By Danesha Marasinghe, Industry Consultant, Supply Chain at Teradata
Supply chains have become increasingly complex and global in their reach. However, many firms are unaware of this. In our previous post, we mentioned that 80% of global automotive supply chains have links to China, in some way or the other. Many of the OEMs may have realised the linkage when they were actually hit by the supply disruption.
Most often, efficiencies in supply chains are achieved by applying lean methods and associated “just in time” inventory policies. While these methods work well during most normal operating conditions, they do not respond well when subject to significant disruptive events.
COVID-19 is a disruptive event affecting both demand and supply at a global level and at an unprecedented speed. Firms are facing severe shocks to their supply chains and will continue to do so in the coming months and quarters.
- The first shock came when manufacturing in China came to a standstill as COVID-19 raged through Wuhan.
- The second shock is ongoing as automotive demand contracts while societies take drastic measure to combat COVID-19. Many smaller suppliers are at risk of going bankrupt. This raises the prospect of further disruptions.
How can automotive firms respond effectively to such disruptions in the long term?
After COVID, we predict a renewed C-level focus in the areas of supply chain strategy. Going forward, the industry may have to focus on building flexible and resilient supply chains that allow firms to rapidly reorient and respond to disruptions. This can be achieved by including risk mitigation as an integral part of the supply chain along with cost and quality.
Here we will discuss three critical capabilities in that area:
1) A deep supply chain awareness: a detailed view of the supply chain, which goes beyond the first and second tier supply base.
- This is a resource intensive exercise which starts with the bill of material (BOM) and maps the supply chain to the raw material level.
- Many companies don’t perform this activity because of the length of time and effort it requires. COVID-19 has shown the value of having this awareness. The small number of companies that invested in mapping their supply chains before the pandemic have emerged better prepared to respond.
- A supply chain map focuses on the key components used in the highest revenue generating products. The goal is to drill down as many tiers into the supply chain as possible to identify any hidden critical suppliers.
- The map should include information on activities the primary supply site performs, any alternative sites the supplier has that can perform the activity and the time it takes to transfer operations to the secondary site.
Achieving this awareness requires firms to manage vast amounts of data on many aspects of their supply chain. Robust cross functional data models are required to create and maintain the map.
2) An early detection capability: the ability to ingest, process and generate insights from data captured from varied sources to provide early warning on potential disruption events.
- Firms must acquire data from news feeds, analyst reports and social media which will include structured and unstructured text and multimedia data.
- An ensemble of classification models (Decision Trees, Self-Organising Maps (SOM), etc.) configured to process event streams can help to identify potential events and alert the firm. This requires sophisticated information modeling and data science capabilities, and the ability to deploy advanced analytics at scale.
Let us consider the traditional approach to risk modeling in a supply chain before moving on.
Modeling Risk in Supply Chain: Traditional Approach
Traditional methods for risk modeling rely on two key data points on potential events that could disrupt a firm’s operations: 1) the likelihood of the event occurring, and 2) the magnitude of the impact caused. This works well for common supply chain disruptions where historical data to quantify risk is available (e.g., poor supplier performance, forecast errors, supplier financial solvency, etc.) This method doesn’t work well for low-probability, high-impact events like COVID-19, because historical data is nonexistent.
Which brings us to our third capability.
3) An advanced supply chain risk modeling capability: a low-probability, high- impact event modeling and scenario planning tool that focuses on potential failure points along the supply chain. This allows a firm to quantify the impact of an event, irrespective of the probability of it happening. It is crucial that the tool is automatable as it must support quick updates and near-real-time execution as the supply chain is in flux during a disruption.
To build this capability, the firms must use data from their deep awareness capability — combined with the potential events identified through their early detection capability –and run optimisation models to generate a risk exposure index score for nodes in the supply chain. The score allows mangers to quickly pinpoint nodes with the highest risks, versus lower risks, and implement mitigation actions.
Overview of advanced supply chain risk modeling
- Map the distribution network (dealerships, warehouses, etc.) and combine it with the supply chain map:
- Integrate data from multiple tiers of the supply chain, including supply information, bill of material, operational and financial measure, inventory levels — both in-transit and on-site — and demand forecasts for each product.
- Represent the supply chain at the lowest granularity. This enables drill-down and roll-up capabilities which allows identification of hidden dependencies within the supply chain.
- Determine the time it takes a node to recover should its operations be impaired by a disruptive event.
All these three capabilities require automotive firms to securely manage and analyse diverse types of data at scale.
We at Teradata help transform supply chains to be more flexible and resilient, through the power of data and analytics at scale, so that industries can be better prepared for such disruptive situations.