The moderating effect of supply base complexity on
the financial damage of a recalling event
Master Thesis
Faculty of Economics & Business
MSc Supply Chain Management
By
Name: Sven Blijleven
Student Number: S2735547
E-mail: s.blijleven@student.rug.nl
Supervisor
Name: dr. Xun (Bruce) Tong
University of Groningen, Faculty of Economics & Business, Department of Operations Management
Co-assessor
Name: prof. dr. D.P. van Donk
ABSTRACT
Product recalls in the pharmaceutical industries are happening more frequently than ever and the consequences for a firm’s financial value are often disastrous. However, current research lacks an understanding of and consensus on these consequences. Firms typically respond by coordinating their supply base. Supply base complexity decreases responsiveness and increases coordinating costs. Hence, this research tests whether recalling events cause a significant negative reaction on the stock markets and whether the degree of supply base complexity has an influence on these consequences. This paper combines the principles of event study and regression analysis in order to test these hypotheses. Recalling events are found to have a significant negative effect on the focal company’s value. Nevertheless, supply base complexity did not show to have a significant effect on this causal relationship. Despite the fact that firms respond to recalling events with their whole supply base, this paper’s supply base complexity measure, which was based on structural complexity, did not show a significant role of supply base complexity. This research offers a foundation for future research to clarify the role of supply base complexity in recalling events.
Table of content
ABSTRACT ... 2
1. INTRODUCTION ... 5
2. LITERATURE REVIEW ... 7
2.1. Recalling events ... 7
2.2. Financial consequences of recalling events ... 7
2.3. Moderating factors on firm value following a recalling event ... 8
2.4. Supply base complexity ... 9
3. HYPOTHESIS DEVELOPMENT ... 11
3.1. Financial consequences of recalling event ... 11
3.2. Supply base complexity as a moderator ... 12
4. METHODOLOGY ... 14
4.1. Data collection and sample selection ... 14
4.2. Variable definitions and measurements ... 15
4.2.1. Dependent variable ... 15 4.2.2. Moderator ... 16 4.2.3. Control variables ... 16 4.3. Research design ... 17 5. RESULTS ... 20 5.1. Event study ... 20
5.1.1. Event study results ... 20
5.1.2. Sensitivity analysis of the event study results ... 21
5.1.3. Deeper analysis of data ... 21
5.2. Regression analysis ... 23
5.2.1. Regression analysis results ... 23
5.2.2. Analysis of different timeframes ... 25
5.3. Additional analysis ... 25
6. DISCUSSION ... 27
6.2. Theoretical contributions ... 27
6.3. Managerial implications ... 29
6.4. Limitations and future research ... 30
7. REFERENCES ... 32
8. APPENDIX ... 36
Appendix A: List of tables & figures ... 36
1. INTRODUCTION
On the 30th of July in 2018 drug company Ei, which is a Florida-based subsidiary of drug
manufacturer “Product Quest, LLC”, was forced by the FDA (Federal Drug Agency) to recall their nasal spray due to contamination and bad product quality. Within a week the company was forced to file for bankruptcy after it became evident that the costs of the recall would end up at over 3 billion US dollars. As a result, approximately 300 people lost their jobs and over 3 billion dollars’ worth of damage was claimed (Bruno, 2018). This is just one example that demonstrates the potential magnitude of the consequences of a product recall. Due to its recent surge in frequency, research on product recalls in the United States is more relevant than ever (Bala, Bhardwaj, & Chintagunta, 2015; Jayaraman, Alhammadi, & Simsekler, 2019). For example, in 2017, 325 products were recalled in the pharmaceutical industry, which is a 12.5% increase compared to 2016. The increase is even more remarkable considering the number of items that were recalled: 206 million. This is 350% of the amount of items in 2016 (“Product recall trend: up strongly in 2018 - Pharmaceutical Commerce,” 2018). Some researchers insist that a major product recall is one of the worst things that could possibly happen to a supply chain because of revenue loss. Next to this, they stat that it is hard, if not impossible, to recover from this loss (Heerde, Helsen, & Dekimpe, 2007).
Substantial research has been done on the consequences of recalling events. Most papers conclude that recalling events have a significant negative influence on the recalling firm’s financial value (Hendricks & Singhal, 2003; Hoffer, Pruitt, & Reilly, 1988; Ni, Flynn, & Jacobs, 2016, 2014). However, literature is not unanimous on this causal effect (Hoffer et al., 1988; Thirumalai & Sinha, 2011). Thirumalai & Sinha (2011), for example, did not find a significant relationship between recalling events and firm value. Even though the consequences of recalling events have been studied extensively, the underlying mechanisms remain largely unexplored. Recalling strategies (Chen, Ganesan, & Liu, 2009; Ni et al., 2014; Zhao, Li, & Flynn, 2013), the type of industry (Zhao et al., 2013) and competition (Ball, Shah, & Wowak, 2018) are argued to have an effect on the relationship between recalling events and firm value. These factors, however, do not take the supply chain into account. This is remarkable since coordination of the whole supply chain is required in order to respond efficiently to such a supply chain disruption (Chao, Iravani, & Savaskan, 2009; Wowak & Boone, 2015).
(Brandon-Jones, Squire, & Van Rossenberg, 2015; Chen et al., 2009; Liu, Liu, & Luo, 2016; Steven, Dong, & Corsi, 2014). Supply base complexity is associated with increasing coordination costs and decreasing supplier responsiveness (Choi & Krause, 2006; Craighead, Blackhurst, Rungtusanatham, & Handfield, 2007; Zsidisin & Wagner, 2011). Multiple researchers have suggested a potential link between supply base complexity and the financial impact of a recalling event (Craighead et al., 2007; Wowak & Boone, 2015). However, to the best of my knowledge, this relationship has not been empirically tested yet. Consequently, this thesis focuses on two research questions. First: Does a recalling event have a negative influence
on a company’s financial vale? Second: Whether and to what extent is there a moderating effect of supply base complexity on the financial value after a recalling event?
The trend of increasing supply base complexity in combination with the high potential impact of recalling events raises the need for a better understanding of the potential moderating role of supply base complexity. This thesis aims to provide two main contributions. First, it aims to provide support to prior studies on the consequences of recalling events, because consequences of recalling events are disputed by many researchers. This research aims to clarify this mechanism once and for all. Second, it aims to provide a better understanding of the supply base characteristics that affect the damage of a recalling event. Several researchers have highlighted the importance of supply base characteristics, but so far the link between supply base complexity and the financial consequences of a recalling event has not been established (Craighead et al., 2007; Wowak & Boone, 2015). The contribution of this paper is confirmatory, which means that this relationship will be empirically tested. The results of this research can be used in order to get a better understanding of the consequences of product recalls. Additionally, this thesis gives an insight in the type of companies that are more likely to suffer from negative consequences than others.
