A Multivariate Time‐Series Examination of Motor Carrier Safety Behaviors

• Author: Jason W. Miller
• Date: 01 December 2017
• DOI: http://doi.org/10.1111/jbl.12162
• Published date: 01 December 2017

A Multivariate Time-Series Examination of Motor Carrier Safety

Behaviors

Jason W. Miller

Michigan State University

Motor carriers’operational safety affects multiple stakeholders including truck drivers, motor carriers, insurance companies, shippers, and

the general public. In this article, I devise and test theory regarding motor carriers’longitudinal performance for three classes of safety

behaviors linked to carriers’accident rates—Unsafe Driving, Hours-of-Service Compliance, and Vehicle Maintenance—tracked by the Federal

Motor Carrier Safety Administration as part of the Compliance, Safety, and Accountability (CSA) program. Specifically, I draw on core con-

cepts from sociological agency theory and resource dependency theory to devise middle-range theory that generates never-before-tested

hypotheses regarding carriers’longitudinal safety performance for these classes of safety behaviors after the start of the CSA program. The

hypothesized predictions are tested by fitting a series of multivariate latent curve models to four years of panel data for a random sample of

484 large, for-hire motor carriers operating in the United States. The empirical findings corroborate the theoretical predictions and remain after

robustness testing. These findings have important implications for scholars, motor carrier managers, procurers of motor carrier transportation ser-

vices, and public policy makers.

Keywords: transportation; motor carrier; safety; agency theory; resource dependency theory; panel data; latent curve model

INTRODUCTION

It is difficult to understate how motor carrier safety affects truck

drivers, motor carriers, shippers, consignees, other supply chain

members, and the general public. Chen et al. (2015), citing

statistics from the Federal Motor Carrier Safety Administration

(FMCSA), note that in 2012 there were 3,464 fatal, 73,000

injury-causing, and 241,000 property-damaging crashes involving

commercial trucks that caused an estimated $99 billion in dam-

ages. Given firms across all tiers of the supply chain use motor

carrier transportation (Wilson 2015), motor carrier safety is an

important concern for all supply chain members, not just carriers

(Cantor 2008).

In an effort to improve carrier safety, in 2010 the FMCSA

implemented the Compliance, Safety, and Accountability (CSA)

program that publically discloses motor carriers’performance

across several safety dimensions (The Volpe Center 2013). This

aligns with the broader societal trend whereby social control

agents (e.g., regulatory agencies) increasingly seek to influence

firms’behaviors via publicizing information to (1) alter metered

firms’behaviors directly and (2) empower stakeholders to pres-

sure metered firms to improve or face consequences (Fung et al.

2007). Importantly for supply chain management (SCM) scholars

and managers, many information disclosure programs (IDPs)

make available information directly pertaining to firms’SCM

activities. Some examples include the Environmental Protection

Agency disclosing firms’toxic emissions through the Toxics

Release Inventory program (Hamilton 1995; Chatterji and Toffel

2010), Los Angeles requiring restaurants to display their hygiene

quality scores (Jin and Leslie 2003, 2009), and the Centers for

Medicare and Medicaid making available data on hospitals’qual-

ity of care and patient satisfaction (Williams et al. 2005; Jha

et al. 2008). Thus, examining carriers’longitudinal safety perfor-

mance since the start of the CSA program would not only enrich

our understanding of carrier safety, but would also provide more

general knowledge regarding how metered firms perform after

the start of IDPs (Chatterji and Toffel 2010).

This article contributes to the literature in several ways. First,

I use the multifaceted conceptualization of safety developed by

the FMCSA to devise and test theory that predicts certain pat-

terns of relations will arise because safety dimensions differ in

the degree they are under carriers’versus drivers’influence. The

underlying logic for these predictions is derived from sociologi-

cal agency theory (SAT) (Kiser 1994, 1999; Shapiro 2005). By

contextualizing more general theoretical principles to this SCM

setting, I devise “middle-range”theory (Merton 1968; Pawson

2000) that furthers our understanding of motor carrier safety and

aligns with recommendations for SCM scholars to devise theo-

ries of this type (Holmstr€

om et al. 2009; Craighead et al. 2016).

