Students

Business analytics: A better way to dig into supply chains

New research by a team of information systems and supply chain experts from the W. P. Carey School of Business shows that adding business analytics to the supply chain toolbox can make the process of evaluating supply chains easier, faster and more targeted.

2016-03-29

Large manufacturers simply can’t have insight into what’s going on with each and every one of their suppliers, much less with those suppliers’ suppliers or with suppliers even further upstream. But opportunities could be out there: a mid-size supplier with an early read on changing economic conditions, or a small supplier with an innovative idea that could pay off big. Risks could also loom: a critical supplier whose factory suffers an explosion, or another whose environmental mishaps taint everyone’s brand.

Until now, getting insights into tier after tier of an original equipment manufacturer’s suppliers has been tedious and time-consuming. But new research by a team of information systems and supply chain experts from the W. P. Carey School of Business shows that adding business analytics to the supply chain toolbox can make the process easier, faster and more targeted. The team’s new approach can help manufacturers learn about more of their suppliers and regain some of the power they’ve too often delegated to top-tier suppliers.

Professor of Supply Chain Management Thomas Choi has been studying the problem ever since he first learned of it while working with consumer electronics giant LG Electronics Inc. during the Great Recession. “All the big manufacturing companies that make consumer goods were looking at the consumer markets for some early signs of the economy improving,” Choi recalled, and LGE’s chief procurement officer was seeking more leverage in his negotiations with a top-tier supplier. It wasn’t until the procurement executive visited one of LGE’s second-tier suppliers for unrelated reasons that he picked up some useful information. He saw that the supplier, which produced semiconductors for a variety of industries, was ramping up its production. The new insights helped LGE lock in then-low prices with their suppliers and saved considerably on its purchasing.

“Wouldn’t it be nice if we can do a study that would identify these critical suppliers that big buying companies may not yet know about?” Choi thought. Some suppliers might not be consequential, but others could be what are known as “nexus suppliers” — those that provide key products or critical information and whose relationships with other suppliers bolster their positions in a supply network as discussed in an article published in the Journal of Supply Chain Management in 2015.

Nexus suppliers, though, are hard to identify when a major manufacturer can easily have more than tens of thousands of suppliers spread from its top tier down toward its raw materials suppliers. Companies that want to trace their way upstream typically examine bills of materials, Choi said — a task that one manufacturer found took a team of more than 100 employees a year to complete. Even then, the proliferation of data and concerns over privacy left the overall picture less than robust.

“The volume of data is tremendous,” said Benjamin Shao, associate professor in the Department of Information Systems. “Business analytics can process voluminous data to derive insights based on the patterns extracted from the data. So there is a perfect match in applying business analytics” to the supplier/network problem.

Creating a model, adding the data

Choi turned to Shao and Zhan Michael Shi, assistant professor of information systems, to tackle the problem. Funded by CAPS Research, a joint venture organization of Arizona State University and the Institute for Supply Management, the team worked on developing a straightforward, mathematical model that could help manufacturers score their second-, third- and fourth-tier suppliers and identify the ones with potential significant impact on the manufacturer’s business.

Shao said they used conventional centrality measures from social network analysis to create a model that captures four different aspects of a supplier’s importance to the manufacturer. The model takes into account a supplier’s degree centrality, or number of links it has with other suppliers; its “betweenness” centrality, or how often it sits on the shortest path between other suppliers; its eigenvector centrality, or how often it connects to important suppliers; and its “farness” centrality or how far it is from other suppliers. The model identifies top-tier suppliers, then their top suppliers, and repeats the process in an iterative fashion as it follows suppliers upstream.

The four centrality measures vary from supplier to supplier. “We do not treat the measures equally,” Shao said. “Instead, our model lets the data speak for itself by picking the appropriate weights on its own.”

The model resulted in the Nexus Supplier Index (NSI), a single measure that assigns a number between 0 and 1 to each supplier. The higher a supplier’s index, the more likely it is to be a nexus supplier worthy of greater attention from the manufacturer.

Shi focused on finding the best data to feed into the model for analysis. “The main difficulty has been the lack of a reliable and comprehensive dataset,” Shi said. “Information is revealed in many different sources, yet each source tends to be incomplete.”

Fortunately for Shi, financial data vendors have started collecting data on supplier-customer relationships. So the team chose to use the Bloomberg Supply Chain Database, a centralized database that culls information from sources including U.S. Securities and Exchange Commission filings, transcripts of earnings calls, website content and industry reports.

To make the research real-world, the team applied its approach to Honda Motor Co. Ltd. and four tiers of its suppliers. The researchers found the automaker had a network of more than 10,000 suppliers across 83 countries and 66 industry sectors. Choi said Honda executives were surprised at some of the suppliers the new approach uncovered. Some were companies the executives hadn’t heard of, and others were big names that the executives hadn’t realized contributed in multiple ways to Honda.

“The data pointed out the blind spots,” Choi said.

Multiple benefits of new approach

A big benefit of the NSI is its ability to give manufacturers leads they can check out and to give them early indicators of which suppliers might pose risks or provide opportunities. Manufacturers can look at the products an upstream supplier produces and how many of them end up in the manufacturer’s products. Managers then can identify which suppliers have promise and are worth establishing direct relationships with.

The index also allows manufacturers to categorize potential nexus suppliers as operational, monopolistic or informational. Operational suppliers have many ties to others in the network, giving them significant roles in the end product. Monopolistic suppliers are common-denominator suppliers that are at the root of supply networks, making them important to the continuity of supplies. Informational suppliers have diverse connections, which can provide market information or technological innovations to the manufacturer.

A final benefit of the new index: what used to take dozens of people a year to accomplish could now be done by one employee in less than three months. However, it is important to note that the Bloomberg data lack resolution compared to other efforts that use the bill of materials.

The researchers recognize that the NSI is just a start. It works at the company level, but it can’t yet suggest to a manufacturer which of a supplier’s factories are most critical, or which of its countries of operation pose the most risk. Nor can it get down to the parts level, identifying which suppliers’ parts are critical to the manufacturer’s business.

Going forward, Choi sees two possible avenues for research. One is to create a longitudinal NSI, which would show a manufacturer how its network evolves over time and what might trigger those changes. Another is expanding the index to the product level, allowing a manufacturer to examine supplier products worth integrating in its own products.

Shao expects the next step to be testing the model with another company’s dataset, and Choi said other manufacturers that are members of CAPS Research already have indicated interest.

“What we did here is considered very futuristic,” Choi said. “We’re trying to identify critical suppliers that you may not know about, and the role of publicly available data sources and business analytics.”

The bottom line

Choi, Shao, and Shi suggest these takeaways for different players in a supply network:

• For manufacturers and other buyers — The NSI gives you a feasible way to get a glimpse at your critical, or “nexus,” suppliers. Those are the ones to consider forging more direct relationships with, whether they are operational, monopolistic or informational suppliers, to either manage risk or find new opportunities.
• For top-tier suppliers — Remember, as the manufacturer becomes more competitive and sells more products, you will profit in the long run, too. You can apply the index to your own network and identify any of your nexus suppliers that might be your “blind spots.”
• For lower-tier suppliers — You might be a few tiers away from the large manufacturer, but there is still opportunity to stand out. If your business is exposed to a variety of industries, you could be in a position to see innovative ideas and become an informational nexus supplier to manufacturers.

Business analytics Big Data KnowIT Faculty Managing IT