2. LITERATURE REVIEW
2.1. Recalling events“A product recall is the removal from the market of a product that was found to be hazardous
(Desai & Patel, 2014), with ‘a substantial risk of injury to the public’ (Rubin, Murphy, & Jarrell, 1988)” (Ni et al., 2016, p. 5404). A recall occurs when a firm is penalized for not
providing a safe product to the market (Kalaignanam, Kushwaha, & Eilert, 2013). The topic gained interest due to heightened customer awareness, increased global and domestic competition and stronger product safety regulations (Luo, 2008). Literature on this topic often refers to recalling events as an important source of supply chain disruptions (Bode & Wagner, 2015; Kalaignanam et al., 2013; Ni et al., 2016; Zhao et al., 2013) and supply chain glitches (Hendricks & Singhal, 2003). Supply chain disruptions and glitches both represent the same issue and can be defined as “the firm’s inability to match supply and demand” (Hendricks & Singhal, 2003, 2005, p. 35).
2.2. Financial consequences of recalling events
Predominantly, literature on the consequences of recalling events has focused on the automobile and pharmaceutical industry, which are both industries that are characterized by a relatively high frequency of recalls (Zhao et al., 2013). According to most research, recalling events have a significant negative impact on the company’s financial valuation (Hartman, 2006; Hoffer et al., 1988; Ni et al., 2016, 2014; Zhao et al., 2013). There is, however, no consensus on the significance of the financial damage of a recalling (Hoffer et al., 1988; Thirumalai & Sinha, 2011). Thirumalai & Sinha (2011, p. 389) for example claim that “Although medical device
recalls represent a firm’s failure to deliver quality over the product life cycle causing significant discomfort to consumers, firms face little by way of capital market penalties”.
Hendricks & Singhal (2003) studied the effect of supply chain disruptions on shareholder wealth. They developed the framework displayed in Figure 2.1 to explain the effect of supply chain disruptions on shareholder value. Hendricks & Singhal (2003) claim that the factors affecting shareholder value can be categorized in operational issues (Cash Flows, Earnings and ROA) and intangible issues (Credibility & Reputation).
Figure 2.1: Linking supply chain performance to shareholder value
(Hendricks & Singhal, 2003)
Supply chain disruptions are disabling a firm to deliver their products. Consequently, the sales drop and operational costs associated with the recalling of products increase (Hendricks & Singhal, 2003). The operational costs are costs that affect the short-term profitability of the company. These include costs that have to be made in order to recall the products. Intangible costs refer to costs, such as brand image, that affect the company’s profitability on a longer term (Luo, 2008).
2.3. Moderating factors on firm value following a recalling event
value (Brandon-Jones et al., 2015). Furthermore, the role of strategies, the type of industry and the effect on the firm’s stock market value has been examined (Zhao et al., 2013). According to Chen et al. (2009) and Wowak & Boone (2015) coordination of the whole supply chain is necessary in order to respond efficiently to a supply chain disruption. The complexity of the supply base could therefore have a strong influence on the efficiency of the response to this disruption (Chao et al., 2009; Craighead et al., 2007; Wowak & Boone, 2015). The quality of the response of a firm to a particular event will be represented by its stock market value (Mackinlay, 1997; Siegel & McWilliams, 1997). Hence, it is striking that the influence of supply base complexity has not been tested before. The next paragraph will elaborate on supply base complexity.
2.4. Supply base complexity
Choi and Krause (2006) developed the most commonly used definition of a supply base. Their definition is based on research on supply networks. They define a supply network as “a network
of firms that exist upstream to any one firm in the whole value system” (Choi & Krause, 2006,
p. 352). This includes firms that are not in direct contact with each other. This is where the distinction between a supply network and a supply base becomes evident. The supply base consists of only those firms the company is in direct contact with. Consequently, Choi and Krause (2006, p. 639) define the supply base as “only those suppliers that are actively managed
through contracts and the purchase of parts, materials and services”.
In order to get a better understanding of supply base complexity, it is important to dive deeper into the definition of system complexity. The foundation of system complexity was established by Simon (1961) who claims that a system is complex if it is “made up of a large number of
parts that interact in a non-simple way” (Bode & Wagner, 2015, p. 79). According to his
theory, there are two factors influencing complexity: structure/size of the system and the behavior between the different nodes in the system. These two factors have been the basis of other classifications of complexity. The structural and behavioral dimensions of complexity are argued to be interrelated (Bode & Wagner, 2015). Bode & Wagner (2015) claim that if the size of the network increases, so does the variety in interactions (behavioral dimension). Simon’s (1961) research has been the basis of most supply base complexity models that are used throughout this research (Choi & Krause, 2006; Vachon & Klassen, 2002).
complex and refer to them as complex adaptive systems (Choi et al., 2001; Choi & Krause, 2006). A complex adaptive system is defined as “a collection of firms that seek to maximize
their individual profit and livelihood by exchanging information, products, and services with one another” (Choi et al., 2001, p. 365). According to Choi et al. (2001) classifying a supply
network as a regular system is too narrow, rather the supply network (and supply base) should be classified as a complex adaptive systems. This is due to the complex and adaptive nature of a supply network. Consequently, they propose that any supply base and network is complex, however the degree of complexity differs.
The factor that has the highest influence on the degree of supply base complexity is the supply base size (Bode & Wagner, 2015; Choi & Krause, 2006; Vachon & Klassen, 2002). This refers to the structural dimension of Simon’s (1961) definition of complexity. It is also in line with Bode and Wagner’s (2015) statement of interrelatedness. They claim that if the supply base increases in size, so does the variety of interactions. Furthermore, Choi and Krause (2006) argue that there are two other factors that are influencing supply base complexity, which refer to the behavioral part of Simon’s definition of complexity: differentiation of suppliers and inter-relationships among suppliers. This implies that higher the differentiation of suppliers and the higher the inter-relationships among suppliers, the higher the complexity (Choi & Krause, 2006). This definition is quite similar to the one proposed by Vachon & Klassen (2002, p. 219):
“supply chain complexity is defined here to include the concepts of numerousness, interconnectivity, and systems unpredictability”. Numerousness and interconnectivity are both
3. HYPOTHESIS DEVELOPMENT
3.1. Financial consequences of recalling eventAs mentioned before, recalling events cause supply chain disruptions and are argued to be one of the worst things that can happen to a firm (Heerde et al., 2007). The financial consequences are often disastrous and, as the example in the introduction shows, can even result in bankruptcy (Bruno, 2018). Hendricks and Singhal’s (2005) framework (Figure 2.1) forms a good foundation for developing a hypothesis on the financial consequences of a recalling event. According to them, costs associated with recalling events can be categorized in operational costs and intangible costs.
Recalling events disrupt the supply chain and disrupt the firm’s operations. To encounter this disruption, costs are to be incurred. According to Thirumalai & Sinha (2011, p. 379), these costs consist of “costs of correcting/replacing the defective product, transaction costs of recall process and costs of unsold inventory”. Eventually this increase in costs will lower the profitability of the firm. According to the signaling theory, this decrease in profitability sends out negative signals, causing shareholders to lose their confidence in the company. This will have a negative impact on the shareholder value of the firm (Connelly, Certo, Ireland, & Reutzel, 2011).
