A Systematic Investigation of the Integration of Machine Learning into Supply Chain Risk Management

Our article on “A Systematic Investigation of the Integration of Machine Learning into Supply Chain Risk Management” was published. The main objective of the paper is to analyze and synthesize existing scientific literature related to supply chain areas where machine learning (ML) has already been implemented within the supply chain risk management (SCRM) field, both in theory and in practice. Furthermore, we analyzed which risks were addressed in the use cases as well as how ML might shape SCRM. For this purpose, we conducted a systematic literature review. The results showed that the applied examples relate primarily to the early identification of production, transport, and supply risks in order to counteract potential supply chain problems quickly. Through the analyzed case studies, we were able to identify the added value that ML integration can bring to the SCRM (e.g., the integration of new data sources such as social media or weather data). From the systematic literature analysis results, we developed four propositions, which can be used as motivation for further research.

Schroeder, M. & Lodemann, S. (2021): A Systematic Investigation of the Integration of Machine Learning into Supply Chain Risk Management, in: Logistics 2021, 5(3), 62; https://doi.org/10.3390/logistics5030062

Supply chain risk identification tools

The first step in the process is to identify supply chain risks. If supply chain risks are not comprehensively identified, only limited control measures can be taken. In risk identification, various instruments can be used to identify supply chain risks:

Collection and search methods for risk identification (excerpt)

In general, a distinction must be made between collection and analytical search methods. While collection methods are predominantly suitable for identifying existing risks, analytical search methods are aimed at identifying future and previously unknown risks (Romeike 2004a, p. 157; Romeike 2004b, p. 174). In the case of collection methods, checklists are frequently used in practice to help identify the sources of risk. In addition to the high degree of aggregation, the lack of completeness of the identified risks is problematic here (Ziegenbein 2007, p. 60; Schorcht 2004, p. 115). Furthermore, the 5 ‘whys’ method—for example, interviews or surveys—can be used to identify risks (Jüttner 2005, p. 128; Burger & Buchhart 2002, p. 69; Schubert 2004, p. 164, Romeike 2004c, p. 185).

In the case of analytical search methods, an additional distinction must be made between analytical and creative methods (Burger & Buchhart 2002, p. 68). Some analytical search methods, such as FMEA (Failure Mode and Effect Analysis), were originally developed for quality management (Bergener 2006, p. 274; Brühwiler 2003, p. 182; Ziegenbein 2007, p. 54). Since risk management has similar process structures as quality management, the established methods are largely transferable between the two (Romeike 2004b, p. 176). Compared to creative methods, analytical methods provide a frame of reference and are characterized by a systematic approach. Creative methods, on the other hand, are based on creative processes characterized by divergent thinking (Romeike 2004b, p. 177). Here, for example, brainstorming or free-writing can be cited as common idea-generation methods (Schorcht 2004, p. 187; Ziegenbein 2007, p. 60).

In addition, when identifying risks, it must be decided whether individual risks or aggregated (summarized) risks should be recorded (Romeike 2004c, p. 183; Singer 2012, p. 65; Winkler & Kaluza 2015, p. 311). Selecting a risk landscape that falls within a relevant scope also plays an important role. For example, only risks within one’s own company can be identified or risks that may occur within the entire supply chain (Kajüter 2003, p. 123). Accordingly, before selecting the appropriate tools, whether and how many supply chain partners should be included in the risk identification step should be decided (Kersten et al. 2013, p. 31). Direct suppliers or customers (Tier 1) can be integrated into the risk identification process, or even extended to the initial supplier or end customer (Tier 2 to n) (Eberle 2005, p. 48ff.).

Extract from Schröder, M. (2019): Structured improvement of supply chain risk management. In: Supply Chain Management – Contributions to Procurement and Logistics, Series-Editor: Essig, M.; Stölzle, W., Kersten, W., Springer Gabler: Wiesbaden

How to audit SCRM? Conducting the Audit (6/9) – List of questions (part a/c)

Below is a list of questions that can be used to audit the SCRM, concerning the topicsTransparency of supply chain structures and processes, Information and dependencies of suppliers and Identification.

Transparency of supply chain structures and processes Are the most important, value-creating corporate processes defined (e.g. development to production to sales)?
  Is there accountability in these operational processes, i.e. are the potential risks and their impacts on different departments known and addressed?

Information and dependencies of suppliers

Is there a list of all direct upstream suppliers?
  Is the most important company data of the upstream suppliers documented (corporate headquarters, production facilities, local contact persons, cell phone numbers, etc.)?
  Are the supplier contacts with decision-making authority known?
  What are all of the value creation processes of the direct upstream suppliers?
  What are the dependencies between upstream suppliers?
  Have they been properly identified?
  Is information about the suppliers obtained on a regular basis?
  Who are the most critical suppliers (critical suppliers can be characterized as the highest proportion of sales, or as the supplier of the material for the most products which cannot be easily replaced, etc.)?
  If the most critical supplier fails, what are the financial consequences?
  How often does the operational business check whether certain deadlines have/will expire?
  How often are the supply contracts checked to see if they need to be transferred to a new company standard?
  Is information about suppliers regularly shared with other departments?
  Does an on-site visit of the most critical suppliers take place every three to five years?
  Is there a structured guidelines in place for communicating with suppliers?


