A

Cmcsaw©

K

1

iltwfS2ti

a

atc2aiont

0d

Accident Analysis and Prevention 39 (2007) 731–742

Financial performance, ISO 9000 standard and safe driving practiceseffects on accident rate in the U.S. motor carrier industry

Eitan Naveh a,∗, Alfred Marcus b

a Technion, Israel Institute of Technology, Faculty of IE&M, Haifa 32000, Israelb University of Minnesota, Carlson School of Management, Department of Strategic Management and Organization, Minneapolis, MN 55455, USA

Received 22 May 2006; received in revised form 26 October 2006; accepted 6 November 2006

bstract

The question this research address is if voluntary certification with ISO 9000 standards can reduce the number of accidents large trucks have.ertification might have a positive effect depending upon the actions a company takes as a result of being certified, for instance, a companyay change driver training and maintenance and how it carries out corrective action. We identified 40 ISO 9002: 1994 certified U.S. trucking

ompanies and compared their safety performance before and after ISO certification. We also compared them to a group of motor carriers that hadimilar characteristics but were not certified. The safety performance of the certified carriers was significantly better after certification than before,nd it also was significantly better than that of non-certified carriers, findings that we believe can be useful to carriers, shippers and regulatorsho wish to better understand the value of certification.

al resp

orsrU2era

2006 Elsevier Ltd. All rights reserved.

eywords: ISO 9000; Motor carrier industry; Traffic accidents; Corporate soci

. Introduction

Over 100 people die daily in traffic accidents in the U.S. whichs equivalent to the crash of a mid-sized commercial jetliner. Aarge number of these accidents involve motor vehicles of morehan 10,000 pounds (4.5 tonnes). Though just 7% of total high-ay travel in the U.S. in 1999, they were involved in 13% of the

atal crashes and 23% of the injuries (National Highway Trafficafety Administration, Fatality Analysis Reporting System,

000).1 Few would dispute that more should be done to improveheir safety. However, insufficient attention has been paid to thessue, and when attention has been paid the focus has been mostly

∗ Corresponding author.E-mail addresses: naveh@ie.technion.ac.il (E. Naveh),

marcus@csom.umn.edu (A. Marcus).1 Note that the classic mistake in safety analysis is the comparison of percent-ge of miles traveled by and percentage of crashes involving a specific vehicleype. The correct figure to be compared to percentage of miles traveled is per-entage of vehicles involved in accidents. Elementary calculations assuming5% of accidents involving one vehicle and 75% involving two will show thatvehicle class (like large trucks) having a 7% share of miles traveled would be

nvolved in 12% of crashes assuming that the class would have an average riskf accident involvement. Hence, large truck involvement in 13% of crashes iso sign of much higher risk, just a bit above average. However, it is clear thathe share of injuries is much higher than average.

spdca

ecrgpaiawt

001-4575/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.oi:10.1016/j.aap.2006.11.004

onsibility (CSR)

n the driver and not on the system that sends the driver on theoad and equips and trains the driver. Therefore, we analyzedafety performance at the system level and applied a number ofelevant theories about safety (Perrow, 1984; O’Connor, 1991;llmann, 1985; McGuire et al., 1988; Nutt, 1986; Klien et al.,001) that suggest that there are likely to be systematic differ-nces in the safety performance of motor carriers, that there iselationship between carrier financial performance and safety,nd that support that argument that certification with standardsuch as ISO 9000 may lead to safety improvements. Fig. 1resents our model which posits that financial performance has airect effect on safety and that it affects safety through ISO 9000ertification and the launching of various safe driving initiativesfter ISO 9000 is implemented.

The methodology of Hendricks and Singhal (2001) ismployed in our study. We matched ISO 9002: 1994 truckingompanies with a similar group of non-certified companies. Weelied on accident data as our main measure of safety. Using lon-itudinal data from U.S. trucking firms we compared the safetyerformance of certified and non-certified motor carriers beforend after certification. We found that the safety of the carriers

mproved after certification and that certified truckers had fewerccidents than control groups of uncertified truckers. We explainhy ISO 9000 may have this effect—it is through the practices

hat are introduced after certification including driver training,

732 E. Naveh, A. Marcus / Accident Analysis and Prevention 39 (2007) 731–742

searc

vd

2

ocUo

2

tutbrpt

Hp

o(pfiwaih

btdol

oh

epTwsoNsaew

Hc

2

tlsmrkto

Fig. 1. Re

ehicle maintenance, and overall safety management. We theniscuss possible alternative explanations for our findings.

. Theory and hypotheses development

Our hypotheses draw on the safety and reliability theoriesf Perrow (1984) and O’Connor (1991), the broader theories oforporate social responsibility found in such works as those ofllmann (1985), and the theories of administrative innovationf Nutt (1986) and Klien et al. (2001).

.1. Financial performance and safety results

With respect to industrial accidents, Perrow (1984) maintainshat minor incidents are nearly “normal,” that is they occur reg-larly and on a continuous basis.2 Following normal accidentheory, one would not expect systematic variation in the num-er and severity of traffic accidents in organizations. However,eliability theory (O’Connor, 1991) suggests that not all criticaloints of exposure and of vulnerability can be eliminated, andhat safety varies systematically by organization. We posit that:

ypothesis 1. There are systematic differences in the safetyerformance of the motor carriers.

Why is there systematic variance in the number of accidents inrganizations? According to the corporate social responsibilityCSR) literature, accidents may be caused by a firm’s financialerformance. Companies face a trade off between safety andnancial performance, since the costs of safety are high; in otherords, only well-to-do firms have the luxury to achieve above

verage safety (Ullmann, 1985). The contrasting point of views that safety costs are not that high and that firms benefit fromigher employee morale and greater productivity. They derive

2 Though fatalities are rare, Perrow (1984) considers them normal, that is, heelieves they are inevitable and that little can be done to prevent them. At theypical industrial facility, there is likely to be more than one such incident peray, but incidents that involve injuries are far less common, perhaps taking placenly once every 2–3 months, and very serious accidents involving major loss ofife are exceedingly rare (Perrow, 1984, p. 65).

psfizlei

Hps

h model.

ther advantages as well, for instance, a reputation for safetyelps in customer relations.

