Discovery‐to‐Recall in the Automotive Industry: A Problem‐Solving Perspective on Investigation of Quality Failures

Published date01 April 2018
Date01 April 2018
AuthorJohn Ni,Xiaowen Huang
DOIhttp://doi.org/10.1111/jscm.12160
DISCOVERY-TO-RECALL IN THE AUTOMOTIVE
INDUSTRY: A PROBLEM-SOLVING PERSPECTIVE ON
INVESTIGATION OF QUALITY FAILURES
JOHN NI AND XIAOWEN HUANG
Miami University
Several recent high-profile product recalls raise the question of why com-
panies take so long to recall defective products from the market. The
recall timing decision is not a simple task, as companies constantly face
multiple, often competing goals during the recall process. In this research,
we examine variations in large automakersrecall timing decisions after
an initial report of a suspected quality failures. Drawing upon problem-
solving theory, we theorize about how five recall attributes impact
discovery-to-recall, defined as the time between a defective products ini-
tial discovery and its officially announced recall. To test our hypotheses,
we assembled a vehicle recall investigation dataset from recall reports filed
by the six largest automakers that sold passenger cars in the United States
from 2000 to 2012. Results from event history analysis reveal that discov-
ery-to-recall is longer for: (1) recalls that are triggered by external initial
reports, rather than internal initial reports; (2) recalls that are attributed
to suppliers, rather than automakers; (3) recalls that are associated with
design flaws, as opposed to manufacturing flaws; and (4) recalls with
more models involved. We also find that cumulative recall experience,
measured as the total number of previous recalls, shortens discovery-to-
recall. These findings improve our understanding of why the timing of
vehicle recalls varies considerably at the individual recall level. They also
highlight the value of problem-solving theory in vehicle recall research, as
well as quality management research.
Keywords: product recall; quality improvement; problem solving; archival data;
event history analysis
INTRODUCTION
Discovery-to-recall, defined as the time between a
defective product’s first discovery to its officially
announced recall, has become the subject of rising
concern, as witnessed in several high-profile product
recalls, such as Toyota’s 20092011 recall of millions
of vehicles due to sudden acceleration and GM’s 2015
recall of vehicles with faulty ignition switches. In the
Toyota recall, the long lapse from 2004, when
the problem was initially discovered, to 2009, when
the company finally announced its recall, raised
curiosity about why the discovery-to-recall spanned
nearly 5 years. The decision of when to recall is not a
simple task, because firms constantly face multiple,
often competing goals during the recall process. On
one hand, announcing and implementing a recall is
costly and can threaten a company’s survival.
A company has good reason to fully investigate sus-
pected defects and avoid a hasty recall that might give
credibility to unsubstantiated defect reports and subse-
quently lead to an ineffective remedy. For example,
Toyota reported savings of $100 million from delay-
ing a full recall of floor mats connected to sudden
acceleration (McCurry, 2010). On the other hand,
delaying a recall may lead to higher direct and indi-
rect costs through increased potential safety hazards,
legal and liability damages, fines, and erosion of
brand image. For example, in 2014 General Motors
was fined $35 million by the U.S. Department of
Transportation (DOT) for its delayed response to the
ignition switch defect in millions of vehicles (Beech &
Cowan, 2014). In this regard, even though a recall is
an adverse event, a quick reaction might attenuate the
damage.
April 2018 71
Journal of Supply Chain Management
2017, 54(2), 71–95
©2017 Wiley Periodicals, Inc.