Signaling theory could also explain the intangible costs associated with recalling events. Based on the available information, stakeholders create their own theory on the performance of a focal firm’s operations. Recalling events send out negative information about the firm’s performance. This new information might be enforced by word to mouth or public boycotts, which could bring the firm in a negative spiral (Folkes, 1984). This information will play part in the shareholder’s evaluation of the firm and might affect how (potential) shareholders view the future profitability of the company (Chen et al., 2009; Folkes, 1984; Thirumalai & Sinha, 2011). Consequently, shareholders will start selling their shares, which could have a negative influence on the firm’s stock value (Ni et al., 2016; Zhao et al., 2013). Hence, looking at the shareholder value in a short timeframe after the event will represent the impact of this recalling event effectively (Siegel & McWilliams, 1997).
Kahneman, 2005). According to the principles of loss aversion, buyers evaluate the stock based on a reference point, rather than on the general state. Recalling events deteriorate the company’s financial position relatively to the prior position (before the recalling event). Hence, they send out a negative signal and are, consequently, more likely to be picked up by a receiver. The receiver in this case is the (potential) shareholder. The costs associated with a product recall identified by Hendricks & Singhal (2005) pose a risk on the (potential) shareholder’s wealth. Consequently, the (potential) shareholder will have an aversion towards the stock of the focal company. These effects will be represented in the stock market in a short timeframe following the event. (Siegel & McWilliams, 1997). Hence, I propose the following hypothesis:
Hypothesis 1: A recalling event has a negative financial impact on stock market value of a focal firm in the pharmaceutical industry.
3.2. Supply base complexity as a moderator
It is hypothesized that a recalling event negatively affects the stock market value of a company. There are, however, factors that might moderate this effect. Craighead et al. (2007, p. 141) argue that “an unplanned event that disrupts a complex supply chain would be more likely to
be severe than the same supply chain disruption occurring within a relatively less complex supply chain”. Complexity decreases the responsiveness (Craighead et al., 2007), which
increases the severity of the impact of a supply chain disruption (Zsidisin & Wagner, 2011). After a literature review on the current stance on recalling research, a similar proposition has been proposed by Wowak & Boone (2015). They further highlight the need for testing the effect of supply chain complexity on the relationship between recalling events and the severity of financial damage. However, until now this proposition has not been empirically tested in the pharmaceutical industry.
(Bozarth, Warsing, Flynn, & Flynn, 2009; Choi & Krause, 2006; Craighead et al., 2007; Zhao et al., 2013). These costs will decrease the firm’s financial performance, which will be represented in the market valuation (Siegel & McWilliams, 1997).
The second underlying mechanism revolves around supply base responsiveness. Supply base responsiveness is defined as “the extent to which channel members respond cooperatively to
environmental changes” (Wu, Yeniyurt, Kim, & Cavusgil, 2006, p. 495). Zsidisin & Wagner
(2011) claim that responsiveness and flexibility are critical to respond well to a recalling event and to minimize the costs associated. Resilient supply chains, which are flexible and responsive, are better able to respond to supply chain disruptions (Choi & Krause, 2006; Peck, 2005; Zsidisin & Wagner, 2011). Smith et al. (1991) claim that complexity decreases responsiveness by delaying or even completely blocking the information that is shared in a network. This statement was later supported by Choi & Krause’s (2006) widely accepted research on supply base complexity. Decreased responsiveness to a supply chain disruption leads to an increase in the severity of financial damage to the focal firm (Choi & Krause, 2006). When a recalling event occurs, coordination is needed to efficiently respond to this supply chain disruption and minimize the potential damage (Chao et al., 2009; Wowak & Boone, 2015). An increase in supply base complexity could increase the severity of the damage this recall does to a focal company by increasing coordination costs and a decreasing supplier responsiveness. Hence, this research hypothesizes that supply base complexity has a negative influence on the responsiveness of the network.
4. METHODOLOGY
Figure 4.1 displays a visual representation of the conceptual model that is tested in this research.
To test the first hypothesis, this research utilizes the principles of event study. Event study is a type of econometric modelling that is used frequently when the magnitude of the consequences of a certain event is studied (Hendricks & Singhal, 2003; Ni et al., 2016; Siegel & McWilliams, 1997; Thirumalai & Sinha, 2011; Zhao et al., 2013). Event study benchmarks the actual returns on a certain stock against the expected (or normal/market) returns of that stock based on an estimation period. For the event study, I assume the semi strong form of the efficient market hypothesis, meaning that stock prices are effectively reflected by external information. An estimation window of 250 days, stretching from Day -270 to Day -21 is used, which is common for this type of research (Mackinlay, 1997; Ni et al., 2014; Siegel & McWilliams, 1997). The date of the earliest recall announcement published by the FDA is the event date. Following the research design of Chen (2009), Hendricks & Singhal (2003) and Ni et al. (2014) the event window is two days (-1,0). The event window should be as short as possible to ensure that no other events influence the abnormal returns (Siegel & McWilliams, 1997).
The second hypothesis tests whether supply base complexity moderates the abnormal returns following a recalling event. This is studied using regression analysis. The regression analysis tests whether and to what extent supply base complexity influences the abnormal returns following a recalling event.
Figure 4.1: Conceptual model
4.1. Data collection and sample selection
events between 2008-2017 were retrieved. 2037 of these recalling events included privately owned firms and were deleted from the sample, leaving a sample size of 662 recalling events. In order to calculate the abnormal returns, there needs to be a clean estimation window. An estimation window is considered to be clean if the focal firm experiences no other recalling events during this period (Thirumalai & Sinha, 2011). 505 of these recalling events included firms that experienced another recall during the estimation period. Hence, these firms were excluded from the sample leaving a sample of 157 recalling events. The analysis required a minimum of 40 valid observations per firm. These observations refer to the amount of daily data that can be extracted. If this criterion was not met, these companies were excluded from the analysis. 61 cases did not meet this criterion, leaving a sample size of 96 recalling events.
Due to the confidentiality surrounding the pharmaceutical industry it is hard to establish a supply base. This is done by analyzing news articles, which are gathered using the “Factiva” database for news announcements. The search string that is used in order to extract the data is presented in Appendix B. By analyzing and coding news announcements of buyer-supplier relationships, an image of the supply base is established. During the coding process, the following codes were used: Buyer/Supplier, Public/Private, Parent company, Country of origin
(of both companies). By coding approximately 9,000 news announcements on buyer-supplier
relationships, a supply base of firms in the pharmaceutical industry was established. Data regarding the stock market valuation of a company was gathered using the online database of CompuStatt from WRDS.