Apart from providing a parsimonious account for an array of

empirical findings, the middle-range theory I propose suggests

several possibilities for future inquiry, meeting the fecundity cri-

terion recommended to evaluate theoretical contributions in SCM

(Wacker 1998). This work also contributes to SAT by enhancing

the theory’s consilience (Thagard 1978) by illustrating that

SAT’s tenets can account for a new class of findings. This is

consistent with arguments from philosophy of science that one

way scholars contribute to general theories is by illustrating how

general theories provide unifying explanatory accounts of many

phenomena (Whitt 1992; M€

aki 2001; Lipton 2004).

A second way this article contributes to the literature is by

increasing our understanding regarding metered firms’perfor-

mance following the start of IDPs. Many IDPs disclose informa-

tion on only a single outcome such as restaurant hygiene scores

(Jin and Leslie 2003, 2009) or patient mortality (Hannan et al.

1994; Chassin 2002), making it impossible to test theory regard-

ing why metered firms may display different patterns of

Corresponding author:

Jason W. Miller, Department of Supply Chain Management, Eli

Broad College of Business, Michigan State University, 632 Bogue

Street, N 370, East Lansing, MI 48824, USA; E-mail: mill2831@

msu.edu

Journal of Business Logistics, 2017, 38(4): 266–289 doi: 10.1111/jbl.12162

© Council of Supply Chain Management Professionals

longitudinal performance across several measures. I leverage the

fact the CSA program tracks multiple safety measures and draw

upon resource dependency theory (RDT) (Pfeffer and Salancik

2003) to offer predictions regarding differences in (1) carriers’

rates of improvement across these safety classes and (2) the cor-

relations between carriers’initial performance and their subse-

quent rates of change. This extends prior research by suggesting

how characteristics of disclosed metrics affect metered firms’

rates of change following IDP rollout. This not only enhances

our understanding regarding IDPs, but also provides guidance for

policy makers designing these programs. It further responds to

Craighead et al.’s (2016) recommendation for SCM scholars to

apply Morgeson et al.’s (2015) event system theory framework

to devise middle-range theories in SCM.

To make these contributions, I assemble a repeated-measures

panel data set consisting eight semiannual measurement occa-

sions for three safety constructs—Unsafe Driving, Hours-of-Ser-

vice (HOS) Compliance, and Vehicle Maintenance—tracked by

the FMCSA as part of the CSA program for a random sample of

484 large, for-hire motor carriers. I then fit a series of multi-

variate latent curve models (MLCMs) (MacCallum et al. 1997;

Blozis 2004) to these data to test the hypothesized predictions.

In doing so, this article follows the tradition of several studies

(Dunn et al. 1994; Garver and Mentzer 1999; Garver et al. 2008;

Coltman et al. 2011; Miller et al. 2013) that have introduced

statistical techniques that enable SCM scholars to answer new

questions in other SCM domains.

The remainder of this article is structured as follows. The first

section contains the literature review. The theory and hypothesis

development follows. The third section summarizes sampling

design, describes the measures, details data preparation, and

introduces the statistical approach. The fourth section describes

the results from the statistical analysis and robustness tests. The

fifth section summarizes theoretical contributions, explains man-

agerial and public policy implications, highlights limitations, and

offers directions for future research.

BACKGROUND LITERATURE

Many studies have investigated motor carrier safety behaviors.

