Many researchers say that their greatest flashes of inspiration come not from extraordinary events but from everyday activities. Intuitive leaps are triggered by observing life unfolding around them, they say. We are all familiar with the anecdote—historically accurate or not—of an apple falling on Isaac Newton’s head while he rested beneath the tree, jolting him to contemplate universal gravitation. And readers of Thomas Friedman’s best-selling book on globalization, The World Is Flat, understand why he conceived of his metaphor of a level global playing field one night after visiting Bangalore, India’s Silicon Valley.
A similar process occurred when Jan Van Mieghem (Professor of Operations Management at the Kellogg School of Management) asked Cort Jacoby and Ruchir Nanda from Deloitte Consulting to make a presentation in his MBA course on operations strategy. Van Mieghem watched while his guests discussed a case in which a client—a $10 billion, high-tech U.S. manufacturer of wireless transmission components—simply wanted to know, “How much should I buy from each of my sources in China and Mexico?” They flashed a graph depicting the minimal costs of various sourcing strategies, and the simplicity of the slightly skewed U-shaped curves inspired Van Mieghem to untangle a real-world dilemma that many businesses face: designing an optimal sourcing strategy when choosing between two global suppliers, one low-cost but distant and sluggish, the other expensive but near and nimble.
It is a simple question on the face of it—but the devil is in the details, especially if you need hard numbers. In the consultants’ case, Mexico was the nearby but expensive source and China distant but cheaper. “How can we align the strengths of each and minimize our total costs?” the client wanted to know. The consultants modeled the costs and benefits at five incremental points, ranging from securing everything from Mexico (0) to securing everything from China (100). The best sourcing solutions allocated about 75 percent to China in most of the cases they analyzed, so the client was urged to follow this rough “three-quarter” rule.
“I looked at their graph and thought, ‘These are such beautiful curves that cry out for mathematical analysis, exactly what we do in our research,’” Van Mieghem recalled. “We build mathematical models to guide optimal operations strategy. Global sourcing is very important in practice, but leads to quite complex analytical problems—and there is no quantitative theory that I’m aware of that can guide the strategic allocation of uncertain demand to two global suppliers.”
Bridging Theory and Practice in Global Dual Sourcing Dilemmas
Van Mieghem said this basic question of how to split demand is one many seasoned executives approach intuitively: they know that China is less expensive and that often more of the demand can be allocated there, but exactly how much to allocate is a question that must balance cost with service, and is so complex that it can cause decision makers to exhaust both their calculators and their patience. So Van Mieghem and his colleague Gad Allon (Assistant Professor of Managerial Economics and Decision Sciences at the Kellogg School of Management) searched for studies that quantify the best global dual-sourcing strategies. What they found was a gap between theory and practice. There were complex mathematical algorithms useful to academics and some oversimplified papers that were not of much use in advising business professionals. So the pair decided to create a middle ground and deliver a useful formula to the business community based on academic theory.
To illustrate what they were seeking to solve, imagine that you run a multinational organization that sells high-end, carbon fiber bicycle frames and buys from manufacturing plants in China and Mexico. Every quarter you must translate your market demand forecasts into target production quotas for both sources. How do you split the demand between the two plants and place a hard number on the orders? The best dual-sourcing strategy in this scenario, according to Van Mieghem and Allon, minimizes the “total landed cost,” a term describing the total expected cost of making and delivering the product, including the cost of holding and maintaining inventory, which comprises most of working capital (shown in Figure 1).
Figure 1: Total Landed Cost
Note: COGS: Cost of goods sold; TLC: Total landed cost. Total landed cost as used by Van Mieghem and Allon includes working capital. Its inventory costs depend on the service level and are the most difficult component to quantify.
A good first step in minimizing total landed cost is to adopt a sourcing policy the authors term “tailored base-surge” (TBS), which splits demand allocation into base and surge demands. Using the TBS policy, base inventory needs would be sourced from China while the plant in Mexico would be reserved for satisfying surges in demand, such as might occur after a Tour de France victory by a cyclist using one of the company’s bicycles. “This is a natural solution to split the constant and certain demand from the uncertain,” Van Mieghem said. “But to put a number on this split is incredibly hard. The math gets very complicated very quickly.”
So even if you intuitively decided that base demand should be more cheaply sourced to China, you might get burned if you overallocated there because hidden costs, such as the cost of holding inventory, could cause the total cost to spike. This point is represented in Figure 2, where the “inventory cost” curve begins a steep vertical climb, pulling the “total cost” curve along with it as sourcing from China increases beyond roughly 75 percent.
Figure 2: Total Cost
Note: The working capital, or inventory costs (light blue line), combines with the sourcing costs (red line) to affect the total cost (dark blue line).
Factoring this into their calculations, Van Mieghem brought out what he called the “high-powered mathematical machinery.” He and Allon devised a formula that incorporated the per-unit cost advantage of using China, the fraction sourced from China, the risk of sourcing from China, the volume expected to be sold, fluctuations in demand, and the per-unit holding cost.
“Basically, the formula captures an intuitive trade-off between cost and responsiveness,” Van Mieghem said. For example, if the per-unit cost advantage increased, the company would increase its sourcing of carbon fiber bicycle frames from China. However, if the holding costs increased or the expected product volume decreased, China would become a less attractive source. Similarly, if the fluctuations in demand decreased, the appeal of China would increase.
A Formula for Balancing Global Sourcing
While the formula captures the complexity of the global dual-sourcing and allocation question, Van Mieghem believes it remains simple enough to be used in real-world business decision making. “The whole theory culminates in this mathematically attractive and simple formula,” he said. “It is a simplification of reality, but in the end it still captures the main tensions in the original decision problem.” This model would be most useful for companies that had a reasonable handle on their cost structures and demand forecasts, he added, noting that the consultants who spoke in his class were already supplying it to their clients for real-world applications. Van Mieghem said they were surprised to find that plugging real numbers into the formula resulted in a rough validation of the consultants’ advice for allocating about three-quarters of the demand to the slower source. While 75 percent is not the golden number for everyone, it could be a good starting point for businesses that do not have very reliable numbers.
Van Mieghem and Allon calculated the relative value of TBS dual sourcing, defining it as the value of dual sourcing from both China and Mexico divided by the cost of sole sourcing from Mexico alone. Figure 3 shows that the bicycle distribution company would be better off using both China and Mexico when the relative value of dual sourcing was positive. However, as China’s cost advantage decreased or levels of demand became more volatile and unpredictable, the relative value of dual sourcing would become negative and the company would be better off placing orders only with the plant in Mexico.
Figure 3: Relative Values of Sourcing from China and Mexico
Note: This graph depicts the relative values of sourcing from China and Mexico. As demand fluctuates and volatility increases, there is less value in sourcing from China.
Van Mieghem cautioned that the model has limitations: it accounts for a single product from just two global sources, assumes a single decision maker and a single market, and assumes that the cost parameters do not change over time. He said that his next steps will be to adapt the model to a multiple-product, multiple-market scenario that includes several decision makers. However, despite its limitations the model makes a valued contribution by formalizing professionals’ intuitions and offering a proven method to calculate meaningful sourcing numbers.
Van Mieghem said the beauty of this project is that it came full circle from posing a research question in his classroom, to filling a void in the academic literature, to being applied both in practice and back in the classroom as a teaching tool. He said, “I am very pleased on a personal level with this paper because it is the first time in my work when theory and practice have gone so beautifully hand in hand.”
Picking Up Pennies for Profit
Predictable stock price patterns