4.2. Variable definitions and measurements 4.2.1. Dependent variable
4.2.2. Moderator
“From the perspective of a focal company, the most immediate area of supply base complexity is the number of suppliers in the supply base” (Choi & Krause, 2006, p. 641). Furthermore,
Bode & Wagner (2015) claim that the structural and behavioral dimensions of complexity are interrelated; if the size of the network increases, so does the variety of interactions and behaviors. Consequently, supply base complexity is estimated and operationalized as the number of suppliers that are present in the supply base. To overcome the limitations of the timeframe of the dataset, the supply base is estimated based on the total number of first-tier suppliers between 2008-2017. This is necessary because suppliers are generally connected to a company for multiple years. However, the data on suppliers of firms that experienced a product recall in 2008 would only include the deals that were initiated and established in 2008. Hence, the average supply base is estimated as the number of first-tier suppliers between 2008-2017.
4.2.3. Control variables
The regression analysis includes three control variables that are commonly used in this type of research: company size, recall size and potential damage of the product (Ni et al., 2016, 2014). The variable company size is operationalized as the number of employees (Ni et al., 2016; Thirumalai & Sinha, 2011). In order to get a better model, the companies are categorized based on their size. These categories are presented in Table 4.1.
Table 4.1: Categories company size Category Number of employees
1 <100 employees
2 101 - 1.000 employees
3 1001 - 10.000 employees
4 10.001 – 99.999 employees
5 >100.000 employees
Recall size is operationalized as the number of units that are recalled (Ni et al., 2014). In order
to get a more meaningful model this variable is standardized. Finally, for potential damage of
the product the FDA’s classification is used. The FDA classifies product recalls into three
Table 4.2: FDA categories Category Description
1 “A situation in which there is a reasonable probability that the use of or exposure to a violative product will cause serious adverse health consequences or death.”
2 “A situation in which use of or exposure to a violative product may cause temporary or medically reversible adverse health consequences or where the probability of serious adverse health consequences is remote.”
3 “A situation in which use of or exposure to a violative product is not likely to cause adverse health consequences.”
All variables and their measures are summarized in the Table 4.3.
Table 4.3: Summary of variables
Variable name Type Operationalization Measure
Stock market value Dependent variable Abnormal returns (0,-1) 𝐴𝑅#$= 𝑅#$− (𝛼#+ 𝛽#𝑅+$)
Supply base complexity
Moderator The number of first-tier suppliers in the period 2008-2017
Number of first-tier suppliers
Recall size Control variable The number of units recalled in the
period 2008-2017
Z-score (Number of recalls)
Company size Control variable The number of employees (based
on the year report of 2018)
Company size categories (1 =<100 employees, 2 = 101 - 1000 employees, 3 = 1001 - 10.000 employees, 4 = 10.001 - 99.999 employees, 5 =>100.000 employees)
Potential harm of the product
Control variable FDA classification for potential damage of product that is recalled
FDA classification for potential damage of product that is recalled
4.3. Research design
For the event study, the single factor market method (or simple market model) is used. This model is commonly used in research focusing on the financial consequences of recalling events (Chen et al., 2009; Hendricks & Singhal, 2003, 2005; Ni et al., 2016; Thirumalai & Sinha, 2011). The abnormal returns in the days surrounding a recall is compared to the expected market return. This event study model is presented below:
𝑅#$ = 𝛼# + 𝛽𝑅+$ + 𝜀#$ 𝑅#$ = 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑓 𝑓𝑖𝑟𝑚 𝑖 𝑜𝑛 𝑑𝑎𝑦 𝑡
𝑅+$ = 𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘𝑠 𝑜𝑛 𝑑𝑎𝑦 𝑡
𝛽# = 𝑆𝑙𝑜𝑝𝑒 𝑜𝑓 𝑟𝑒𝑔𝑟𝑒𝑠𝑠𝑖𝑜𝑛 𝑅#$ 𝑜𝑛 𝑅#+
The abnormal returns are calculated as following:
𝐴𝑅#$ = 𝑅#$− (𝛼# + 𝛽#𝑅+$)
And for the event window as a whole I use the Cumulative Abnormal Return (CAR):
𝐶𝐴𝑅#(𝑡E, 𝑡G) = H 𝐴𝑅#$
$I
$J$K (Siegel & McWilliams, 1997)
The estimation of 𝛼# and 𝛽# is done using the ordinary least square regression of the normal returns over the estimation window on the market returns. One parametric test (independent sample t-test) and two non-parametric tests (Wilcoxon signed-rank test and binomial test) are used to test the significance of the following hypothesis:
𝐻M: 𝐴𝑅 = 0
𝐻E: 𝐴𝑅 ≠ 0
This means that the null hypothesis (H0) assumes that the abnormal returns following the event are not significantly different zero. If the null hypothesis is rejected, it means that there are abnormal returns, which indicates that the stock prices are affected by the (recalling) events.
A regression analysis is performed in order to test the second hypothesis. This regression analysis tests whether supply base complexity has a significant effect on the abnormal returns following the recalling event. The regression analysis will test the significance of the following model: 𝐴𝑅# = 𝛼 + 𝛽E∗ 𝑆𝑢𝑝𝑝𝑙𝑦𝐵𝑎𝑠𝑒𝐶𝑜𝑚𝑝𝑙𝑒𝑥𝑖𝑡𝑦 + 𝛽G∗ 𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝑆𝑖𝑧𝑒 + 𝛽U∗ 𝑅𝑒𝑐𝑎𝑙𝑆𝑖𝑧𝑒 + 𝛽V ∗ 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙𝐷𝑎𝑚𝑎𝑔𝑒 + 𝜀 𝑤ℎ𝑒𝑟𝑒, 𝐴𝑅# = 𝑇ℎ𝑒 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 𝑓𝑜𝑟 𝑓𝑖𝑟𝑚 𝑖 𝛼 = 𝑖𝑛𝑡𝑒𝑟𝑐𝑒𝑝𝑡 𝛽] = 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑜𝑓 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑥 𝜀 = 𝑡ℎ𝑒 𝑒𝑟𝑟𝑜𝑟 𝑡𝑒𝑟𝑚
5. RESULTS
5.1. Event study5.1.1. Event study results
The descriptive statistics of the daily abnormal returns are presented in Table 5.1. The first column of Table 5.1 represents the day before the event (Day -1), the second column represents the day of the event (Day 0) and the third column represents both days in the form of the cumulative abnormal return (CAR). The day before the recall announcement date was included since there is a possibility that information regarding the recall reaches the market before the recall announcement. The average abnormal return and median abnormal return on this day are both -0,02%, which is very close to 0. Furthermore, the independent t-test (t-statistic = -0,127 & p-value = 0,991), Wilcoxon signed-rank test (Z = -0,011 & p-value = 0,991) and Binomial test (p-value = 0,919) all indicate that the abnormal returns do not significantly differ from zero. Hence, it can be assumed that prior released information does not significantly influence the abnormal returns in the sample. Therefore, the focus of the rest of the analysis will be on the abnormal returns of the event date (Day 0).
Table 5.1: Daily abnormal returns and test statistics
Day -1 Day 0 Event Period (CAR) (-1,0)
Mean abnormal return -0,02% -1,61% -1,63%
T-statistic -0,127 (p-value = 0,899) -1,376 (p-value = 0,172) -1,339 p-value = 1,84
Median abnormal return -0,02% -0,01% -0,08%
Wilcoxon signed-rank test statistic -0,011 (p-value = 0,991) -0,782 (p-value = 0,434) -0,603 (p-value = 0,547) Percentage of negative abnormal returns 51,04% 51,04% 51,04%
Binomial test significance 0,919 0,919 0,919
be rejected, and it cannot be concluded that the abnormal returns are significantly different from 0.
5.1.2. Sensitivity analysis of the event study results
In order to calculate the abnormal returns, the simple market model was used. There are, however, different econometric models that can calculate the abnormal returns following an event. This section compares the results of an additional model, the market-adjusted model, to the model that was used in the prior analysis.
The simple market model benchmarks the actual returns of a focal firm’s stock to the expected market return, while the market-adjusted model benchmarks the actual market return against actual return on the focal firm’s stock. Consequently, the Abnormal Returns of the stock of firm 𝑖 (𝐴𝑅#$) is calculated as follows:
𝐴𝑅#$ = 𝑟#$− 𝑟+$
A comparison of some key descriptive statistics between the market model and the market-adjusted model is presented in the Table 5.2. The summary statistics of both models are very similar and there does not seem to be a significant difference between the two models. To confirm this, the abnormal returns are tested using the Wilcoxon signed-rank test. The results are not significant (Z-score of -0,640 and p-value of 0,522). This confirms that the null hypothesis cannot be rejected, meaning that the abnormal returns of both models do not significantly differ.
Table 5.2: Comparison event study models (day 0)
Mean abnormal return (%) Median abnormal return (%) Negative values (%)
Market- Model -1,61% -0,01% 49 (51,04%)
Market-adjusted model -1,49% -0,01% 50 (52,10%)
5.1.3. Deeper analysis of data
Table 5.3: Ranks Wilcoxon signed-rank test
Class N
1 AN_D0 - 0 Negative ranks 8a
Positive ranks 2b
2 AN_D0 - 0 Negative ranks 30a
Positive ranks 23b
3 AN_D0 - 0 Negative ranks 11a
Positive ranks 22b
a = AN_D0 < 0, b = AN_D0 > 0, c = AN_D0 = 0
Table 5.4: Descriptive statistics per recall class
Class 1 Recall Class 2 Recall Class 3 Recall
Number of observations 10 53 33
Mean abnormal return (%) -7,54% -0,68% -1,30%
Median Abnormal Return (%) -0,60% -0,23% +0,36%
Negative Abnormal Returns (%) 8 (80%) 30 (56,60%) 11 (33,33%)
T-statistic -1,090 (p-value = 0,304) -2,357 (p-value = 0,022*) -0,486 (p-value = 0,630)
Wilcoxon signed-ranked test statistic -1,988 (p-value = 0,047*) -1,961 (p-value = 0,050*) -2,242 (p-value = 0,025**)
Binomial test significance 0,109 0,410 0,080
*= based on positive ranks significant (sig. <0,05). **= based on negative ranks significant (sig. <0,05)
Classifying the data and rerunning the tests offers some interesting insights. According to the Wilcoxon signed-rank test, the abnormal returns of class 1 recalls are significantly lower than 0. However, based on the one sample t-test and binomial test significance, the abnormal returns of class 1 recalls are not significantly different from zero. However, this is most likely solely due to the small sample size (N = 10), which is likely to cause violations of the normal distribution. Hence, this research relies on the Wilcoxon signed-rank test, which does not assume normality. This test statistic is significant (test statistic = -1,988 & p-value = 0,047) on negative ranks. Consequently, it can be concluded that companies involved in a class 1 recall experience abnormal returns that are significantly lower than zero.
Based on the Wilcoxon signed-rank test it can be concluded that the abnormal returns of class 3 recalls are significantly different from zero (Z-score = -2.242, p-value = 0,025) . However, this is based on “positive ranks”, which means that the abnormal returns are significantly higher than 0, which contradicts hypothesis 1. The binomial test (sig. = 0,080) and independent sample t-test (t-statistic = -0,468 and p-value = 0,630) are both not significant. This research relies on non-parametric tests (Wilcoxon signed-rank test) over parametric tests. Hence, it can be concluded that companies experiencing class 3 recalls abnormal returns that are significantly higher than 0.
In conclusion, H1 is confirmed for both class 1 and class 2 recalls. However, there seems to be a reverse effect for class 3 recalls, where abnormal returns were, according to the Wilcoxon signed-rank test, even significantly higher than 0 instead of lower. In order to understand these results, it is important to get a grasp of the factors that influence the degree to which a recall leads to abnormal returns. This is tested using regression analysis in the next section. In chapter 5.3, I will shed a light on these conflicting results using the outcomes of the regression analysis.
5.2. Regression analysis
5.2.1. Regression analysis results
The second hypothesis states that supply base complexity has a negative influence on the abnormal returns following a recalling event. This is tested using a regression analysis with the abnormal returns (of day 0) as a dependent variable and supply base complexity as an independent variable. This following model (model 1) is used in order to test the second hypothesis:
𝐴𝑅# = 𝛼 + 𝛽E∗ 𝑆𝑢𝑝𝑝𝑙𝑦𝐵𝑎𝑠𝑒𝐶𝑜𝑚𝑝𝑙𝑒𝑥𝑖𝑡𝑦 + 𝛽G∗ 𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝑆𝑖𝑧𝑒 + 𝛽U∗ 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙𝐻𝑎𝑟𝑚 + 𝛽V∗ 𝑅𝑒𝑐𝑎𝑙𝑙𝑆𝑖𝑧𝑒 + 𝜀.
Table 5.5: Descriptive statistics variables regression analysis
1 2 3 4 5
1. Abnormal return 1
2. Supply base complexity 0,069 1
3. FDA classification 0,102 -0,073 1
4. Company size (categories) 0,280 0,279 0,151 1
5. Recall size (Z-score) 0,015 -0,025 -0,027 0,091 1
Mean -1,61% 3,54 2,24 3,04 0,00
Standard Div. 11,5% 7,48 0,63 1,11 1,00
The statistics pf two models are calculated. Model 1 contains all control variables and supply
base complexity. Model 2 contains all control variables but does not contain supply base complexity. Both models will be assessed in order to determine whether adding supply base complexity will improve the model.
Table 5.6: Regression models
Variables Model 1 Model 2
(Constant) -0,126b -0,126b
Supply base complexity <0,01
Company size 0,028b 0,028a Recall classification 0,011 0,11 Size of recall 0,002 -0,001 F-value 2,036 2,738c R2 0,082 0,082 Adjusted R2 0,042 0,052
a = significant (sig. <0,01). b = significant (sig. <0,025). c = significant (sig. <0,05
Table 5.6 presents the result of the regression analysis. Model 1 is not significant with an F-value of 0,453 and significance of 0,503. R2 and adjusted R2 are 0,005 and -0,006 respectively,
which indicates that the model does not explain the variance of the dependent variable well. Moreover, the coefficient of supply base complexity approximates 0 and is not significant with a p-value of 0,997. Model 2 is a better model; with an F-value of 2,738 and significance of 0,048 the model has a significant fit. Furthermore, the adjusted R2 is higher for model 2, which
indicates that model 2 better explains the variance of the dependent variable.