The argument of Marcus and Nichols (1996) is that safety lev-ls in organizations are not the highest technically and humanlyossible but depend on resources organizations have available.he risk organizations tolerate relies on a utilitarian calculusherein they view safety as desirable but costly and choose a

afety level by balancing the benefits of safety against the costsf safety improvement (Marcus and Nichols, 1996). Marcus andichols (1996) draw on prior corporate social responsibility

tudies such as that of McGuire et al. (1988) to suggest thatn organization’s financial performance is more likely to influ-nce its level of safety than the reverse. Based on this literature,e posit that:

ypothesis 2. Correlations exist between motor carrier finan-ial performance and its safety performance.

.2. Adoption of new management practices

Studies about the implementation of administrative innova-ions (Nutt, 1986; Klien et al., 2001) are consistent with the CSRiterature in emphasizing resources’ influence. These studieshow that the availability of resources positively affects imple-entation success. Financial performance as an indicator of

esource availability represents success, experience gained, andnow-how. Successful implementation of management prac-ices then leads to the accumulation and the development ofrganizational capabilities (Dierickx and Cool, 1989).

Thus, we hypothesize that implementation of managementractices partially mediates between financial performance andafety. By mediation (Baron and Kenny, 1986) we mean thatnancial performance impacts the implementation of the organi-ation’s management practices. The management practices thenead to safety improvement. Partial mediation means that a directffect also exists of financial performance on safety. The effect

s not only an indirect through the management practices.

ypothesis 3. The implementation of management practicesartially mediates the effects of financial performance on carrierafety performance.

lysis

2

arRii1tbtaaaIaAtooTttt(dm

Cthip9hwtw2st1so

4fmUeia

gaNat

ti

h(iwpiabocmb

p2ltttea

E. Naveh, A. Marcus / Accident Ana

.3. ISO 9000 and safe driving practices

Management standards, such as ISO 9000, are a subset ofbroader category of voluntary management practices, often

eferred to as administrative innovations (Brunsson et al., 2000).ecent studies suggest that there is likely to be a gain in operat-

ng performance when administrative innovations like ISO aremplemented (Anderson et al., 1999; Greve, 1999; Naveh et al.,999).3 ISO has a number of roles to play, which may bring abouthis improvement. The motivation for being registered com-ines the need to satisfy customer requirements with the needo make quality improvements (Anderson et al., 1999). To guar-ntee compliance, independent, third party auditors evaluate anpplicant’s procedures and carry out site visits. Twice a year theuditors return to the facility to re-verify that it is in compliance.n the U.S., the American National Standards Institute’s (ANSI)nd American Society of Quality (ASQ) administer ISO 9000.NSI with ASQ have a separate affiliate, the Registrar Accredi-

ation Board (RAB), which accredits third party auditors.4 Manyrganizations report tangible and intangible improvements inrganizational performance after ISO 9000 implementation.5

he standard is a tool for organizational improvement underhe following conditions: (1) the existence of internal motiva-ion for implementation; (2) customization of the standard to

he specific context, needs and conditions of the organization;3) the standard being routinely used; (4) the certified companyeveloping effective learning processes based on the standard’sandates (such as learning from mistakes and internal auditing

3 New models for quality have continually evolved and been refined over timeole (1999). ISO 9000 is a quality standard that requires that firms have set rou-

ines and procedures and that they demonstrate adherence to these routines. Byaving these routines and demonstrating that they adhere to them, their qualitys expected to increase. The International Standards Organizations in Genevaublished ISO 9000 in 1987 but its origins are in earlier U.S. (1960s—MIL-Q-858A and MIL-I 45208A) and British standards (BS 5750). Over 100 countriesave adopted ISO 9000. As of 2000 more than 300,000 facilities were certifiedorldwide (IQNet, 2000). Until January 2000, organizations implement either

he ISO 9001 or the ISO 9002, depending on whether they were associated alsoith research and development. These standards were replaced by the ISO 9001:000. Certifications based on the older standards remained valid during a tran-ition period that lasted till end of 2003. Since then ISO 9000 can only refero the new standard. The 2000 version of ISO 9000 is quit different from the994. It is more abstract, more oriented to the business process, and the customeratisfaction, while the older version was clearly focused on the management ofperational processes.4 By July 2005, based on a search of the ABI Inform Index search, there were937 journal papers and articles about ISO 9000. Some of the factors neededor successful ISO 9000 certification include the role of the certification body,anagement attitudes and commitment, and the role of the external consultants.ndoubtedly these factors influence whether certification has an impact on safety

nhancement. It is probably the case that the effects on safety in the motor carrierndustry depend on the degree to which risks are better identified and prioritiesre established for dealing with these risks.5 ISO 9000 is based on the assumption that high organizational performance isained by quality processes carried out before-the-fact and preventive in naturend not those carried out after-the fact and corrected only because of inspection.aveh and Halevy (1999) have developed a mechanism to assess the cost that is

ssociated with poor quality. This cost is a motivation for companies to improveheir quality assurance systems.

a

pBmhaaop

He9

3

3

oOd

nsNmha

gh

and Prevention 39 (2007) 731–742 733

o prevent mistakes); (5) a policy of going beyond the rules beingmplemented.

Implementation of the ISO standard is a discrete, notomogenous phenomenon. Variations must be consideredWestphal et al., 1997; Naveh et al., 1999). Yet each ISO 9000mplementation does include practices safety experts maintainould reduce accidents. In the motor carrier industry, theseractices are related to driver hiring, licensing procedures, work-ng hours, training, safety programs, instructions about limitedccess roads (roads with fewer intersections) and monitoring ofoth drivers and their vehicles.6 In other words, implementationf ISO 9000 can affect carrier safety by upgrading driving pro-edures, improving vehicles and enhancing overall carrier safetyanagement (these three aspects are explained in Section 3

elow).While a gap might exist between the rhetoric of a quality

rogram like ISO 9000 and the reality (Zbaracki, 1998; Zohar,000), a semblance of conformance to ISO 9000 guidelines isikely to apply once an organization is certified. The reason thathere is this semblance is that certification requires bi-annualhird party audits to which we referred previously. For instance,hough there is great variety in how ISO 9000 is implemented,xternal audits will verify if and how a company has establishednd uses a system for identifying and learning from its mistakesnd other mishaps.