The objective of this study was to investigate varia-
tions in large automakers’ timing of vehicle recalls
after a suspected defect is initially reported. This
study makes two important contributions to the pro-
duct recall and quality literature. First, with the
exception of Hora, Bapuji and Roth’s (2011) article
on toy recalls, little academic research attention has
been given to understanding variations in the timing
of recalls. The majority of the existing product recall
literature has focused on empirically estimating the
negative consequences of product recalls, including
reduction in shareholder wealth (Chen, Ganesan &
Liu, 2009; Hoffer, Pruitt & Reilly, 1988; Thirumalai
& Sinha, 2011) and change in product demand (Bar-
ber & Darrough, 1996; Dowdell, Govindaraj & Jain,
1992). Few studies have examined how factors such
as company strategy (Chen et al., 2009; Gray, Roth
& Leiblein, 2011; Haunschild & Rhee, 2004), R&D
focus (Thirumalai & Sinha, 2011), characteristics of
the production process and product attributes (Shah,
Ball & Netessine, 2016), and previous recalls (Kalaig-
nanam, Kushwaha & Eilert, 2013) impact the likeli-
hood of a recall. While these studies have advanced
our understanding of the impact and causes of pro-
duct recalls, they also manifest the paucity of
research on product recall processes, specifically, the
actual recall behavior of a company. By examining
which factors affect the length of discovery-to-recall,
our first contribution is to answer the call from lead-
ing operations and supply chain scholars for more
research on the product recall process (Lyles, Flynn
& Frohlich, 2008; Marucheck, Greis, Mena & Cai,
2011) in general, and the first stage of a product
recall process in particular. Arguably, the process of
determining and announcing a recall after an initial
report of a suspected defect is the most important
stage of the entire recall process. Decisions made
during this stage largely determine how the entire
recall process unfolds and thus can have significant
legal and financial consequences for a company, as
well as safety-related consequences for consumers
(Marucheck et al., 2011).
We conceptualize the process of determining and
announcing a recall after an initial report of a sus-
pected defect as a product recall investigation: The
task of the investigation is to understand the causes of
a quality failure and to develop a remedial solution.
We examine the impact of five recall attributes on
the speed of discovery-to-recall in the U.S. automotive
industry: the source of the initial defect report, the
party responsible for the defect, the type of defect,
the total number of vehicle models involved, and the
number of previous recalls by the company. Drawing
upon problem-solving theory, we argue that difference
in these five recall attributes poses distinct challenges
in problem diagnosis and solution search, which lead
to variations in task uncertainties associated with
problem-solving activities. We further posit that, as
task uncertainties increase, so do information process-
ing needs and subsequently, the length of the
discovery-to-recall process.
We apply event history analysis to empirically test a
large vehicle recall investigation dataset assembled
from automotive recall reports filed by the six largest
automakers that sold passenger cars in the United
States from 2000 to 2012. Our findings offer insights
on when companies are likely to swiftly announce a
recall after a suspected defect is initially reported.
These findings confirm the predictions of problem-sol-
ving theory that highlight the impact problem charac-
teristics have on problem-solving performance. They
also imply that automakers’ recall timing decisions are
primarily based upon the completion of recall investi-
gations. This view is corroborated by GM’s stance in
its 2014 investigation of Cobalt sedan steering defect:
“GM is redoubling efforts on pending product reviews
to bring them forward and to resolve them quickly.
We will not sacrifice accuracy for speed” (Healey &
Meier, 2014).
Second, this study is one of the first to address
product recalls from a problem-solving perspective.
Problem-solving theory provides us with a coherent
theoretical theme and guides our identification and
discussion of salient explanatory factors. Some
explanatory factors in this study have been previ-
ously unexamined, and others are offered as alterna-
tive explanations for the timing of a recall. It is
particularly noteworthy that using a problem-solving
lens allows us to focus on recall attributes as the
explanatory variable and enables us to examine vari-
ations in discovery-to-recall at the individual recall
level. According to this study’s dataset, while varia-
tions in average discovery-to-recall exist between the
six largest automakers, with the shortest at
6.37 months and the longest at 14.04 months, we
also observe considerable variations in discovery-to-
recall within each automaker. We note that the stan-
dard deviation of discovery-to-recall within each
automaker ranges from 5.78 to 16.29 months. Nis-
san, which had the lowest among the six automak-
ers, has a discovery-to-recall of 6.37 months on
average, with a minimum of 1 month and a maxi-
mum of 39 months. Given the significant variation
among discovery-to-recall in each company, it is
imperative to examine the recall timing issue beyond
the company level. This research complements the
existing supply chain and operations recall literature
conducted at the company level by empirically
examining factors that contribute to performance
variations within a company.
From a managerial standpoint, this research also
advances our understanding of the impediments to
Volume 54, Number 2
Journal of Supply Chain Management
72

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT