AN EXPLORATIVE STUDY INTO LEARNING FROM NEAR-MISSES TO BUILD SUPPLY CHAIN ROBUSTNESS

AN EXPLORATIVE STUDY INTO LEARNING FROM NEAR-MISSES TO BUILD SUPPLY CHAIN ROBUSTNESS

Master thesis, MSc, specialization Supply Chain Management University of Groningen, Faculty of Economics and Business

20

of February 2018

ESTHER N. VAN DER MATEN Student number: S2988275 Email: e.n.van.der.maten@student.rug.nl

Supervisor University of Groningen Dr. K. Scholten

Co-assessor University of Groningen X. Tong

ACKNOWLEDGEMENTS

Writing this master thesis would not have been possible with the support from others, they have enabled me to learn and grow. At first, I would like to thank Dr. Kirstin Scholten for her thorough feedback, valuable comments and opportunities to let me see situations differently.

Secondly, I would like to thank Boris Fenneman for acknowledging the importance of near-

misses, his valuable input, enthusiasm, giving me possibilities to visit the plant and discussing

this topic with employees. Thirdly, I would like to thank Lorena for her cooperation

throughout the period. Finally, I would like to thank my friends and family who helped me to

sharpen my thoughts and have given their countless support.

ABSTRACT

Purpose – This study aims to found which learning approaches are used to build supply chain robustness by studying near-misses. Despite the increased interest in supply chain robustness, a way to proactively managing changes via learning of near-misses is rare.

Design/methodology/approach– An exploratory qualitative study based on semi-structured interviews is conducted. Data are gathered from thirteen interviews at twelve organizations.

Findings – The findings suggest that most frequently exploitative and collective learning from near-misses enhance supply chain robustness. The learning approaches show eleven underlying mechanisms to build supply chain robustness.

Originality/value – This is one of the first papers to research supply chain near-misses and its influence on the relationship between learning and supply chain robustness.

Keywords – Near-miss, Exploitative Learning, Exploratory Learning, Individual Learning, Collective Learning, Supply Chain Robustness

Paper type – Case study

Word count – 8875

INTRODUCTION

Supply chains are vulnerable to unexpected events due to continuous and severe disruptions in the business environment (Jüttner & Maklan, 2011). A first step is identifying such unexpected events at an early stage to support the robustness of the supply chain (Brandon- Jones, Squire, Autry, & Petersen, 2014). Redesign strategies to resist or avoid unexpected events reduces the vulnerability of the supply chain in the future (Vlajic, Van Der Vorst, &

Haijema, 2012), hence increase robustness (Durach, Wieland, & Machuca, 2015). Managing disruptions effectively requires an organization to learn (Mitroff and Pearson, 1993 in Chowdhury & Quaddus, 2016). Although, it is well described how organizations learn, little is known how organizations learn in order to increase its robustness.

Currently, supply chain management literature has described supply chain robustness as the driver for enhancing business performance (Wieland & Wallenburg, 2012). This requires an organization to proactively manage disruptions during times of change (Brandon-Jones et al., 2014) and has been described as a ‘strategy to resist and avoid change’ (Durach et al., 2015:

123). Ways to increase supply chain robustness and prepare for future disruptions can be done by including redundancy in the supply chain (Azadegan, Patel, Zangoueinezhad, &

Linderman, 2013), for example using a multiple supplier strategy (Tang, 2006a). Another way to increase the capability of the organization to deal with future disruptions is by learning from unexpected events (Zsidisin & Wagner, 2010). Learning can be done by ‘exploration of new possibilities’ and ‘exploitation of old certainties’ (March, 1990: 71) and takes place on individual and collective level (Anderson Jr. & Lewis, 2014).

Even though it is known that explorative and exploitative learning enhances future

preparedness (Chowdhury & Quaddus, 2016), it is unknown whether these two types of

learning also enhance supply chain robustness. In order to understand how organizations learn to build robustness, the focus should not just be on the outcome of disruptions but also on the causes that leads to these disruptions (Soyer & Hogarth, 2015). A possible way to research the causes of disruptions, is by looking at so-called near-misses (Dillon, Tinsley, Madsen, &

Rogers, 2016). Near-misses characterize as ‘often unmarked small failures that permeate day- to-day business but cause no immediate harm’ (Tinsley, Dillon, & Madsen, 2011: 90). These will be helpful in understanding how organizations learn to build supply chain robustness, since they show potential causes which consequently lead to disruptions before they actually happen (Andriulo & Gnoni, 2014). Hence, near-misses do provide valuable insights in order to increase supply chain robustness (Tinsley et al., 2011). This results in the following research question: How do organizations learn from near-misses to increase supply chain robustness?

Exploring near-misses in high-risk/high-reliability organizations, like the chemical processing industry, is a great opportunity to investigate the impact of these near-misses, because this sector focuses on total elimination of disruptions and total elimination of trial-and-error learning (Weick, 1987 in Ellis & Davidi, 2005). By interviewing multiple organizations in the chemical processing industry a first step has been taken to analyze exploration and exploitation of learning in order to enhance supply chain robustness. Moreover, this paper extends the literature on responses to near-misses by focusing on learning approaches organizations used to resist and avoid unexpected events and by defining near-misses in supply chain management. On a managerial level this paper shows that near-misses are detectable and describes how the circumstances to learn form near-misses can be enhanced.

Furthermore, this paper acknowledges which ex ante measure contribute to a new state of the

robustness of the supply chain.

THEORETICAL BACKGROUND Supply Chain Robustness

Robustness is a broad term and has different descriptions depending on the field of research (Jen, 2005; Bundschuh et al., 2006; Qiang et al., 2009 in Vlajic et al., 2012). In supply chain management one of the first articles describes supply chain robustness as ‘a system’s ability to resist an accidental event and return to do its intended mission and retain the same stable situation as it had before the accidental event’ (Asbjørnslett & Rausand, 1999: 220). Many other researchers have used this definition to describe a robust supply chain as an unchanged structure in which robustness can only be threatened by disruptions (Vlajic et al., 2012).

Durach, Wieland and Machuca (2015) argue that the proposed definition addresses various characteristics and propose the following definition: ‘the ability of a supply chain to resist or avoid change’ (Durach, Wieland, & Machuca, 2015: 123). This definition differs from Asbjørnslett and Rausand (1990) because a robust supply chain should be able to cope with change to construct an even more stable situation. Also, the definition by Durach et al. (2015) describes changes rather than accidents. So, the definition of Durach et al. (2015) addresses a wider variety of possible impacts on the supply chain and focuses not just on ways to resist change but also on the ability of the supply chain to have back-up solutions when changes occur. This definition will be used throughout this paper.

It is generally accepted that supply chain robustness needs a proactive strategy (e.g. Wieland

& Wallenburg, 2012) and involves ex ante measures to manage change. By doing so the supply chain does not require modifications during times of change (Durach et al., 2015).

These measures are ‘cause-related measurements that strive for lowering the probability of

risk occurrence’ (Thun & Hoenig, 2011: 245). Measures that resist and/or avoid change are

shown in table 1.

TABLE 1: EXAMPLES OF EX ANTE MEASURES TO INCREASE SUPPLY CHAIN ROBUSTNESS Supply

chain robustness

Brandon- Jones et al.

(2014)

Durach et al. (2015)

Kleindorfer and Saad (2005)

Tang (2006a) Thun and Hoenig (2011)

Chowdhury and

Quaddus (2016)

Vlajic et al. (2012)

Wieland (2016)

Wieland and

Wallenburg (2012)

Resistance Buffers,

Human capital, Strategic storing inventory

Multi-supplier strategy, Responsive pricing strategy, Postponement, VMI and CPFR

Flexibility Financial strength, Market capability, Collaboration

Do not supply materials from specific regions, Modular product design

Multiple sources of supply Product design Logistical network design Resistance

and Avoidance

Leadership commitment, Information sharing, Risks management orientation, Dense supply chain

Coordination, cooperation and

collaboration among supply chain partners

Managed system, Managing system, Information system

Avoidance Visibility of demand information

Sharing risk- related information

Certified suppliers, Supplier development, Production sites in safe areas and tracking and tracing

Visibility Inventory

In general, the approaches in table 1 enhance the ability of an organization to continue operating during time of change (Tang, 2006a). Sometimes multiple supply chain members are involved and a combination of ex ante measures is required to both resist and avoid change. This combination of ex-ante measures is often focused on changes on the supplier side. Wieland & Wallenburg (2012) explain that changes on the supplier side are most of the times easier to predict and thereby help to proactively resist and avoid change. However, a distinction between resistance and avoidance is necessary to fully comprehend the supply chain robustness concept.

Resistance

Resistance itself is described as ‘the ability of a supply chain to withstand change’ (Durach et al., 2015: 123). Measures relating to resistance concentrate on ways to secure the availability of products or ways how products can be used in multiple processes. For example, having multiple suppliers for the same product helps to resist disruptions (Vlajic et al., 2012) as it allows an organization to continue producing when a change occurs at another supplier. The above-mentioned approaches might increase costs. They also provide a way to easily use different sources during time of change (Brandon-Jones et al., 2014).

Avoidance

Avoidance is ‘the ability of a supply chain not to be affected by change’ (Durach et al., 2015:

123). Measures relating to avoidances focus on collaboration with suppliers and on the presences of excess resources (Marley, Ward, & Hill, 2014). This way a buffer of e.g.

inventory is available in case of a disruption. Chopra and Sodhi (2004) show that a balance of

inventory, capacity and other requirements throughout the supply chain is necessary. Sheffi

(2015) takes this one step further and suggests that gaining an understanding of the supply

chain, the impact on others and the visibility of supply chain members enhances the ability to avoid changes.

Although, resistance and avoidance are dealt with separately both need to be in place in order to build robustness (Kamalahmadi & Parast, 2016). The ex ante measures of resistance and avoidance help to overcome come unexpected events (Chowdhury & Quaddus, 2016). To further improve supply chain robustness it is necessary to learn from unexpected events and adjust business processes accordingly (Madsen, 2009).

Learning

Extensive literature about learning is available, but this paper focuses on literature regarding learning from changes and finding new ways to deal with unexpected events (Chowdhury &

Quaddus, 2016) to enhance supply chain robustness. In order to deal with unexpected events an organization needs to recapture previous events, to discard the unstable situation, to expect the unexpected and to enhance the ability of the organization to learn from disruptions (Berkes, 2007). Expecting the unexpected is an oxymoron and stands for ‘having the tools and the codes of conduct to fall back on when an unexpected event happens (Hewitt 2004)’

(Berkes, 2007: 289). Learning from unexpected events improves the capability of the supply chain to deal with similar events in the future (Zsidisin & Wagner, 2010). Learning from previous similar events in order to prevent these events from happening (Madsen, 2009) can be done in two ways ‘exploitation of old certainties’ and ‘exploration of new possibilities’

(March, 1990: 71). Besides, this learning can also be done on two different levels: individual

and collective (Argote & Miron-Spektor, 2011).

Ways of learning

Exploitation is explained as continuously ‘testing and revising of the arrangement and environmental knowledge of organizations’ (Folke et al., 2002 in Berkes, 2007). By exploitative learning an organization establishes reliability of its experience (Holmqvist, 2004). An example of an organization that uses exploitative learning is Dell: ‘Dell focuses on its time-to-market production, efficient logistics, enhanced service operations, and an innovative distribution model (Parise and Henderson 2001; Kraemer and Dedrick 2002).

Although Dell has an R&D intensity of only 1.6%; it relies heavily on overseas design houses and its Tier 1 suppliers with larger R&D expenditure in design and manufacture of new products (Ricadela 2002)’ (Azadegan, Dooley, Carter, & Carter, 2008: 22). This example shows that Dell is successful by organizing and operating their business based on exploitative learning.

Exploration is described as learning based on probabilities (March, 1990). By exploratory

learning an organization establishes variety in its experience (Holmqvist, 2004). In contrast to

Dell, an example of an organization that uses exploratory learning is Sun Microsystems: ‘Sun

designs its own microprocessors and internally develops (instead of purchase) major

components for its computer systems (Parise and Henderson 2001). Sun’s R&D supports an

innovation-focused product strategy. Compared with its competitors, Sun shows high

preference towards internal search and risk taking in developing new products. The company

develops a wide range of technologies’ (Azadegan et al., 2008: 22). This example shows that

by exploratory learning the organization has developed a strategy to deal with change and not

being affected by change.

In sum, exploitation involves efficiency, refinement and implementation whereas exploration can be seen as search and risk-taking (March, 1990). If an organization focuses too much on exploitative learning it might stagnate whereas an organization that focuses too much on exploratory learning might be unable to extract valuable information from its current way of operating (Baum & Dahlin, 2007). However, both types of learning are required to enhance supply chain robustness.

Levels of learning

Individual learning is described as: ‘how people process the information that enables them to interpret changes in situations, assess the consequences of their own and others’ actions in situations, and uses this understanding to refine or radically refine their subjective theories or mental model of how the world, or that part of their world that immediately concerns them, operates’ (Hayes & Allinson, 1998: 851). Collective learning is ‘a group or any assemblage of individuals whose collaboration produces knowledge embedded at the level of the collective’

(Anderson Jr. & Lewis, 2014: 358). Collective learning is enabled by the learning of its individuals which directly or indirectly affects the collective (Kim, 1993). In order to establish collective learning, the gained knowledge of the individual needs to be ‘embedded in a supra-individual repository so that others can access it’ (Argote & Miron-Spektor, 2011:

1126). An often-used example to illustrate the effects of individual learning and collective

learning is given by de Geus (1997): ‘In the early 20th century milk bottles in Britain had no

top on and birds quickly learnt to siphon off the cream from the top of the milk. Robins and

blue tits were particularly adept at this. Then, aluminum seals were placed on milk bottles. By

the early 1950's the entire blue tit population of the UK, about a million birds, had learned

how to pierce the aluminum seals. Conversely, the robins, as a family, never regained access

to the cream. Occasionally, an individual robin learns how to pierce the seals of the milk bottle, but the knowledge does not pass to the rest of the species.

In short, the blue tits went through an extraordinarily successful institutional learning process.

The robins failed, even though individual robins had been as innovative as individual blue tits.

Moreover, the difference could not be attributed to their ability to communicate. As songbirds, both the blue tits and the robins had the same range of communication methods:

behavior, movements and song. The explanation could be found only in the social propagation process - the way blue tits spread their skill from one individual to members of the species as a whole.’ This example shows the different outcomes of individual and collective learning and hence, suggests distinct ways to build supply chain robustness.

Even though, learning can happen in different ways and on different levels, in order to understand how organizations learn to build robustness the focus should not just be on the outcome of disruptions but also on the causes that leads to these disruptions (Soyer &

Hogarth, 2015). A possible way to research just this, the causes of disruptions, is by looking at so-called near-misses (Dillon et al., 2016). Near-misses are often studied in other sectors (e.g. automotive, healthcare and engineering) than in a supply chain context. In these sectors valuable information was distracted from near-misses to increase performance (Barach &

Small, 2000).

Context

Near-misses will be helpful in understanding how organizations learn to build supply chain

robustness. They show potential causes and provide valuable insights (Tinsley et al., 2011) of

disruptions before these disruptions actually happen (Andriulo & Gnoni, 2014). However,

near-misses are currently not defined in supply chain management. Other fields of research have developed several definitions of near-misses, which are presented in table 2 below. The different definitions help to extract a definition of a supply chain near-miss.

TABLE 2: DEFINING NEAR-MISSES

Definition Field of

research Source A hazardous situation where the event sequence could

lead to an accident if it had not been interrupted by a planned intervention or by a random event

Automotive Andriulo & Gnoni (2014)

An event, a sequence of events, or an observation of unusual occurrences that possesses the potential of improving a system's operability by reducing the risk of upsets some of which could eventually cause serious damage.

Banking Muermann &

Oktem (2002)

An event or situation that, if a small set of behaviors or conditions had been slightly different, would have led to a consequential adverse event.

Engineering Corcoran (2004)

Qualitative descriptions of events that could have harmed patients but did not due to a timely intervention or a convenient evolution of the circumstances.

Healthcare Cure, Zayas-Castro

& Fabri (2011) Successful outcomes in which chance plays a critical

role in averting failure

Management Dillon & Tinsley (2008)

Close calls, had it not been for chance, would have been worse

Management Tinsley, Dillon &

Madsen (2011) When an event could have happened (for example,

because of hazardous conditions) but did not.

Medicine, Gambling

Dillon & Tinsley (2008)

Having little if any immediate impact on individuals,

processes, or the environment. Process

industry Phimister, Oktem, Kleindorfer &

Kunreuther (2003) Where an accident could have happened had there been

no timely and effective recovery

Process industry

Van Der Schaaf &

Kanse (2004)

All definitions presented have four elements in common. They all describe an unexpected

situation with in potential an unwanted, negative outcome. Thereby they mention a planned

intervention or a random event to prevent the unwanted outcome. Relating this to the supply

chain context leads to the following definition ‘a potentially problematic situation that could

have lead to severe performance impact on the supply chain if it had not been interrupted by a random event or planned intervention.’

By interpreting near-misses as a warning signal, they show a potential to learn from them (Dillon et al., 2016) and consequently they are an opportunity to build supply chain robustness. Yet, there are some reasons that make it difficult to learn from near-misses: (1) outcome bias (Dillon et al., 2016), (2) normalizations of deviance (Tinsley et al., 2011) and (3) hindsight bias (Gephart, 1993, Fischhoff, 1975, Fischhoff & Beyth, 1975 in Dillon &

Tinsley, 2008). In case of near-misses, decision makers are apt to value near-misses with no specific impact, therefore, they tend to not value the outcome; outcome bias (Dillon et al., 2016). Secondly, when making judgments about risky situations the tendency exists to accept abnormalities as normality; normalization of deviance (Tinsley et al., 2011). Thirdly, decisions which are based on information that seemed relevant at the time of the event are overvalued; hindsight bias (Gephart 1993, Fischhoff 1975, Fischhoff and Beyth 1975 in Dillon & Tinsley, 2008). The reasons mentioned make it more difficult to learn from past events with no negative outcome (Dillon & Tinsley, 2008).

Moreover, having a positive safety climate, enhancing alertness and awareness of organizational members increases the recognition of near-miss events (Dillon et al., 2016).

The recognition of events that almost happened prompts critical evaluation and learning

(Markman and Tetlock 2000, Kray et al. 2006 in Dillon & Tinsley, 2008). Thus, the near-

misses will be helpful in understanding how organizations learn to build supply chain

robustness.

In conclusion, individuals and collectives have the ability to exploitive and exploratory learn from near-misses by evaluating previous events. The outcome of this learning process has the potential to enhance the robustness of the supply chain because it exposes the measures need to be taken to proactively cope with near-misses. This relationship can be seen in figure 1: the conceptual model.

FIGURE 1: CONCEPTUAL MODEL

METHODOLOGY Research Design

As little was known about the role of near-misses in supply chain robustness and existing theory did not offer many answers to what concepts of learning can be applied to do so, this paper used an explorative qualitative study (Yin, 2003). Qualitative research was particularly suitable for explaining a phenomenon that is not well understood (Ellram, 1996), in this case learning from near-miss events. By doing so, qualitative research provided in-depth and rich information of that phenomenon (Ellram, 1996). In order to explore near-misses in supply chain management, different organizations were interviewed and hence supply chain organizations are the unit of analysis.

Near-Miss Learning

• Way of Learning o Exploitation o Exploration

• Level of Learning o Individual o Collective

Supply Chain Robustness

• Resistance

• Avoidance

Research Setting

The setting of this research was the chemical processing industry in Germany and The Netherlands. This sector was especially interesting, because it is known to use approaches to identify near-miss events (Gnoni, Andriulo, Maggio, & Nardone, 2013) whereas other sectors do not. Moreover, this sector focused on total elimination of disruptions and total elimination of trial-and-error learning (Weick, 1987 in Ellis & Davidi, 2005), because disruptions had severe consequences in this sector. The chemical processing industry was broadly described as industries that either processes or manufacture chemicals (Ebnesajjad, 2016). The industry had been specified into six different categories: (1) petrochemicals, (2) polymers, (3) basic inorganics, (4) specialty chemicals, (5) consumer chemicals and (6) bio based chemicals and polymers (Ebnesajjad, 2016).

Data Collection

Data was gathered through semi-structured interviews between November and December 2017. Before the interview took place the interviewee was send an introductory letter to be informed about the topic of this paper and the procedure of the interview. Next to this, the interview questions were sent before the interview took place and it was asked to communicate the closed questions before the interview (appendix 1 shows the interview protocol). The first part of the closed questions focused on the risk culture of the organization of the interviewee and the second and final part of the closed questions was about two near- miss examples. This way the interviewee could be prepared before the interview took place.

Such a protocol increases the reliability of the study (Yin, 2003). The open interview

questions focused on a detailed description of two near-miss events and how the organization

of the interviewee responded to them. By asking the same open questions to each interviewee

the ability to compare interviews and reliability of data was improved (Voss, Tsikriktsis, &

Frohlich, 2002). Moreover, a pilot interview tested if the interview questions asked were in accordance with the expected answers. This increases the reliability of the study (Ellram, 1996). And an expert in the field of supply chain resilience checked the interview questions beforehand. The interviews were done by two researchers, which ensured that all questions were answered and comprehended. This way the reliability of findings was increased (Eisenhardt, 1989).

The interviews took place in a face-to-face manner and one interview took place via telephone. During the interview the interviewee had the possibility to discuss questions in more depth. Also, the interviewee was asked to sign a consent clarification (see appendix 1) and had the possibility to leave the interview at any moment. The interview was voice recorded and transcribed afterwards. Moreover, the transcription of the interview was send to the interviewee to verify and was adjusted if necessary. This improves construct validity (Ellram, 1996). An interview database had been developed and contained all recorded and transcribed interviews, such an approach enhanced reliability (Ellram, 1996). The goal of the interviews was to discover how organizations overcome near-misses and ultimately learned from these situations. Next to this, out of these twelve organizations the opportunity was given to analyze one organization more close by being present at this organization for a day.

It provided extra information on how an organization coped with near-misses.

Interviewee selection

The interviewees were asked to participate because their job dealt with disruptions, had some

experience in reflecting on disruptions (hence experience in the field) and operated in a

supply chain related job (e.g. purchasing, planner). They were approached via personal

business contacts, LinkedIn, e-mail or telephone. The following table, table 3, depicted the

position of the interviewee, the length of the interview, in which chemical category the organization of the interviewee operated in, the amount of employees at the organization of the interviewee and the frequency of near-misses occurring. The different characteristics of the organization of the interviewee (sector, # of employees and frequency of near-misses) were valuable for the analysis and showed why interviewees showed different ways of learning to build robustness.

TABLE 3: OVERVIEW OF SELECTED INTERVIEWEES

O rgan iz ati on In te rvi ew P os iti on of in te rvi ew ee Le n gth of in te rvi ew C h emi cal cate gor y of in te rvi ew ee ’s or gan iz ati on N u mb er of emp loye es at in te rvi ew ee ’s or gan iz ati on F re q u en cy of a n ear -mi ss at in te rvi ew ee ’s or gan iz ati on 1 1 Project & account

manager 35 minutes Consumer chemicals > 20.000 Weekly 2 2 Global

procurement director

52 minutes

Machinery and software for chemicals

> 20.000 Daily

3 3 Commodity

manager 80 minutes Machinery for chemicals

75.000 Monthly 4 4 Sales & operations

planner 33 minutes Consumer chemicals 1.314 Weekly 5 5 Supply chain

manager Northern Europe

59 minutes

Machinery for chemicals

167 Once every five years 6 6 Procurement

director 69 minutes Machinery for chemicals

230 Monthly

7 7 Global strategic supply chain manager

73 minutes Machinery and packaging for chemicals

20.000 Monthly

8 Project leader 95 minutes 8 9 Procurement

manager 55 minutes Polymers 1.400 Weekly

9 10 Senior supply

chain planner 62 minutes Petrochemicals 1.500 Weekly 10 11 Production

coordinator 56 minutes Specialty chemicals 8.300 Yearly 11 12 Site director 56 minutes Petrochemicals > 10.000 Yearly 12 13 Senior analyst

supply chain solutions

60 minutes Petrochemicals 40.000 Monthly

Manager supply

chain solutions

Data Analysis

The data was analyzed using coding based on the approach taken by Miles and Huberman

which had been elaborated in Miles, Huberman and Saldana (2014). This approach is

threefold: (1) data reduction, (2) data display and (3) drawing and verifying conclusions. Data

reduction had been done by reading the interviews multiple times and afterwards segmenting

and coding the interview data. As a result some near-miss examples had been excluded from

the analysis, because interviewees did not describe near-misses, but disruptions instead. This

was the case in the second near-miss example of interview: 2, 5, 6, 7 and 9. Quotes that were

found relevant were transferred to an Excel file and summarized. Afterwards, they were

focused on either resistance or avoidance and categories were identified. Next, a link to a way

and a level of learning had been made. Some interviews did not show learning from near-

misses, which were labeled ‘no learning’. An example of the coding scheme had been shown

in table 4 and a part of the coding scheme had been presented in appendix 2. The process had

been shown in figure 2. Moreover, data display had been taken place, of which an aggregate

framework can be seen in the findings: table 5, 6 and 7. This helped to compare interviews in

order to find similarities and differences between the interviewees. These are also shown in

the finding section. At last, drawing and verifying conclusions were shown in the discussion,

which contained propositions based on the findings and its interpretation.

FIGURE 2: DATA ANALYSIS PROCESS

TABLE 4: EXAMPLE OF CODING TREE

Case

Focus

Quote Data

reduction Category

Link to learning

R es is tan ce A voi d an ce Ex p loi tati on Ex p lor ati on In d ivi d u al C ol le cti ve N on e 11 ✓ ‘For me it is the

business continuity plan. It is never good to have a single source.’

No single source

Multiple supplier strategy

✓ ✓

8 ✓ ‘Than we decided

to work on some other projects based on internal

priorities and now we wished to continue with this one and then it turned out the lead time increased in the last six months and more than doubled.’

Project

management Coordination ✓ ✓ Data reduction

Descriptive codes

Link to learning:

Exploitation or Exploration, Individual or Collective or

None Focus on supply chain

robustness:

Resistance,

Avoidance

FINDINGS

This paper aimed to find out which ways of learning (i.e. exploratory and exploitative) and which levels of learning (i.e. individual and collective) were used, in order to learn from near- misses to build supply chain robustness. Eleven underlying mechanisms were identified from the interview data: collaboration, coordination, financial management, information management, inventory management, modularity, multiple supplier strategy, human resource management, risk management, supplier development and training. The findings provided support for a relationship between learning from near-misses and both aspects of supply chain robustness: resistance and avoidance. Noteworthy, some organizations recognized near- misses, but were unable to learn from these events to increase supply chain robustness.

Resistance

It was found that to build resistance, organizations applied exploitative and exploratory learning from near-misses and learning from near-misses on individual and collective level.

Organizations learned that collaboration, coordination, financial management, information

management, inventory management, multiple supplier strategy, human resource

management, risk management and/or supplier development were effective measures to

enhance resistance. Modularity and training were not seen in this context. An overview of

these results is shown in table 5 on the following page.

TABLE 5: SUMMARY FINDINGS RESISTANCE

Level of learning Way of Learning

Exploitation Exploration

Individual Collaboration (7, 10) Collaboration (9) Financial management (10)

Multiple supplier strategy (11) Risk management (6,10)

Collective Collaboration (3, 5, 11, 12) Coordination (1, 5, 9)

Coordination (3, 4, 7) Information management (7) Financial management (3) Inventory management (1, 2, 3) Information management (3, 7, 12) Human resource management (9) Inventory management (1, 3, 4, 6,

8, 9)

Risk management (7, 11) Multiple supplier strategy (7, 10)

Risk management (3, 7, 11, 12) Supplier development (12)

Ways of learning

Eleven out of twelve cases used exploitative learning from near-misses to increase resistance.

In particular, it was found that six organizations learned that collaboration, inventory management and/or risk management build resistance. In order to do so, organizations learned that proactive communication upstream, internal and downstream was required (collaboration) and managing stock (inventory management) helped to increase resistance.

For example, case 5 described collaboration as ‘that is what [communication about the state

of events] we did upfront, what we did continuously.’ This resulted in in developing plans in

consultation with supply chain members to be able to respond proactively to unexpected

events. Regarding risk management, it was learned that several methods enhance resistance,

e.g. scenario planning (case 3), not using the just-in-time principle (case 6), recognizing

warning signals (case 7), outsourcing (case 10), root cause analysis (case 11) and scenario

planning and using the risk matrix (case 12). Case 11 described risk management as ‘most of

the times, the issue is a disturbance in an operational line. Of course, root cause analysis is

done to prevent disturbances for the future.’ Moreover, in four of the cases it was learned that

coordination, information management and/or multiple supplier strategy build resistance. The

organizations learned that coordination of preventive plans enhance resistance. Also, organizations learned that information management by storing knowledge helped to be better prepare for future events. Case 12 explained this as ‘we make a sort of ‘draaiboek’ (plan of approach) if that [near-miss] happens: what do we need, whom do we need and how will we approach it, because you can be prepared for a lot of these things beforehand.’ By exploitative learning organizations found that a multiple supplier strategy helped to build resistance. In only a few cases, organizations learned that financial management or supplier development enhance resistance. The organizations used exploitative learning to manage the stock of the supplier (financial management) and to adept the selection process of the logistics service provider (supplier development) to build resistance. At last, one case did not use exploitative learning at all, but instead used exploratory learning.

Regarding, the effect of exploratory learning from near-misses to enhance resistance, the

findings were less pronounced. Seven out of twelve cases used exploratory learning to build

resistance. Especially, three cases learned that coordination and/or inventory management

were measures to enhance resistance. In doing so, coordination of the production process and

realignment of stock (inventory management) were learning outcomes that increase

resistance. Case 2 gave an example of inventory management as in which stock locations can

be predetermined to continue operating during a near-miss: ‘although it [near-miss] led to a

discussion whether to use or continue to use country A as a hub of that inventory’. In only a

few cases organizations learned that risk management, collaboration, information

management and/or human resource management helped to build resistance. These learning

outcomes, regarding building resistance, entailed: identifying consequences of near-misses

(risk management), proactive communication between supplier and focal organization

(collaboration), storing knowledge of certification (information management) and hiring a new employee to overview the stock of (toxic) raw materials (human resource management).

To conclude, both ways of learning were used to enhance resistance. Although, there appeared to be a slight preference for exploitative learning from near-misses.

Levels of learning

Five out of twelve cases used individual learning from near-misses to increase resistance.

Especially, in three cases it was learned that collaboration, as in the communication with the supplier or internally, enhances resistance. Two cases used individual learning and found that risk management helped to build resistance. For example, case 10 explained risk management as: ‘the only option would be a third silo from our backup supplier, maybe it is cheaper to take the chance to not produce for two days.’ It was learned which options need to be taken to create future preparedness and thus enhances resistance. Moreover, in one interview it was found that either financial management or a multiple supplier strategy was required to build resistance. The case that learned that a multiple supplier strategy was required also learned that a single supplier is not sufficient to resist near-misses in the future. Or as mentioned in case 11: ‘for me it is the business continuity plan. It is not desirable to have a single source.’

Much more remarkable were the findings about collective learning from near-misses to

enhance resistance, as all twelve cases used collective learning to increase resistance. In

particular, seven cases learned that in order to build resistance inventory management of

safety stock and buffers were an effective measure. In six cases collective learning was

applied to build resistance by coordination of preventive checks and plans. For example, case

7 described coordination as ‘we have a decent plan, of course sometimes machinery or people

move to other areas, but we have those plans available to make sure we get back to our

business in a preferable timeframe. We make sure that we do not have a disruption in

business, which affects our customers negatively.’ Furthermore, risk management and/or

collaboration came forward in four cases as a learning outcome to enhance resistance. Risk

management was seen by means of scenario planning (case 3), recognizing warning signals

(case 7), root cause analysis (case 11) and scenario planning and using the risk matrix (case

12). Collaboration entailed upstream, internal and downstream communication to build

resistance. Three cases learned that in order to enhance resistance, information management

was required. This involved having preventive information of internal and downstream supply

chain members available and to reassessing this information to be better prepared for future

near-misses. Only a couple of cases learned that multiple supplier strategy, financial

management, supplier development and/or human resource management helped to build

resistance. Financial management was done every three months and supplier development

was used in adapting the selection process of the logistics service provider. Case 12 described

this as ‘proactive: not only wait until it [near-miss] happens, but also try upfront

[communicate requirements]. And in our selection process of LSP (logistic service provider)

that is very important and of course there is a focus on costs, yes there is, but it is the

combination of safety and performance, quality and costs.’ Moreover, it was learned that in

order to build resistance, human resource management required hiring a new employee in

order to manage raw material stock. This was described in case 9 as ‘the impact on the supply

chain was when we assigned a new planner, currently it takes half a work week of time to

plan all the raw materials for all the production lines.’

To conclude, both levels of learning (individual and collective) were used to enhance resistance. Despite the impact of both levels on resistance, learning on a collective level occurred more frequently than learning on an individual level.

Avoidance

It was found that exploitative and exploratory learning from near-misses and individual and collective learning from near-misses increase avoidance. Organizations learned that collaboration, coordination, financial management, information management, inventory management, modularity, multiple supplier strategy, modularity, risk management and/or training enhance avoidance, while supplier development was not seen in any of the cases. The findings are summarized in table 6 below.

TABLE 6: SUMMARY FINDINGS AVOIDANCE

Level of learning Way of Learning

Exploitation Exploration

Individual Collaboration (10) Collaboration (4)

Coordination (7) Coordination (5, 7, 9, 10)

Information management (6, 8) Inventory management (9) Modularity (9)

Risk management (3, 5)

Collective Collaboration (1, 7, 8, 11, 12) Collaboration (3, 7) Coordination (1, 2, 3, 4, 6, 7, 8, 11,

12)

Coordination (1, 3, 8, 9, 12) Financial management (3, 7, 11, 12) Financial management (3) Information management (1, 7, 8, 11,

12)

Information management (2, 3, 12)

Inventory management (1, 2, 4, 6, 9,

11, 12) Inventory management (1, 7)

Multiple supplier strategy (6, 12) Multiple supplier strategy (12)

Risk management (3, 7, 8, 9, 11, 12) Risk management (2, 5)

Training (4)

Ways of learning

All cases used exploitative learning in order to increase avoidance. In particular, it was found that all twelve cases learned that coordination was an effective measure to build avoidance.

Coordination entailed extra manual handling (case 1, 2, 3, 6, 7, 8, 11), rearranging transport (case 4, 5, 10) and alignment of production schedule (9, 11, 12). In seven cases it was learned that inventory management and/or risk management build avoidance. Inventory management consisted of the availability of stock (case 1, 4), securing supply (case 1, 2, 4, 6) and adjusting the production planning (case 9, 11, 12). Risk management was seen in multiple ways, for example checks on supplier (case 3, 8, 12), communication about circumstance towards the customer (case 5, 12), preventive actions (case 7, 12), product checks (case 9, 11) and a crisis- team (case 12). Half of the cases learned that collaboration by proactively communicating upstream, internal (specifically about product analysis) and downstream helped to build avoidance. Case 1 described this as: ‘we had to cooperate [with upstream and internal supply chain members] to do an analysis on the products.’ In five cases information management of safety, product and customer related information came forward as a learning outcome to enhance avoidance. Case 8 described the following example: ‘the biggest change is internally:

to inform the product engineers even better about the changing market circumstances, because

a lot of them work in our company for a long time. They have a lot of historical knowledge,

but they still base their decisions on that historical knowledge. So, in their minds an average

time of delivery is eight weeks, but at this moment the average time of delivery is sixteen

weeks.’ In only a few cases it was learned that financial management, modularity, multiple

supplier strategy and/or training helped to build avoidance. Financial management of the

supplier and the organization and training involved gaining a better understanding of their

Enterprise Resource Planning system, both appeared to build avoidance. Case 9 described the

effect of modularity as: ‘not only the customer uses this product, but we use this product as a

raw material for other products as well. We would not been able to produce our other products [in case of the near-miss]. So, the effect would have been much, much bigger.’