We therefore hypothesize that safe driving practices willartially mediate the effect of ISO 9000 on carrier safety results.y mediation (Baron and Kenny, 1986) we mean that adoptinganagement practices such as those that ISO 9000 mandates

as an impact on driver-related safety, vehicle-related safety,nd safety management; and these practices lead to fewerccidents. Partial mediation means that there is also direct effectf ISO 9000 on safety, not one that is just through these drivingractices.

ypothesis 4. Safe driving practices partially mediate theffects of implementation of management practices, like ISO000, on carrier safety performance.

. Methods

.1. Sample

We included all U.S. motor carriers with large trucks inur sample.7 Three data sets were used to do the analyses.ne is the American Trucking Association (ATA) foundationata set, which includes financial data on all U.S. motor

6 Because ISO 9000 certification is designed to improve both a carrier’s inter-al auditing and corrective action procedures, it can have a positive impact onafety. In research for the Nuclear Regulatory Commission (NRC) Marcus andichols (1996) found that organizations often drift to a safety border, but whatakes a difference in terms of determining whether they go over this border and

ave serious accidents is various organizational processes they have in place forlerting them to the danger and to take corrective action.7 Our definition of motor carriers is that their trucks are over 10,000 poundsross vehicle weight rating. Thus, we are restricting our study to carriers thatave large trucks.

7 lysis

cTpfi

aosAcaautt

3

amotatsvficihwSv

1

2

ia

oh

3

3

hpt2wpn

ep(p(dgw

maTye

34 E. Naveh, A. Marcus / Accident Ana

arriers. The second is the federal government’s Department ofransportation (DOT) SAFER data set, which includes safetyerformance data on all U.S. carriers. The third data set comesrom McGraw-Hill. It names all ISO 9000 registered companiesn the U.S. and gives their year of registration.

Comparing the ATA data with the McGraw Hill, we wereble to identify 40 ISO 9002: 1994 registered companies outf 1742 additional non-registered carriers (at the time of thetudy there were no ISO 9000: 2000 version certified carriers).s indicated, being registered means that an external auditor

hecks for compliance with the ISO standard twice a year. Onlysmall percentage of carriers can afford to be registered. They

re found mainly in the automotive industry, where major man-factures require that suppliers be certified and among carriershat transport hazardous chemicals, where great care has to beaken in the transport of goods.

.2. Measures

The main financial measure we used as an independent vari-ble was return on assets (net income/total assets).8 The safetyeasures that we used as dependent variables were the number

f injury, fatal and tow accidents the carrier had as recorded inhe SAFER database. The tow accidents class is used in the USAnd it is not a standard class used throughout the world. It referso trucks towed due to damage from the crash scene, and thus ituggests information on the damage included. We included thisariable in our analyses for the completion of the analyses andt with the USA data collection method. Mediating variablesame from SafeStat, which calculates a score to rank carriersn descending order by their safety status. A carrier with theighest SafeStat score is the worst in terms of safety. A scoreith a value 75–100 is considered deficient.9 We used three ofafeStat Safety Evaluation Areas (SEAs) as indicators: driver,ehicle and management.

. Within the Driver SEA, SafeStat evaluates a carrier’s driver-related safety performance and compliance. The Driver SEAValue reflects a carrier’s driver safety posture relative to itspeers. SafeStat calculates the Driver SEA Value based onthe Driver Inspections Indicator (DII), the Driver ReviewIndicator (DRI) and the Moving Violations Indicator (MVI).The DII is based on driver roadside out-of-service (OOS)inspection violations. The DRI is based on the violationsof driver-related acute and critical regulations discoveredduring a compliance review. The MVI is based on seriousmoving violations recorded in conjunction with roadside

inspections.

. Within the Vehicle SEA, SafeStat evaluates a carrier’svehicle-related safety performance and compliance. The

8 We used other financial measures (total assets, total operating revenue, totalnsurance expenses and net operating income) in alternative runs of our analysesnd had similar results.9 Detailed explanation of the measures is available in the U.S. Departmentf Transportation, Federal Motor Carrier Safety Administration’s Internet sitettp://ai.volpe.dot.gov, where the SafeStat data can be found.

ns

Grit

0

and Prevention 39 (2007) 731–742

Vehicle SEA Value reflects a carrier’s vehicle safety pos-ture relative to its peers. SafeStat calculates the VehicleSEA Value based on the Vehicle Inspections Indicator(VII) and the Vehicle Review Indicator (VRI). The VIIis based on vehicle roadside OOS inspection violations.The VRI is based on the vehicle-related violations ofacute and critical regulations discovered during compliancereviews.

. The Safety Management SEA Value reflects the carrier’ssafety management posture relative to its peers. The SafetyManagement SEA Value is based on the EnforcementHistory Indicator (EHI), the Hazardous Material ReviewIndicator (HMRI) and the Safety Management ReviewIndicator (SMRI). The EHI uses the Enforcement SeverityMeasure (ESM) based on the results of violations cited inclosed enforcement cases. The HMRI and the SMRI useviolations of hazardous material-related acute and criticalregulations and violations of safety management-relatedacute and critical regulations, respectively, that werediscovered during a compliance review.

.3. Creating the matched samples

There is no single, agreed upon approach in the literature forow to create matched samples and then test for the long-termerformance effects on companies resulting from their adop-ion of different management practices (Hendricks and Singhal,001). Following the approach of Hendricks and Singhal (2001),e developed a series of matched samples with which to com-are the 40 ISO 9002: 1994 certified carriers with the1742on-certified carriers.

