's more limited andgeographically isolated workforce. Since then, overtime utilizationin Vermont has moderated somewhat, but still remains well above thenational average for the construction industry.

Expanding construction activity also influences a state'sworkers' comp system through its tendency to spawn large numbers ofnew, smaller construction firms. These firms tend to employ lessexperienced workers and to have weaker accident preventionprograms, and as a result, to have higher than average accidentfrequency rates.

In Vermont, for example, smaller firms (non-employers and smallcommercial) accounted for almost 90 percent of the state's totalconstruction employment in 2002, while smaller firms made up onlyabout 65 percent of construction employment nationally. Withrespect to a carrier's underwriting results, increasing numbers ofthese smaller firms tend to intensify the impact of deterioratingaccident frequency, since they are too new and small to beexperience rated.

While in-depth analysis of employee characteristics and workingconditions might yield a more granular understanding of the factorsdriving deteriorating accident frequency, substantial insights andpractical value can be gained simply by monitoring industry growthrates.

Over the 2004-2006 period, construction industry employmentgrowth is forecast to be quite muted for the United States as awhole, averaging only one-half percentage point growth annually.Yet employment in New Jersey's construction industry is forecast toaverage about 1.7 percent annually over this period, placing it inthe ranks of the fastest growth states for construction exposureslikely with adverse development along with the likes of Nevada andArizona.

In most cases, it is not possible for carriers to eliminate thelagged adjustment process that characterizes current workers' compratemaking. However, the rise and fall of the business cyclescausing the problem are: (1) state-specific, (2) industry-specificand (3) readily forecast with substantial accuracy. As a result,carriers can use knowledge of anticipated business cycle phenomenonto reduce their exposure to workers' comp classes in specificstates that are having or are forecast to have adverse developmentin accident frequency.

Fritz Yohn is the developer of “MarketStance,” a market analysistool for U.S. commercial property-casualty insurers and is aregistered trademark of IntelliStance, LLC, in Middletown,Conn.

Reproduced from National Underwriter Edition, April 30, 2004.Copyright 2004 by The National Underwriter Company in the serialpublication. All rights reserved.Copyright in this article as anindependent work may be held by the author.