Less explicit were the findings regarding exploratory learning: nine out of twelve cases used exploratory learning from near-misses in order to increase avoidance. Especially, half of the cases learned that coordination by extra manual handing (case 1, 3, 9, 12), delivery (case 7) and project management (case 8) helped to enhance avoidance. In three of the cases it was learned that information management and/or collaboration played a role to build avoidance.

Information management consisted of gaining knowledge of internal and upstream parties to build avoidance. Collaboration was seen by internal and downstream communication. Only a couple of cases learned that inventory management, multiple supplier strategy and/or risk management enhance avoidance. Inventory management focused on the stock levels and risk management consisted of being aware of risks and finding which mitigation actions were required to further develop avoidance. This is described in case 5 as: ‘If that [previous near- miss] would not have happened, we would not have changed our view on risks.’ An example of exploratory learning to enhance avoidance of multiple supplier strategy is given in case 12:

‘we have multiple things [supply sources] to handle from a so to say near-miss perspective.

So, that is a balancing act.’

In conclusion, both ways of learning appeared to enhance avoidance. Even though there was a preference to use exploitative learning.

Levels of learning

Eight out of twelve cases used individual learning to increase avoidance. In particular, it was

found that four cases learned that coordination of production planning enhanced avoidance.

Only a few cases learned that collaboration, information management, modularity and/or risk management increased avoidance. Collaboration consisted of internal and downstream communication. Case 10 described this as ‘I think our customers appreciate it, when we proactively contact them.’ Information management to enhance avoidance was described in case 8 ‘we only have an order confirmation and an alarm system when it is close to the delivery date.’ Moreover, modularity and/or risk management were an outcome of individual learning to build avoidance.

The findings regarding collective learning are more distinct. Eleven out of twelve cases used collective learning to increase avoidance. Particularly, it was found that ten cases learned that coordination of extra manual handling (case 1, 2, 3, 6, 7, 8, 11), alignment of production schedule (case 9, 11, 12), rearrangement of transport (case 4) and delivery (case 7) were used to enhance avoidance. Eight cases learned that inventory management and risk management were applied to build avoidance. Inventory management consisted of the availability of stock (case 1, 4), securing supply (case 1, 2, 4, 6, 7) and adjusting the production planning (case 9, 11, 12). Risk management concerned checks on the supplier (case 3, 8, 12), communication about the circumstances towards the customer (case 5, 12), preventive actions (case 2, 7, 12), product checks (case 9, 11) and a crisis-team (case 12). In about half of the cases it was learned that collaboration and/or information management helped to enhance avoidance.

Collaboration was used to proactively communicate upstream, internal (about product specifications) and downstream. Information management entailed gaining knowledge of upstream, internal and downstream supply chain members. In a third of the cases it was learned that financial management enabled to build avoidance. Case 11 described this as ‘per line we checked if we would stop the line, whether we could do maintenance during this stop.

How much money we would lose during that phase and if we could afford that.’ In only a few

cases it was learned that multiple supplier strategy and/or training increased avoidance. Case 4 described that by collective learning training enabled an enlargement of avoidance: ‘It is training actual people to use the SAP system.’ At last, the case that did not use collective learning to build avoidance, used individual learning instead.

In summary, both levels of learning were applied to increase avoidance. Especially, a strong preference was seen for collective learning.

No learning

At some point, organizations were unable to learn from near-misses to build supply chain robustness. In relation to resistance, this was found in six out of twelve cases and in relation to avoidance, this concerned nine out of twelve cases. In all cases the near-misses were recognized, but the consequences were interpreted as ‘business as usual’. This made the organizations unable to learn from near-misses to enhance their robustness. Different mechanisms were mentioned (see table 7) in the interviews as adequate procedures to enhance the robustness of the supply chain.

TABLE 7: SUMMARY FINDINGS NO LEARNING Supply chain robustness No learning

Resistance Collaboration (8)

Information management (3, 6, 10, 11) Inventory management (11)

Multiple supplier strategy (2, 11) Risk management (10)

Avoidance Collaboration (3, 7, 8) Coordination (2, 5)

Information management (3, 5, 8, 11) Inventory management (9, 10)

Modularity (9)

Risk management (10)

Supplier development (1)

‘No learning’ in respect of resistance was regarded by the lack of a system (information management, inventory management) and a complex supplier relation or a sole supplier (collaboration, multiple supplier strategy and risk management). ‘No learning’ regarding avoidance was referred as ‘no availability’ or ‘no sharing’ of information (information management) and no collaboration between the supplier and the focal organization. Other possible measures were found, but not employed due to different reasons: interpreting a near- miss as a failure (coordination), outsourcing stock and a new production plan (inventory management) or not having responsibilities for stocks (modularity, risk management).

In sum, the inability to learn from near-misses influenced both resistance and avoidance.

Organizations that did not learn, showed slightly more impact on avoidance. Furthermore, organizations that did not learn from near-misses to enhance resistance or avoidance, did recognize the consequences. However, measures were neither in place nor taken or were outsourced.

Overall, different ways and different levels of learning from near-misses were used to build

supply chain robustness. Exploitative and collective learning from near-misses were most

frequently applied to build both resistance and avoidance. However, the underlying

mechanisms that enhance supply chain robustness differed regarding resistance and

avoidance. Resistance was most enhanced by organizations that learned from near-misses to

apply collaboration, risk management and inventory management. Moreover, all cases

learned from near-misses about coordination to increase avoidance and it was frequently

learned that collaboration, information management, inventory management and risk

management enhance avoidance.

DISCUSSION

This study aims to explain how organizations learn from near-misses to build supply chain robustness. The first and most remarkable finding is that all cases are able to detect near- misses and learn from them. This is surprising, because from literature it is known that near- misses often remain unrecognized (Soyer & Hogarth, 2015). It is also known that the recognition of near-misses can be increased when safe, anonymous reporting channels exist to communicate about near-misses (Tinsley et al., 2011). In this study, the interviews have been a stimulus to create awareness of near-misses amongst organizations by explaining the phenomenon and to make organizations explicitly think about near-misses and their effects on the supply chain. In addition, current literature describes that a positive, safe business climate and a significant learning environment need to be in place, to be able to learn from near- misses (Dillon et al., 2016). The findings of this study demonstrate that the aforementioned evaluation of near-misses consequently showed the learning abilities of the organization to build supply chain robustness.