How did we create these matched samples? We started withach ISO certified company and gave each non-certified com-any an index value based on the similarity of the goods shippedGoods Index). We compared certified and non-certified com-anies with respect to each of the 29 categories of goods hauledgeneral freight, household goods, metals, etc.). When they wereifferent (the certified company hauled a particular category ofoods and the non-registered company did not—or vice versa),e added a one to the index.10

We also did a calculation of the difference in highwayiles traveled between certified and non-certified companies,process that we repeated for each non-certified company.

he measure of size was highway miles traveled in a particularear. We considered it the best indicator of size. The Size Indexxpressed a relative size comparison between certified andon-certified carriers. Simply put, it was the percentage of themaller company’s size in proportion to the larger company.

The non-certified companies then were sorted using bothoods Index and the Size Index. The Goodness of Fit Index

anged from 0 to 100. It was the weighted average of the twondexes, first the Goods Index, scaled to be a 0–100, and secondhe Size Index, also scaled in this fashion. The main Goodness

10 Thus, the index’s range, as first conceived, was 0–29. A perfect match waswhile complete difference between carriers was 29.

E.N

aveh,A.M

arcus/A

ccidentAnalysis

andP

revention39

(2007)731–742

735

Table 1Financial and safety performance of five best matched samples prior to and after ISO 9002: 1994 implementation

ISO 9002: 1994 certified Non-ISO 9002: 1994 certified matched sample

1 2 3 4 5

Sample size 40 40 40 40 40 40

Mean Goodness of Fit Indexes (S.D.)Main index 93.87 (6.38) 92.62 (6.93) 91.79 (7.44) 90.59 (7.62) 89.81 (7.72)Goods 94.69 (5.18) 92.57 (6.94) 92.96 (5.8) 91.37 (8.17) 90.71 (6.87)Size percentage 92.23 (15.74) 92.74 (17.44) 89.43 (16.18) 89.02 (19.04) 88 (17.34)

Financial resultsPre-implementation period (event month): −24 to 0

Mean financial measures (S.D.)Total miles on highway 35,728,943 (12,641,955) 37,043,347 (11,655,983) 42,840,340 (14,626,167) 32,476,444 (15,086,467) 30,408,421 (8,717,405) 33,078,251 (11,092,638)Total assets 30,463,262 (11,639,518) 47,192,961 (21,162,419) 43,144,428 (16,012,863) 31,952,010 (10,620,695) 36,025,732 (11,332,018) 24,592,349 (7,603,969)Total operating revenue 56,902,724 (23,384,875) 43,084,074 (44,915,390) 51,905,662 (27,292,017) 49,248,047 (16,727,104) 55,747,302 (15,119,451) 56,079,629 (19,257,635)Total insurance expenses 1,900,703 (699,697) 1,003,649 (1,071,865) 2,286,337 (767,802) 1,100,668 (281,944) 1,737,900 (465,447) 2,710,541 (928,085)Net operating income 2,579,035 (1,109,842) 1,853,628 (4,350,536) 2,115,496 (2,709,200) 3,910,769 (2,050,691) 3,834,515 (1,325,029) 3,516,384 (1,357,180)Net income 3,724,732 (806,658) 3,179,080 (2,668,894) 4,113,244 (1,777,926) 3,519,563 (2,014,210) 1,230,884 (843,859) 1,556,496 (1,078,504)

Mean of the paired difference (S.D.)Total miles on highway 5,067,230 (3,543,454) −7,111,397 (7,058,228) 3,252,498 (6,076,606) 5,320,522 (5,001,881) 2,650,692 (1,004,124)Total assets −3,548,906 (9,982,928) −12,681,166 (1,319,412) −1,488,748 (2,425,185) −556,270 (7,914,838) 5,870,867 (5,920,235)Total operating revenue −3,461,721 (5,126,425) −5,002,937 (7,556,223) 7,654,677 (11,946,882) 7,155,422 (3,775,495) 823,095 (2,526,473)Total insurance expenses −874,985 (408,423) −385,633 (806,542) 800,035 (540,432) 162,803 (420,388) −809,837 (313,635)Net operating income 1,438,154 (3,765,403) 1,536,460 (2,081,450) −1,331,734 (1,945,365) −1,255,479 (1,090,204) −937,348 (950,581)Net income 1,398,877 (2,403,005) −2,705,640 (1,485,588) −2,111,958 (1,894,933) 823,280 (791,332) 1,148,891 (935,346)

Paired differences that are positive (%)Total miles on highway 47.5 40 40 50 57.5Total assets 57.5 40 52.2 50 47.5Total operating revenue 50 37.5 32.5 40 40Total insurance expenses 60 47.5 32.5 37.5 57.5Net operating income 57.5 32.5 40 65 57.5Net income 60 47.5 47.5 57.5 47.5

T-statistic for the paired differenceTotal miles on highway 1.11 −1 .53 1.06 .26Total assets −1.35 −1.12 −.11 −.7 .99Total operating revenue −1.21 −.85 1.47 .8 .08Total insurance expenses −1.14 −.47 1.48 .38 −1.28Net operating income 1.24 1.49 −.68 −1.15 −.98Net income 1.24 −1.42 −1.11 1.04 1.22

Financial resultsPost-implementation period (event month): 1–24

Mean financial measures (S.D.)Total miles on highway 42,519,155 (16,014,200) 43,239,012 (21,204,275) 46,918,357 (16,238,027) 39,202,413 (13,359,479) 36,875,816 (11,952,472) 41,840,262 (15,160,286)Total assets 40,109,933 (15,882,261) 50,234,611 (8,588,970) 53,808,507 (20,953,295) 41,212,382 (14,071,849) 43,891,172 (11,384,003) 34,525,192 (13,919,583)

736E

.Naveh,A

.Marcus

/AccidentA

nalysisand

Prevention

39(2007)

731–742

Table 1 (Continued )

ISO 9002: 1994 certified Non-ISO 9002: 1994 certified matched sample

1 2 3 4 5

Total operating revenue 61,056,020 (26,060,946) 44,727,965 (55,461,692) 59,535,259 (21,785,862) 51,613,089 (22,096,792) 62,410,240 (9,526,928) 52,535,189 (11,661,095)Total insurance expenses 2,037,007 (721,540) 1,170,794 (1,583,002) 2,198,465 (741,324) 1,579,631 (558,727) 1,917,004 (598,122) 3,077,005 (2,119,802)Net operating income 3,651,034 (1,230,195) 2,120,954 (5,571,084) 3,742,408 (3,743,997) 3,993,570 (2,198,494) 3,170,283 (2,226,211) 2,270,846 (1,507,438)Net income 4,165,508 (518,577) 3,502,067 (3,445,518) 4,300,891 (2,547,624) 3,733,140 (2,241,822) 1,995,236 (1,132,696) 1,801,192 (1,741,309)