Are operational supply chain risks systematically identified at regular intervals?
  Are strategic supply chain risks systematically identified at regular intervals?
  Is there a catalog for the systematic identification of supply chain risks?
  Are external as well as internal corporate risks considered during the identification of risks?
  Are the interfaces with supply chain actors considered during risk identification?
  Is it always possible to add newly identified supply chain risks to the catalog?
  Are the results of the supply chain risk identification from each individual workplace recorded in writing?
  Are the results of the supply chain risk identification from the entire department recorded in writing?
  Are risks from supply chain partners that may have a negative impact on the company considered in the identification process?
  Is the identified supply chain risk list up-to-date?
  Is the operational supply chain risk assessment complete?

Extract from Schröder, M. (2019): Structured improvement of supply chain risk management. In: Supply Chain Management – Contributions to Procurement and Logistics, Series-Editor: Essig, M.; Stölzle, W., Kersten, W., Springer Gabler: Wiesbaden

Changes in Risk Management via Big Data

The meaningful and beneficial evaluation of large amounts of unstructured data is the next important step for effective risk management as a whole to take. This challenge, along with several other new, pertinent challenges, will be briefly summarised below (Rogers 2017): 

  • The integration of Big Data into the risk management process will be further expanded in the future. Thus, both the accessibility and validation of the data and the use of corresponding analysis and evaluation tools will play a central role. 
  • Within the scope of risk identification, preliminary considerations should therefore be made to address questions such as “Who feeds the data?”, “How can the timeliness, completeness and compatibility of the data be ensured?”, and “How can data quality be guaranteed? 
  • It may also be necessary to adapt customized procedures in order to use limited resources sparingly: for example, it is advisable to first define the goal of data analysis (e.g. predicting the risk of Ebola infection) instead of trying to draw as many evaluation options as possible from an available data set without a specified goal in mind.
  • For better, more extensive data availability in the future, risk management must be geared more strongly towards interdisciplinary cooperation within companies. At the same time, as shown above, a multitude of departments may benefit from the evaluation of the results. 
  • For data processing, the correct IT tools–Amazon Elastic MapReduce, Microsoft Azure Cortana Analytics Suite–must be used to efficiently identify the required information (e.g., implementing machine learning and artificial intelligence).
  • However, the continuous analysis of unstructured data cannot be mastered by IT tools alone; the support of experts, i.e. data scientists, who perform technical data analysis and provide decision assistance is also necessary. This includes, among other things, a manual cleansing of raw data to increase data quality and the programming of algorithms for the automatic use of real-time data.
  • The increased emphasis on data analysis in risk management will thus lead to a division of responsibilities between data scientists and risk managers.
  • In the future, one of the risk manager’s core tasks will be to consider the results of data analysis as a forecasting aid in management decisions. 
  • The primary tasks will consequently become more data-driven, and, as a result, the demands on the qualifications of risk managers will also change.
  • The automated research and evaluation of real-time information will lead to a reduction in the manual activities of the risk manager. This leaves more time for analytical and strategic tasks, i.e. medium and long-term solutions can be developed based on the results of Big Data Science (Al-Khazrage 2018, p. 58). 

It should be noted that the use of Big Data in risk management is associated with various outcomes. In order to prepare for these outcomes, companies should–depending on their resources–initiate appropriate measures at an early stage. 

Excerpt from Schröder et al. 2019, Neue Anwendungsmöglichkeiten für Risikomanagement durch die Einsatz von Big Data – Zwei Fallbeipspiele, in: Schröder, Meike; Wegner, Kirsten (eds.) Logistik im Wandel der Zeit, Springer: Wiesbaden, pp. 121-136.


MBP 9/16 Measures and Best Practices on “Supply Chain Risk Identification”

“Of the many risks, the top five will be discussed again together in a meeting.” (Interview – Head of Additive Manufacturing Solutions, aerospace industry) 

In the course of the risk identification process, significant risks to the company are systematically recorded. Only those risks that are identified here can be assessed and managed in the following. The risk identification phase is, therefore, often regarded as particularly significant as it has a direct impact on the effectiveness of the entire process (Kersten et al. 2012, p. 293). 

In the numerous expert interviews and focus groups with company representatives and scientists for the development of the maturation model, the following measures were identified:

  • Identify the operational and strategic supply chain risks at regular intervals.
  • Also, try to proactively identify risks.
  • Categorize your identified risks: e.g. in procurement, process, control, demand, and environmental risks.
  • Try to use analytical methods, such as FMEA and morphological, as well as the standard creativity methods like brainstorming and free-writing to identify supply chain risks.
  • Ensure that the process of supply chain risk identification can be adapted to quickly changing risk situations by developing a versatile risk catalogue appropriately and maintaining easily adaptable surveys.
  • Ensure that the identified supply chain risks are up-to-date.
  • Ensure that the identified supply chain risks are complete (especially via the principle of dual control and with the involvement of a control authority)
  • Discuss the identified supply chain risks and possible consequences not only within departments but also across departments.  
  • Ensure that an overview of the most critical supply chain risks is available on an ad-hoc basis.  
  • When appropriate, you should include supply chain partners in the risk identification process.
  • Where possible, also use external data to identify supply chain risks.
  • Perform supply chain risk identification at regular intervals and which are formulated independently of previous results (start from zero).
  • If possible, use key-figure, extrapolation-oriented, and indicator-oriented early warning systems to identify supply chain risks well in advance.  

Extract from Schröder, M. (2019): Structured improvement of supply chain risk management. In: Supply Chain Management – Contributions to Procurement and Logistics, Reihen-Hrsg.: Essig, M.; Stölzle, W., Kersten, W., Springer Gabler: Wiesbaden

Suche in OpenEdition Search

Sie werden weitergeleitet zur OpenEdition Search