Moreover, the findings show that organizations use both exploitative and exploratory learning from near-misses to enhance supply chain robustness. In particular, exploitative learning is applied more often than exploratory learning. Hence, an exploitative learning approach seems a natural way for organizations to learn form near-misses to build robustness. Even though opportunities may lie in the further development of explorative learning from near-misses.

Characteristics of explorative learning are search, discovery and experimentation (March,

1990) and may lead to innovative solutions to improve supply chain robustness. An example

of explorative learning from near-misses is given in one of the interviews (case 9), in which

the organization has learned that hiring a new employee to oversee the stock of raw materials

was a successful approach. This leads to the following proposition:

P1. Development of explorative learning from near-misses has a positive impact on mechanisms to build supply chain robustness.

Focusing on the levels of learning, the findings of this study show that individual and collective learning from near-misses are both used to build supply chain robustness, although there is a clear indication that collective learning takes place more frequently. Collective learning happens when the collective recognizes something that suggests a more effective way of operating (Hayes & Allinson, 1998). For example, some cases (1, 5, 9, 11, 12) learned from the near-miss, that rearrangement of the production process was needed, to be better prepared for the future. Supply chain management describes: to be able to improve the response to unexpected events, disruptions are a trigger to apply collective learning for adapting the supply chain (Zsidisin & Wagner, 2010). More specifically and in line with supply chain disruption theory: analyzing previous events leads to a better understanding of the process and helps to prevent similar disruptions in the future (Bode, Wagner, Petersen, &

Ellram, 2011). This means that disruptions and near-misses show similar characteristics and a preference exists for collective learning from both to deal with unexpected events. Also, learning from near-misses prevents future near-misses and inherently helps to build supply chain robustness. This has led to the following propositions, which read as follows:

P2. The greater the collective learning capability from near-misses, the higher will be the impact on supply chain robustness.

Another important finding is that several organizations do recognize near-misses, but were

unable to learn and therefore were unable to build supply chain robustness. From literature it

is known that it can be difficult to learn from near-misses (Dillon et al., 2016). A bias can be a

barrier for potential learning. An example of outcome bias is described in one of the

interviews (case 7) where a near-miss is interpreted as a failure with no specific impact. When organizations perceive the near-miss as ‘business as usual’, it obstructs learning, because it is a confirmation that their existing knowledge is already a trustworthy representation of the ordinary. This is in accordance with the theory proposed by Madsen and Desai (2010), who describe that it is easier to learn from failures, than from small failures or near-misses (successes), because failures challenge the existing knowledge and successes stabilize this search. This has led to the following proposition:

P3. A safer business environment, increased awareness and appropriate evaluation procedures, will create conditions for better learning from near-misses to improve supply chain robustness.

The interviews show that it was learned that a range of underlying mechanism helped to improve supply chain robustness. Considering these measures and their influence on resistance or avoidance, one thing stands out: collaboration is the only underlying mechanism that emerged at both ways and at both levels of learning. This is also seen in literature that collaboration among supply chain members should be in place to avoid and resist disruptions (Kleindorfer & Saad, 2005). Kleindorfer and Saad (2005) take collaboration one step further and suggest that testing of early warning and crisis management systems throughout the supply chain are necessary to resist and avoid change. This step of testing was recognized during the interviews when internal audits or training took place. However, none of the organizations tested their collaboration among other supply chain members. Although this study did not involve finding successful or less successful mechanisms to build supply chain robustness, collaboration seems unique and important to mention. This has led to the following proposition:

P4. Collaboration is positively related to build supply chain robustness.

CONCLUSION

This paper examined the role of learning from near-misses to build supply chain robustness.

At first, it is concluded that all interviewed organizations were capable of recognizing near- misses. Secondly, it was found that by studying near-misses all four learning approaches (exploitative, exploratory, individual and collective learning) enhanced supply chain robustness. In particular, exploitative and collective learning were most frequently seen.

Finally, in some cases organizations were unable to learn, despite recognizing the near-miss, because measures were not in place, not taken or were outsourced.

Managerial Implications

The theoretical propositions of learning methods from near-misses to enhance supply chain robustness also suggest some managerial implication. At first, learning in the context of near- misses does not happen by itself. To be able to learn from near-misses, managers should be aware that near-misses ought to be recognized and evaluated, because near-misses show valuable information to prevent future near-misses (Tinsley et al., 2011). The importance of awareness of near-misses has been seen in other sectors than supply chain management and helped in many ways. The American National Safety Council (NSC) (2013: 1) describes approaches to create awareness of near-misses: ‘leadership must establish a reporting culture reinforcing that every opportunity to identify and control hazards, reduce risk and prevent harmful incidents must be acted on.’ and ‘Investigate near-misses to identify the root cause and the weaknesses in the system that resulted in the circumstances that led to the near-miss.’

Furthermore, a cost reduction may be assumed when near-misses are recognized and acted

upon.

Moreover, theory suggest that to be able to learn from near-misses, a positive, safe business climate and a significant learning environment need to be in place (Dillon et al., 2016). The findings showed that the cases had the right climate and environment to recognize near- misses. Thus, managers should provide circumstances that make learning from near-misses possible.

The interviewed organizations most often used exploitative and collective learning from near- misses to enhance supply chain robustness. More innovative solutions to improve supply chain robustness are possible by exploratory learning. Managers themselves play an important role in learning processes and should also facilitate these processes. In order to maintain the knowledge of near-misses, a systematic effort of discussion and circulation of reports is required (Kletz, 1984). Moreover, Hora and Klassen (2013) suggest that training (e.g.

periodical workshops and meetings) can enhance learning of the organization.

Limitations and Future Research

Being one of the first studies to research the effect of learning on supply chain robustness results in some limitations. As only a very general approach of learning methods has been taken into account, while other types of learning might be relevant as well. Therefore, future research should look more in-depth as to which learning styles/approaches will help to build supply chain robustness.

Secondly, the chemical processing industry helped to take the first step in researching near-

misses in supply chain management. However, other industries might have a different

understanding of supply chain near-misses and learn in different ways to establish supply

chain robustness. Thus, future research should investigate the concept of supply chain near-