Mean of the paired difference (S.D.)Total miles on highway 5,542,592 (1,212,740) −4,399,202 (3,543,202) 3,316,742 (3,510,459) 5,643,339 (6,321,801) 678,893 (4,136,514)Total assets 5,033,913 (2,829,007) −3,698,573 (2,525,609) −1,102,448 (7,751,658) −3,781,238 (5,731,910) 3,584,741 (7,020,546)Total operating revenue 5,082,171 (6,159,716) −3,479,238 (4,865,136) 4,442,931 (9,368,759) −6,354,219 (3,751,459) −6,479,168 (6,090,145)Total insurance expenses 1,833,342 (845,552) −161,457 (722,196) 457,376 (385,151) 120,003 (321,981) 2,039,997 (1,477,297)Net operating income 2,353,564 (4,775,401) 1,091,373 (2,869,723) −1,342,535 (2,063,965) 2,519,248 (1,463,713) 980,188 (1,958,261)Net income 1,818,512 (3,065,727) 1,135,382 (2,161,042) −2,567,631 (207,6618) −829,728 (1,042,356) 1,266,700 (2,055,255)

Paired differences that are positive (%)Total miles on highway 65 46 50 46 65Total assets 65 42 50 42 50Total operating revenue 50 34 34 42 38Total insurance expenses 50 34 34 46 53Net operating income 57 50 46 50 53Net income 57 50 53 50 38

T-statistic for the paired differenceTotal miles on highway 1.72† −1.24 .94 .89 .16Total assets 2.08* −1.09 −.06 −.92 .42Total operating revenue 1.95† −.71 .74 −.46 −.78Total insurance expenses 2.16* −.22 1.86† .37 1.68†Net operating income 2.16* 1.77† −.65 1.72† .19Net income 2.22* 1.91† −1.2 −.79 .61

Safety resultsPre-implementation period (event month): −24 to 0

Mean safety measures (S.D.)Fatal .72 (.27) .92 (.28) .75 (.24) .52 (.15) .55 (.2) .66 (.26)Injury 8.2 (2.53) 11.13 (3.3) 11.52 (3.44) 7.44 (2.92) 7.9 (2.14) 11.27 (3.27)Tow 6.7 (4.44) 6.4 (3.45) 3.97 (3.07) 6 (2.76) 5.91 (1.38) 4.16 (2.65)

Mean of the paired difference (S.D.)Fatal .13 (.24) −.02 (.16) .19 (.24) −.16 (.21) .05 (.4)Injury −1.21 (1.87) −3.3 (2.04) .77 (1.6) .3 (1.3) −3.05 (2.28)Tow −2.47 (1.9) −3.8 (1.76) .83 (1.48) .86 (1.21) 2.33 (2.2)

Paired differences that are negative (%)Fatal 25 27.5 37.5 30 27.5Injury 30 40 32.5 25 30Tow 37.5 35 30 40 40

T-statistic for the paired differenceFatal .46 .6 .53 .53 .4Injury .53 .53 .4 .4 .53Tow .4 .4 .4 .4 .4

E. Naveh, A. Marcus / Accident Analysis

Safe

tyre

sults

Post

-im

plem

enta

tion

peri

od(e

vent

mon

th):

1–24

Mea

nsa

fety

mea

sure

s(S

.D.)

Fata

l.6

9(.

24)

.84

(.36

).7

7(.

23)

.36

(.19

).6

3(.

24)

.66

(.53

)In

jury

7.27

(2.2

5)9.

86(3

.27)

8.69

(2.8

2)7.

69(3

.53)

6.63

(1.8

)7.

97(2

.96)

Tow

6.05

(2.1

6)7.

94(2

.49)

7.11

(1.9

9)5.

69(2

.54)

5.77

(1.5

4)4.

3(2

.86)

Mea

nof

the

pair

eddi

ffer

ence

(S.D

.)Fa

tal

−.1

(.34

)−.

08(.

2)−.

33(.

24)

−.05

(.31

).0

2(.

41)

Inju

ry−1

.26

(1.2

6)−1

.4(1

.32)

−.41

(2.3

)−.

63(1

.31)

−.69

(1.5

2)To

w−.

57(.

9)−.

88(.

9)−.

52(1

.74)

−.44

(1.0

4)2.

08(1

.4)

Pair

eddi

ffer

ence

sth

atar

ene

gativ

e(%

)Fa

tal

6562

.555

57.5

57.5

Inju

ry65

62.5

8050

80To

w57

.567

.557

.557

.557

.5

T-s

tatis

ticfo

rth

epa

ired

diff

eren

ceFa

tal

.11

−.43

−2.0

8†−.

11.

84†

Inju

ry.2

9−.

63−.

9−1

.19

−.36

Tow

−2.2

5*−2

.23*

−.9

−1.7

6†−.

46

*p

<.0

5;† p

<.1

.

oWhscoTmw

stfiFtIfiif

3

dcoecm

teHbttici

spsieIo

7Sb

sa

t

and Prevention 39 (2007) 731–742 737

f Fit Index was the “weighted” average of the two indices.e assumed that if carriers were similar in the types of goods

auled and the amount of miles traveled, then they would beimilar in other variables (driver experience, turnover rates,rash types, age of trucks, maintenance policies, technologiesf the trucks, time of driving—day, night or working 24 h, etc.).he top 20 non-registered companies that were the 20 bestatches for each certified company were used in the analyseshich were carried out.11

The ISO 9002: 1994 registered companies financial andafety annual measures were taken for 2 years before regis-ration, 1 year before registration, the year of registration, therst year after registration and the second year after registration.or each matched company, financial and safety measures were

aken for the same calendar years as for the company that wasSO 9002: 1994 certified (i.e. if 1998 was the certification year,nancial and safety measures both for the ISO 9002: 1994 reg-

stered company and for its 20 matched companies were takenrom 1996 and up to 2000).