Supply base complexity is not significant in model 1. Furthermore, the model that contains the
variable supply base complexity (model 1) is less significant than the model that does not contain supply base complexity (model 2). Hence, it can be concluded that supply base complexity does not have a significant influence on the abnormal returns following a recalling event.
5.2.2. Analysis of different timeframes
This research focuses on a short timeframe after the event. This timeframe includes only the event date. The decision for this timeframe might have biased the results. For example, the influence of some mechanisms, such as supplier responsiveness, might take longer to become apparent. To ensure that this set timeframe did not cause a bias to the results, I will also analyze whether supply base complexity had an influence on the firm value up to ten days after the event. This is done using the cumulative abnormal returns (CAR) as the dependent variable instead of the abnormal returns of day 0. Two additional time frames are analyzed. One stretching from the event date until the fifth day after the event and one until the tenth day after the event. The results are presented in the Table 5.7:
Table 5.7: Analysis of different timeframes
*= sig.<0,05
Supply base complexity remains non-significant in every model. Therefore, it can be concluded that the results were not biased due to the timeframe that was set. Consequently, supply base complexity remains non-significant until ten days after the event.
5.3. Additional analysis
In the regression analysis the variable potential damage of the product did not have a significant influence on the abnormal returns. Hence, it is likely that another factor could explain why companies experiencing class 1 or class 2 recalls have significantly negative abnormal returns and companies experiencing class 3 recalls do not. In the regression analysis, company size showed a strong significant influence on the abnormal returns following a recalling event.
Time frame Mean CAR (%) Model 1 (F-value Model 2 (F-value) Significance supply
base complexity (in model 1)
0 (event date) -1,61% 2,032 2,738* 0,981
(0,5) -2,00% 1,077 1,030 0,274
Differences in company size might explain why class 1 and 2 recalls were followed by significant negative abnormal returns and class 3 recalls were not. In order to test whether company size explains the different results among classes the descriptive statistics of the variable company size split per class need to be analyzed. The descriptive statistics of the variable “company size” split by class are presented in Table 5.8.
Table 5.8: Descriptive statistics Company Size
FDA Classification N Minimum Maximum Mean Std. Deviation
1 10 29 90200 16744,10 27227,981
2 53 29 134000 20194,75 34301,908
3 33 16 134000 29656,79 36702,593
Class 3 recalls included larger companies with a mean that is approximately 46% higher than companies that had class 2 recalls and 77% higher than class 1 recalls. Comparing the means of class 1 & 2 recalls with the mean of class 3 recalls using the independent samples t-test provides a significance of 0,06, which is significant using a 10% significance. However, it is not significant using the significance level of this paper: 5%. This is likely due to the fact that the sample size is quite small, and the assumption of normality is not met. The means are, however, very far apart and it is very likely that the means would have been significantly different on a 5% significance level if the sample size was larger. According to the regression analysis, company size can cause the abnormal returns to be less negative (or more positive). Therefore, it is very likely that H1 was not confirmed for class 3 recalls because class 3 recalls included larger companies, who are able to better respond to recalling events compared to smaller companies.
6. DISCUSSION
6.1. Major findingsAccording to the analysis, hypothesis 1 is partially supported. This means that in essence a recalling event is followed by a negative reaction on the financial markets which devaluates the stock market value of the focal company. This is in line with most research (Hartman, 2006; Hoffer et al., 1988; Ni et al., 2016, 2014; Zhao et al., 2013), but is contradicting the claims of Thirumalai & Sinha (2011). The abnormal returns were significant when the recalls were classified by potential damage of the product. It is concluded that only class 1 and 2 recalls, which are the recalls with the highest potential damage, are followed by significant negative abnormal returns. This was in line with the findings of Ni et al. (2014). Prior research suggested that greater potential harm of the product would intensify investor’s loss aversion, which would be translated in the company’s valuation (Ni et al., 2014). However, the regression analysis did not reveal a significant relationship between the abnormal returns and the potential harm of the product. This means that there is another variable that likely causes class 1 and class 2 recalls to be significantly negative and class 3 recalls not. Deeper analysis revealed the importance of taking company size into account. The analysis showed that the negative causal effect of a recalling event on a company’s financial value can be (partially) offset by company size. Larger companies experienced less negative (or even positive) abnormal returns compared to smaller companies.
The second hypothesis was not supported in the analysis. Hence, it can be concluded that supply base complexity did not have a significant influence on the abnormal returns following a recalling event. This was contradictory to the propositions of Craighead et al. (2007) and Wowak & Boone (2015).
6.2. Theoretical contributions
Second, the role of potential harm of the product is repositioned by this research. In prior literature, researchers have claimed that potential harm was significantly enforcing the negative abnormal returns following a recalling event (Ni et al., 2014). By combining event study with regression analysis this thesis was able to offer a different interpretation of this mechanism. Class 1 and class 2 recalls were determined to have a significant negative effect on the abnormal returns, while class 3 recalls did not. An in-depth analysis revealed that the differences in size of the companies that experienced these types of recalls could explain the different outcomes. The regression analysis revealed that company size has a positive influence on the abnormal returns. Companies that experienced class 3 recalls were much larger than companies that experienced class 1 (77%) or class 2 (46%) recalls. Steven et al. (2014) claimed that firm size should be accounted for in research on recalling events since larger companies typically have a larger product portfolio. A larger product portfolio could have two implications on the abnormal returns. First, firms are able to spread the risk with larger product portfolio. Van Heerde et al. (2007) demonstrated that companies owning multiple brands can (partially) offset the negative abnormal returns following a product-crisis involving one of their brands by investing in their other brands. This indicates that for larger companies, with a larger product base, product-crises will have a less negative impact on their stock value, since it is compensated by the results of their other brands. Second, a larger product portfolio increases the complexity. As hypothesized in this paper and multiple others, this supply base complexity could potentially increase the costs associated with a recalling event (Craighead et al., 2007; Wowak & Boone, 2015). This research demonstrated that an increase in company size has a positive effect on the abnormal returns following a recalling event. Hence, this suggests that the advantages (spread of risk) outweigh the disadvantages (increased complexity). However, more in-depth research is required to confirm this.
research neglects behavioral complexity as a measure of supply base complexity. Even though structural and behavioral complexity are argued to be interrelated (Bode & Wagner, 2015), many researchers claim that there are more dimensions of supply base complexity that should be considered (Choi & Krause, 2006; Vachon & Klassen, 2002). Potentially, the behavioral dimension of complexity could have a larger influence compared to the structural dimension. Third, this research focused on a short timeframe. Even though the analysis focuses on the effect of supply base complexity on the abnormal returns until ten days after the event, it could be that the real influence of this variable becomes evident after this period. Especially the underlying mechanism of supplier responsiveness could possibly play a larger role when a longer timeframe is considered.
6.3. Managerial implications
This research adds support to the claim that recalling events are negatively affecting a firm’s financial value. Consequently, exposure to recalling events pose a large risk for the company’s existence. Companies should therefore take extra measures to prevent recalling events from happening in the first place. The results show that especially smaller companies experience higher financial risks. This is likely due to the fact that larger companies are able to spread the risk among a more extensive product base. Steven et al. (2014) claim that companies should not invest in proactive marketing efforts for their affected brand since this is often a waste of resources. Rather, firms should invest in their other brands in order to compensate the losses of the affected brand. This is in line with the claims of Chen et al. (2009) who advocate the advantage passive recall strategies over proactive strategies. The hypothesis development section highlighted the role of signals that are send out by the company. Managers should be very careful when releasing certain signals to the public.
dimensions of complexity that fall outside of the scope of this research but are relevant when responding to a recalling event.
6.4. Limitations and future research
It should be noted that this research has some limitations. First, this research only tests the effect of recalling events on public firms and private firms are excluded from the sample. The largest proportion of pharmaceutical firms in the United States are private and they do not deal with shareholders. Hence, signaling theory would be less relevant for these firms. Furthermore, this research focuses on the United States. It could well be that the response of the capital markets would be different in other geographic locations. As mentioned before, the operational definition of supply base complexity was focused on structural supply base complexity. This was based on the assumption that the number of nodes in the supply base (structural complexity) increases the degree of variety of interactions (behavioral complexity) (Bode & Wagner, 2015). Focusing more on behavioral supply base complexity could yield different results. Cultural aspects, as proposed by Choi & Krause (2006) are, for example, not considered in this research. Future studies could investigate whether and how the different dimensions of supply base complexity have an influence on the abnormal returns. Furthermore, the supply base in this research was established based on publicly available news announcements. This might not capture the full supply base. The data on supply base complexity also only included suppliers that were announced in a news announcement between 2008 and 2017. It could well be that there are more suppliers that are tight longer to the company. Furthermore, the total amount of different suppliers was taken in order to estimate the supply base at the time of the event. This might, however, not be an accurate estimation of the supply base at the time of the recalling event.
7. REFERENCES
Bala, R., Bhardwaj, P., & Chintagunta, P. K. (2015). Product Recalls, Category Effects and Competitor Response. Ssrn, (February 2019). https://doi.org/10.2139/ssrn.2546122 Ball, G. P., Shah, R., & Wowak, K. D. (2018). Product competition, managerial discretion,
and manufacturing recalls in the U.S. pharmaceutical industry. Journal of Operations
Management, 58–59(April), 59–72. https://doi.org/10.1016/j.jom.2018.04.003
Bergh, D. D., Connelly, B. L., Ketchen, D. J., & Shannon, L. M. (2014). Signalling theory and equilibrium in strategic management research: An assessment and a research agenda.
Journal of Management Studies, 51(8), 1334–1360. https://doi.org/10.1111/joms.12097
Bode, C., & Wagner, S. M. (2015). Structural drivers of upstream supply chain complexity and the frequency of supply chain disruptions. Journal of Operations Management, 36, 215–228. https://doi.org/10.1016/j.jom.2014.12.004
Boulding, W., & Kirmani, A. (2002). A Consumer-Side Experimental Examination of
Signaling Theory: Do Consumers Perceive Warranties as Signals of Quality? Journal of
Consumer Research, 20(1), 111–123. https://doi.org/10.1086/209337
Bozarth, C. C., Warsing, D. P., Flynn, B. B., & Flynn, E. J. (2009). The impact of supply chain complexity on manufacturing plant performance. Journal of Operations
Management, 27(1), 78–93. https://doi.org/10.1016/j.jom.2008.07.003
Brandon-Jones, E., Squire, B., & Van Rossenberg, Y. G. T. (2015). The impact of supply base complexity on disruptions and performance: The moderating effects of slack and
visibility. International Journal of Production Research, 53(22), 6903–6918. https://doi.org/10.1080/00207543.2014.986296
Bruno, J. (2018). Kannapolis-based drug maker filed for Chapter 11 bankruptcy | WSOC-TV. Retrieved February 10, 2019, from
https://www.wsoctv.com/news/local/drug-manufacturingplant-abruptly-shuts-down-weeks-after-nasal-spray-recall/829185176 Chao, G. H., Iravani, S. M. R., & Savaskan, R. C. (2009). Quality Improvement Incentives
and Product Recall Cost Sharing Contracts. Management Science, 55(7), 1122–1138. https://doi.org/10.1287/mnsc.1090.1008
Chen, Y., Ganesan, S., & Liu, Y. (2009). Does a Firm’s Product-Recall Strategy Affect Its Financial Value? An Examination of Strategic Alternatives during Product-Harm Crises.
Journal of Marketing, 73(6), 214–226. https://doi.org/10.1509/jmkg.73.6.214
Choi, T. Y., Dooley, K. J., & Rungtusanatham, M. (2001). Supply networks and complex adaptive systems: Control versus emergence. Journal of Operations Management, 19(3), 351–366. https://doi.org/10.1016/S0272-6963(00)00068-1
transaction costs, risks, responsiveness, and innovation. Journal of Operations
Management, 24(5), 637–652. https://doi.org/10.1016/j.jom.2005.07.002
Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011). Signaling Theory: A Review and Assessment. Journal of Management, 37(1), 39–67.
https://doi.org/10.1177/0149206310388419
Craighead, C. W., Blackhurst, J., Rungtusanatham, M. J., & Handfield, R. B. (2007). The severity of supply chain disruptions: Design characteristics and mitigation capabilities.
Decision Sciences, 38(1), 131–156. https://doi.org/10.1111/j.1540-5915.2007.00151.x
Crippa, L., Caridi, M., Tumino, A., Sianesi, A., & Perego, A. (2009). Do virtuality and complexity affect supply chain visibility? International Journal of Production
Economics, 127(2), 372–383. https://doi.org/10.1016/j.ijpe.2009.08.016
Desai, P., & Patel, D. N. (2014). Identifying Backfire of Communication on Perceived Hazard Level during Product Recall. Int. J. Retail. Rural Bus., 3(2), 969–977.
Folkes, V. S. (1984). Consumer Reactions to Product Failure: An Attributional Approach.