1

While a few studies have predominantly focused on identifying

the prevalence of unsafe behaviors (e.g., speeding, violating

HOS rules) (Hertz 1991; Chen et al. 2015), most studies have

sought to identify driver-, haul-, and carrier-level factors that pre-

dict such behaviors. At the driver level, factors found to affect

safety behaviors include owner-operator status (Mayhew and

Quinlan 2006; Williamson et al. 2009; Cantor et al. 2013), driver

age/experience (Braver et al. 1992; Sullman et al. 2002), lack of

sleep (McCartt et al. 2000), drivers’attitudes toward safety

(Swartz and Douglas 2009), and pay structure (Monaco and Wil-

liams 2000; Shaw et al. 2002). At the haul level, factors found

to affect safety behaviors include having unrealistic delivery

schedules (Beilock 1995, 2003; Crum and Morrow 2002), the

presence of late delivery penalties (Sabbagh-Ehrlich et al. 2005),

and long waiting times to load or unload (McCartt et al. 2008;

Williamson and Friswell 2013). At the carrier level, factors

include type of commodity hauled (Horrace and Keane 2004;

Tzamalouka et al. 2005), use of electronic monitoring (Cantor

et al. 2009), carriers’financial performance (Beard 1992; Corsi

2004; Naveh and Marcus 2007; Britto et al. 2010), driver turn-

over (Shaw et al. 2005), unionization (Corsi et al. 2012), and

firm size (Cantor et al. 2016). Other work has examined whether

drivers and managers have similar perceptions of safety problem

prevalence and its causes (Arnold et al. 1997). Scholars have

also examined management practices at carriers with strong

safety performance (Mejza et al. 2003).

Far less is understood regarding carriers’longitudinal perfor-

mance on safety behaviors. To date, a few studies have utilized

repeated-measures designs to examine carriers’safety behaviors

following either managerial or regulatory interventions. Naveh

and Marcus (2007) find carriers that implemented ISO 9002 saw

subsequent improvements in their driver and vehicle safety, as

measured by the SafeStat

2

system. Hickman and Hanowski

(2011) find that the implementation of on-board monitoring tech-

nologies reduced unsafe driving events. Miller and Saldanha

(2016) find that publically traded motor carriers had reduced

unsafe behaviors since the rollout of the CSA program. Miller

et al. (2017) also find that large, for-hire motor carriers had

reduced unsafe behaviors after the start of the CSA program,

although they found evidence that rates of improvement slowed

over time.

Although little work exists regarding carriers’longitudinal per-

formance as it pertains to safety behaviors following interven-

tions like the CSA program, a substantial body of research has

examined metered firms’performance following the start of other

IDPs (Fung et al. 2007). In particular, scholars have utilized both

difference-in-difference designs (Jin and Leslie 2003; Bennear

and Olmstead 2008; Chatterji and Toffel 2010; Doshi et al.

2013) and post-IDP-only designs (Hannan et al. 1994; Konar

and Cohen 1997; Blackman et al. 2004; Williams et al. 2005) to

examine whether metered firms’show performance changes after

IDP rollout. These studies have found strong evidence that

metered firms, on average, improve their performance. Regard-

less of research design, the primary explanation given for this

observed improvement is that publicizing firms’performance

increases their incentives to improve in disclosed domains or

face negative repercussions from stakeholders (Marshall et al.

2000; Stephan 2002; Berwick et al. 2003; Fung et al. 2007).

Other studies have subsequently identified factors that affect (i.e.,

moderate) the amount of change in disclosed metrics including

ownership type (Jin and Leslie 2009), performance at the start of

1

Studies examining factors that predict accidents and accident

rates (e.g., Moses and Savage 1994; Cantor et al. 2010) are

excluded from this literature review given a strong theoretical

distinction is made between safety behaviors (e.g., sudden lane

changes) and safety outcomes (e.g., accidents) in that the former

are more under firms’and individuals’influence, whereas the lat-

ter are oftentimes the product of many factors beyond actors’

control (Christian et al. 2009). This exclusion is not meant as a

critique of these works; rather, this delineation better positions

this article in the context of prior safety behavior investigations.

2

SafeStat was an IDP that was the forerunner to the CSA pro-

gram (Federal Motor Carrier Safety Administration 2011).

Safety Change Over Time 267