.4. Analyses

We first analyzed our data following the statistical proce-ure used by Hendricks and Singhal (2001). Differences wereompared for each of the 20 matched samples paired with eachf the 40 ISO 9002: 1994 registered companies. These differ-nces were calculated as the differences between the certifiedompany’s measure and the non-certified, but paired, company’seasure.12

Going beyond Hendricks and Singhal (2001) methods, wehen relied on a multi-level approach and used Hierarchical Lin-ar Models (Bryk and Raudenbush, 1992) to test our hypotheses.ierarchical Linear Models (HLM) for analyzing the data wereuilt (Bryk and Raudenbush, 1992) taking into to considera-ion: (1) the 40 ISO 9002: 1994 registered companies as regardo their financial and safety performance before and after reg-stration and (2) each ISO 9002: 1994 registered 20 matchedompanies as regards to their financial and safety performancen similar years.13

According to the mediation model called for by our hypothe-es (Fig. 1), financial performance (ROA) and new managementractices (ISO 9002: 1994) should have a significant effect onafe driving practices and result in fewer accidents. Safe driv-ng practices also should yield fewer accidents, but to a lesser

xtent than financial performance and the management practice.f mediation holds, then ISO 9002: 1994 should lessen the effectf the financial variable on safety.

11 The mean main index (Goodness of Fit) for the 20th matched sample was7 with a standard deviation (S.D.) of 12, the mean Goods Index was 91 with a.D. of 6 and the mean Size Index was 50 with a S.D. of 38. This mean droppedelow 50 for the 21st best sample.

12 Following Hendricks and Singhal (2001) this analysis was done using a T-tatistic to determine if these differences were significant. The t-test was runfter a test was carried out to see if the measures were normally distributed.13 Twenty times 40 yielded 800 non-certified companies, with the addition ofhe 40 certified carriers the N was 840.

738 E. Naveh, A. Marcus / Accident Analysis and Prevention 39 (2007) 731–742

s—pr

4

9paTc9

diop

Fig. 2. Safety results of ISO 9002: 1994 carrier

. Results

Table 1 presents descriptive statistics, which compare the ISO002: 1994 registered companies and the 5 best matched sam-les. The first part of the table presents the pre-implementation

nd the post-implementation period for financial performance.he table shows how similar the ISO certified and the non-ISOertified carriers are. The Goodness of Fit Indices are close to0% or more.

ps

a

Fig. 3. Safety results of ISO 9002: 1994 carriers and of the contro

e and post implementation (results of Table 1).

In the pre-implementation period, there are no significantifferent means (no significant t tests among the paired compar-sons). In the post-implementation period, there are a numberf significantly different means, which show improved financialerformance after the carriers become ISO certified. For exam-

le, in comparison to sample 1, the ISO certified carriers haveignificantly more income.

The second part of the table presents the pre-implementationnd the post-implementation period for safety performance. To

l groups—pre and post implementation (results of Table 1).

E. Naveh, A. Marcus / Accident Analysis and Prevention 39 (2007) 731–742 739

Table 2Pre-implementation and post-implementation periods mean, standard deviation and correlationsA,B

Pre-im01; †

bpiig

tIinsIrienno

ociicssarp

sati

ict

e

d1iBnrccpa

fmBtitmar

bctct

9om(tpartial mediation (see Fig. 1) is that both ROA and ISO 9002:1994, as predicted, significantly impact safe driving practices(models 1, 2 and 3).

AISO 9002: 1994 registered companies and the best twenty matched samples; B

correlations are above the diagonal; CNet income/total assets; *p < .05; ***p < .0

etter illustrate the data it is also presented in two figures. Fig. 2resents the trend among ISO 9002: 1994 carriers pre and postmplementation. Fig. 3 presents the data of before and aftermplementation in comparing to the control non-ISO certifiedroups.

Using a paired t-test for the performance of the carriers inhe 2 years before and after we found: Thirty-six out of the 40SO 9002: 1994 certified companies statistically significantlymproved their financial performance (the rest are with no sig-ificant change), and 28 statistically significantly improved theirafety performance (the rest are with no significant change).n the control groups, for financial improvement the numberanged between 5 and 11 carriers that statistically significantlymproved their performance (the rest did not improved or wors-ned their financial performance). For safety improvement theumber ranged between 10 and 18 carriers that statistically sig-ificantly improved their performance (the rest did not improvedr worsened their safety performance).

As can be seen in Table 1, in the first period, there are no casesf significant differences between the ISO certified and non-ISOertified carriers with regard to their safety performance, but then the second period the ISO certified carriers begin to shownstances of significantly better safety records. For example, inomparison to samples 1 and 2, the ISO certified carriers haveignificantly fewer tow accidents. In particular, with regard toimple accidents involving towing, as opposed to more severeccidents involving loss of life or injuries, the ISO certified car-iers do significantly better in the second, post-implementationeriod.

These results provide support for the first hypothesis that theafety performance of the carriers is likely to vary. Carriers thatre similar in most other ways seem to diverge when it comeso their safety performance depending on whether they haventroduced ISO 9002: 1994 or not.

Table 2 presents pre-implementation and post-mplementation periods means, standard deviation, and

orrelations of the ISO 9002: 1994 registered companies andhe best twenty matched samples.

This table provides additional evidence for the first hypoth-sis. After implementation (see the correlations above the

mt

plementation correlations are below the diagonal, while post-implementationp < .1.

iagonal), there are significant relationships between ISO 9002:994 and fatal, injury and tow accidents. Carriers that havemplemented the standard have fewer of these kinds of crashes.efore implementation (see the correlations below the diago-al), these relationships do not hold. There are no significantelationships between the carriers that become ISO 9002: 1994ertified and the safety indicators. In contrast, ROA is signifi-antly related to safety in both the pre and post implementationeriods. The higher the ROA of a carrier, the fewer fatal, injurynd tow crashes it has.

Table 3 presents the results of hierarchical models, which testor the effects of financial performance and the adoption of theanagement practice, ISO 9002: 1994, on safety performance.14

ecause that in these analyses the variables used as explana-ory variables (independent variables), are not independent wencluded the interaction terms between these variables as addi-ional explanatory terms. These interactions had a near-zero

agnitude, insignificant effect. Also, since theoretically an inter-ction relationship were the less parsimonious explanation, weerun the models without these interactions (Cohen, 1988).

Hypothesis 2 is strongly supported in that the carriers withetter financial performance, that is higher ROA, have signifi-antly better safety records. This result is uniform across all ofhe safety indicators (models 1–12). The better a carrier’s finan-ial performance, the better the driving practices and the fewerhe accidents.

Hypothesis 3 receives some support in the analyses. ISO002: 1994s presence in the model reduces the effect on ROAn safety, but just slightly (models 5, 6; 8, 9 and 11, 12). ROA isore predictive of fatal and injury accidents than ISO 9002: 1994

models 5, 6 and 8, 9). However, ISO 9002: 1994 is more predic-ive of tow accidents (models 11 and 12). What adds weight to the

14 As expected with regard to the control variable, the more total highwayiles, the worse a carrier’s safety (models 1–12). This result is uniform across

he safety indicators.

740E

.Naveh,A

.Marcus

/AccidentA

nalysisand

Prevention

39(2007)

731–742

Table 3Result of hierarchical models testing the effect of financial performance and standardization on safety performance

Vehicle-related safety Safety management Driver-related safety Fatal Fatal Fatal

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

Intercept 5.2 (1.7)*** 5.2 (2.7)*** 5.3 (.6)*** .49 (.03)*** .49 (.03)*** .49 (.03)***

Total miles on highway .0000007 (.00000001)*** .0000007 (.00000001)*** .0000007 (.00000001)*** .00000001 (.000000001)*** .00000001 (.000000001)*** .00000001 (.000000001)***

ROAA −16.6 (5.2)*** −21.2 (6.1)*** −14.2 (6.9)*** −.32 (.18)*** −.3 (.15)*** −.27 (.15)***

ISO 9002: 1994 −3.2 (1.3)*** −5.3 (2.1)*** −3.1 (2.3)*** −.17 (.07)* −.12 (.05)†Vehicle-related safety .59 (.2)** .13 (2.1)** .05 (.05)Safety management .1 (.05)* .04 (.02)* .05 (.03)Driver-related safety .02 (.01) .3 (.01)* .06 (.004)N 840 840 840 840 840 840−2 res log likelihood 3252 3428 3621 3719 3714 3708

Injury Injury Injury Tow Tow Tow

Model 7 Model 8 Model 9 Model 10 Model 11 Model 12

Intercept 7.7 (1.2)*** 9.8 (1.2)*** 4.2 (1.2)*** 6.2 (1.3)*** 11.2 (2.5)*** 1.6 (1.8)***

Total miles on highway .00000008 (.00000001)*** .00000009 (.00000001)*** .00000009 (.00000001)*** .000000011 (.00000001)*** .000000011 (.00000001)*** .000000011 (.00000001)***

ROAA −3.6 (1.8)*** −3.5 (1.8)*** −2.4 (1.8)** −4.2 (1.3)*** −3.9 (1.2)*** −2.9 (1.2)*

ISO 9002: 1994 −2.1 (.9)** −1.4 (.5)** −4.8 (2.2)*** −3.2 (1.3)**

Vehicle-related safety .01 (.1) .1 (.06)**

Safety management .01 (.02)* .05 (.02)**

Driver-related safety .008 (.003)† .007 (.05)N 840 840 840 840 840 840−2 res log likelihood 9672 9658 9650 9368 9358 9351

Standard errors in parentheses.A Net income/total assets.* p < .05.

** p < .01.*** p < .001.† p < .1.

lysis

9aeb9attm

5

tdctptvashlttabbb

itaIdtbwmbtfw

5

tnfr9

va

ipmpcafactpifipIacvmf

5

talsfimeltsrqfictsbtib

R

A

B

E. Naveh, A. Marcus / Accident Ana

Hypothesis 4 is fully supported in that in each model, ISO002: 1994 is significantly related to safer driving (models 1–3)nd fewer accidents (models 5, 6; 8, 9 and 11, 12), and this holdsven in models where ROA and the safe driving practices areoth present (models 6, 9 and 12). In all of the regressions, ISO002: 1994 continues to have a significant effect in reducingccidents, whether they are fatal, injury or tow accidents. Con-rolling for the other variables in our models, the adoption ofhis management standard has a positive effect on the safety of

otor carriers.

. Conclusion

While many studies deal with traffic safety performance athe national, state and individual driver levels, only few studieseal with traffic safety performance of organizations. The mainontribution of this study is by showing that a management prac-ice, a quality standard, ISO 9000, can lead to better traffic safetyerformance. Such an argument has not before been supported inhe organizational literature and it brings a new and fresh point ofiew to the study of traffic safety. Our study suggests that man-gement practices mediate between financial performance andocial performance. Organizations with better financial resultsave a higher tendency to implement management practices thatead to better social performance. The same management prac-ices have an effect on financial performance, which suggestshat organizations can doubly benefit from implementing man-gement practices like ISO 9000 (specifically, this conclusion isased on the analyses showed in Table 1. This conclusion cannote supported by the analyses showed in Tables 2 and 3 whichased on correlations which do not prove a causal effect).

Alternative explanations for the findings are possible. First,t might be that carriers that implement ISO 9000 do so becauseheir customers require them. This may make these companiesware to the importance of customers and other stakeholders.mprovement in performance can be gained as a result of theevelopment of organizational culture (Naveh and Erez, 2004)hat is more open to side effects of the business. Second, it mighte that concert safety requirements of customers that do notant to be associated with accident made the change. Third, itight be that companies that implement ISO 9000 are likely to

e more innovative and have better management practices thanhe average of their competitors. Since safety management is aunction of the general management function, this may explainhy these companies are able to manage safety more effectively.