Journal of Consumer Research, 10(4), 398–409. https://doi.org/10.1086/208978
Hartman, R. S. (2006). Product Quality and Market Efficiency: The Effect of Product Recalls on Resale Prices and Firm Valuation. The Review of Economics and Statistics, 69(2), 367. https://doi.org/10.2307/1927247
Heerde, H. Van, Helsen, K., & Dekimpe, M. G. (2007). The Impact of a Product-Harm Crisis on Marketing Effectiveness. Marketing Science, 26(2), 230–245.
https://doi.org/10.1287/mksc.l060.0227
Hendricks, K. B., & Singhal, V. R. (2003). The effect of supply chain glitches on shareholder wealth. Journal of Operations Management, 21(5), 501–522.
https://doi.org/10.1016/j.jom.2003.02.003
Hendricks, K. B., & Singhal, V. R. (2005). An Empirical Analysis of the Effect of Supply Chain Disruptions on Long-Run Stock Price Performance and Equity Risk of the Firm.
Production and Operations Management, 14(1), 35–52.
https://doi.org/10.1111/j.1937-5956.2005.tb00008.x
Hoffer, G. E., Pruitt, S. W., & Reilly, R. J. (1988). The Impact of Product Recalls on the Wealth of Sellers: A Reexamination. Journal of Political Economy, 96(3), 663–670. https://doi.org/10.1086/261556
Jayaraman, R., Alhammadi, F., & Simsekler, M. C. E. (2019). Managing Product Recalls in Healthcare Supply Chain. In IEEE International Conference on Industrial Engineering
and Engineering Management. https://doi.org/10.1109/IEEM.2018.8607403
Future Product Reliability and Future Accidents: Evidence from the Automobile Industry. Journal of Marketing, 77(2), 41–57. https://doi.org/10.1509/jm.11.0356 Kannan, V. R., & Tan, K. C. (2007). The impact of operational quality: A supply chain view.
Supply Chain Management, 12(1), 14–19. https://doi.org/10.1108/13598540710724356
Liu, A. X., Liu, Y., & Luo, T. (2016). What Drives a Firm’s Choice of Product Recall Remedy? The Impact of Remedy Cost, Product Hazard, and the CEO. Journal of
Marketing, 80(3), 79–95. https://doi.org/10.1509/jm.14.0382
Luo, Y. (2008). A Strategic Analysis of Product Recalls: The Role of Moral Degradation and Organizational Control. Management and Organization Review, 4(2), 183–196.
https://doi.org/10.1111/j
Mackinlay, A. C. (1997). Event Studies in Economics and Finance. Source: Journal of
Economic Literature (Vol. 35). Retrieved from
https://www-jstor-org.proxy-ub.rug.nl/stable/pdf/2729691.pdf?_=1464007850793&seq=1
Ni, J., Flynn, B. B., & Jacobs, F. R. (2016). The effect of a toy industry product recall announcement on shareholder wealth. International Journal of Production Research,
54(18), 5404–5415. https://doi.org/10.1080/00207543.2015.1106608
Ni, J., Flynn, B., & Jacobs, R. (2014). Impact of product recall announcements on retailers financial value. International Journal of Production Economics, 153, 309–322. https://doi.org/10.1016/j.ijpe.2014.03.014
Novemsky, N., & Kahneman, D. (2005). The Boundaries of Loss Aversion. Journal of
Marketing Research, 42(2), 119–128. Retrieved from
https://journals-sagepub-com.proxy-ub.rug.nl/doi/pdf/10.1509/jmkr.42.2.119.62292
Peck, H. (2005). Drivers of supply chain vulnerability: An integrated framework.
International Journal of Physical Distribution & Logistics Management, 35(4), 210–
232. https://doi.org/10.1108/09600030510599904
Product recall trend: up strongly in 2018 - Pharmaceutical Commerce. (2018). Retrieved March 3, 2019, from https://pharmaceuticalcommerce.com/manufacturing-and-packaging/product-recall-trend-up-strongly-in-2018/
Rubin, P. H., Murphy, R. D., & Jarrell, G. (1988). Risky products, risky stocks. Regulation,
12(1), 35–39.
Siegel, D., & McWilliams, A. (1997). Event Studies in Management Research : Theoretical and Empirical Issues. Academy of Management Journal, 40(3), 626–657. Retrieved from
Simon, H. A. (1961). The architecture of complexity. Proceedings of the American
Philosophical Society, 106(6), 467–482.
Smith, K. G., Grimm, C. M., Gannon, M. J., & Chen, M.-J. (1991). Organizational
Information Processing, Competitive Responses, and Performance in the U.S. Domestic Airline Industry. Academy of Management Journal, 34(1), 60–85.
https://doi.org/10.5465/256302
Steven, A. B., Dong, Y., & Corsi, T. (2014). Global sourcing and quality recalls: An
empirical study of outsourcing-supplier concentration-product recalls linkages. Journal
of Operations Management, 32(5), 241–253. https://doi.org/10.1016/j.jom.2014.04.003
Thirumalai, S., & Sinha, K. K. (2011). Product Recalls in the Medical Device Industry: An Empirical Exploration of the Sources and Financial Consequences. Management
Science, 57(2), 376–392. https://doi.org/10.1287/mnsc.1100.1267
Vachon, S., & Klassen, R. D. (2002). An exploratory investigation of the effects of supply chain complexity on delivery performance. IEEE Transactions on Engineering
Management, 49(3), 218–230. https://doi.org/10.1109/TEM.2002.803387
Wowak, K. D., & Boone, C. A. (2015). So Many Recalls, So Little Research: A Review of the Literature and Road map for Future Research. Journal of Supply Chain Management,
51(4), 54–72. https://doi.org/10.1111/jscm.12079
Wu, F., Yeniyurt, S., Kim, D., & Cavusgil, S. T. (2006). The impact of information
technology on supply chain capabilities and firm performance: A resource-based view.
Industrial Marketing Management, 35(4), 493–504.
https://doi.org/10.1016/j.indmarman.2005.05.003
Zhao, X., Li, Y., & Flynn, B. B. (2013). The financial impact of product recall
announcements in China. In International Journal of Production Economics (Vol. 142, pp. 115–123). https://doi.org/10.1016/j.ijpe.2012.10.018
Zsidisin, G. A., & Wagner, S. M. (2011). DO PERCEPTIONS BECOME REALITY? THE MODERATING ROLE OF SUPPLY CHAIN RESILIENCY ON DISRUPTION OCCURRENCE. Journal of Business Logistics, 31(2), 1–20.
8. APPENDIX
Appendix A: List of tables & figuresTable 4.1: Categories company size ... 16
Table 4.2: FDA categories ... 17
Table 4.3: Summary of variables ... 17
Table 5.1: Daily abnormal returns and test statistics ... 20
Table 5.2: Comparison event study models (day 0) ... 21
Table 5.3: Ranks Wilcoxon signed-rank test ... 22
Table 5.4: Descriptive statistics per recall class ... 22
Table 5.5: Descriptive statistics variables regression analysis ... 24
Table 5.6: Regression models ... 24
Table 5.7: Analysis of different timeframes ... 25
Table 5.8: Descriptive statistics Company Size ... 26
Figure 2.1: Linking supply chain performance to shareholder value (Hendricks & Singhal, 2003) ... 8
Appendix B: Search strings Factiva