.1. Limitations and future research

The 1994 version of the ISO 9000 standard updated in 2000o ISO 9001: 2000. The new version is more oriented to busi-ess processes, and customer satisfaction while the 1994 versionocused on the management of operational processes. Futureesearch should check the new version of the ISO 9000–ISO

001: 2000.

While this study set the theoretical argumentation and pro-ided evidence at the level of the commonly use three types ofccidents future studies can get to the causes of type of accidents

B

B

and Prevention 39 (2007) 731–742 741

n more detail. Voluntary ISO 9000 certification does have theotential to alleviate the regulatory burden and improve overallotor carrier safety. However, certification is a relatively new

henomenon in this industry, and companies that have beenertified may be unique. Whether ISO 9000 is likely to havepositive effect on safety is likely to depend on a variety of

actors—the reason a company becomes certified, the auditorsnd consultants it engages, the process and the content of itsertification. Size and visibility of the company also are likelyo be important. Our analysis suggests certification can have aositive effect on safety but the conditions under which this pos-tive impact exists require further investigation. To what extent,or instance, does certification lead to the introduction of newnternal auditing procedures, and what kind of learning takeslace? Further investigation of the circumstances under whichSO 9000 leads to improvements in safety could lead to the cre-tion of a quality assurance standard for motor carriers. It alsoould lead to changes in regulation, wherein regulators relied onoluntary ISO 9000 implementation in place of, or as a comple-ent to, actions they currently take. All these aspects should be

arther studied.

.2. Implications

The overriding implication to be derived from this study iso improve motor carrier safety and reliability and avoid trafficccidents that can cause deaths and injuries, quality standardsike ISO 9000 can play a useful purpose. This study is the firstolidly based research exploring and comparing the safety andnancial performance of ISO 9000 certified and non-certifiedotor carriers companies before and after certification. In it, we

xamined and identified the conditions under which ISO 9000 isikely to have a positive effect on carrier safety. Understandinghe connection between a better quality assurance system andafety performance in this industry should provide guidance toegulators, shippers and firms in the industry that the ISO 9000uality assurance standard can have a positive effect on bothnancial and safety performance. With ISO 9000 in place, motorarrier regulators can feel more confident that a carrier is likelyo exhibit good driving practices and have fewer accidents. Thistudy also can serve to develop shipper awareness of the linketween certification and safety. Inasmuch as this study helpso define a modern quality assurance system for the truckingndustry as a tool for increasing safety its implications are veryroad.

eferences

nderson, S.W., Daly, J.D., Johnson, M.F., 1999. Why firms seek ISO 9000 certi-fication: regulatory compliance competitive advantage? Prod. Oper. Manage.8, 28–43.

aron, R.M., Kenny, D.A., 1986. The moderator-mediator variable distinc-tion in social psychological research: conceptual, strategic, and statistical

considerations. J. Personality Social Psych. 51 (December (6)), 1173–1182.

runsson, N., Jacobsson, B., et al., 2000. A World of Standards. Oxford Uni-versity Press, New York.

ryk, A.S., Raudenbush, S.W., 1992. Hierarchical linear models: applicationsand data analysis methods. Sage Publications, Newbury Park, California.

7 lysis

C

CD

G

H

I

K

M

M

N

N

N

N

N

O

P

U

W

42 E. Naveh, A. Marcus / Accident Ana

ohen, J., 1988. Statistical Power Analysis for The Behavioral Sciences, seconded. Erlbaum, Hillsdale, NJ.

ole, E.R., 1999. Managing Quality Fads. Oxford University Press, New York.ierickx, I., Cool, K., 1989. Asset stock accumulation and sustainability of

competitive advantage. Manage. Sci. 35, 1504–1511.reve, H.R., 1999. The effect of core change on performance: inertia and regres-

sion toward the mean. Admin. Sci. Q. 44, 590–614.endricks, K.B., Singhal, V.R., 2001. The long-run stock price performance of

firms with effective. TQM programs. Manage. Sci. 47, 359–368.QNet Association, 2000. Bern, Switzerland, http://www.iqnet-

certification.com.lien, J.K., Conn, A.B., Sorra, J.S., 2001. Implementing computerized technol-

ogy: an organizational analysis. J. Appl. Psychol. 86, 811–824.arcus, A., Nichols, M., August 1996. Acquiring and using knowledge in

response to unusual events in a hazardous industry. In: Presented at theAnnual Meeting of the Academy of Management, Cincinnati, Ohio.

cGuire, J.B., Sundgren, A., Schneeweis, T., 1988. Corporate social responsi-

bility and firm financial performance. Acad. Manage. J. 31, 854–872.

ational Highway Traffic Safety Administration, 2000. Fatality AnalysisReporting System.

aveh, E., Halevy, A., 1999. How much do we lose on non-quality? Total QualityManage. J. 10, 1037–1045.

Z

Z

and Prevention 39 (2007) 731–742

aveh, E., Marcus, A., Moon, H.K., Allen, G., 1999. ISO 9000 Survey ’99:An Analytical Tool to Assess the Costs, Benefits and Savings of ISO 9000Registration. McGraw-Hill.

aveh, E., Erez, M., 2004. Innovation and attention-to-detail in the qualityimprovement paradigm management. Science 50, 1576–1586.

utt, P.C., 1986. Tactics of implementation. Acad. Manage. J. 29, 230–261.

’Connor, P., 1991. Practical Reliability Engineering, third ed. John Wiley, NewYork.

errow, C., 1984. Normal Accidents: Living with High-risk Technologies. BasicBooks, New York.

llmann, A.A., 1985. Data in search of a theory: a critical examination of therelationships among social performance, social disclosure, and economicperformance of U.S. firms. Acad. Manage. Rev. 10, 540–557.

estphal, J.D., Gulati, R., Shortell, S.M., 1997. Customization or conformity?An institutional and network perspective on the content and consequencesof TQM adoption. Admin. Sci. Q. 42, 366–394.

baracki, M.J., 1998. The rhetoric and reality of total quality management.Admin. Sci. Q. 43, 602–636.

ohar, D., 2000. A group-level model of safety climate: testing the effect ofgroup climate on microaccidents in manufacturing jobs. J. Appl. Psychol.85, 587–596.