To err is human, but most research on supply chain management doesn't take psychological, functional, incentive-related, and other biases into account. HBS professors Rogelio Oliva and Noel Watson have devised their latest research to learn how such behavioral dynamics can affect the making of forecasts as well as decisions about inventory.
In this interview with HBS Working Knowledge, Oliva and Watson say a key to understanding and avoiding supply chain inefficiency may be a better understanding of the behavioral side of the equation.
Sarah Jane Johnston: Your research explores the behavioral dynamics that may exist in supply chain management. What led you to study this in particular?
Rogelio Oliva and Noel Watson: Supply chains are incredibly complex with numerous moving parts. To appreciate this complexity, it is useful to first realize that activities in the supply chain space can be divided into strategic and operational activities. Strategic activities include, among others, long-term capacity planning and network configuration of warehouses, distributors, and retail stores. Examples of operational activities include short-term demand planning (including forecasting and inventory management), production, and logistics.
Many of these [academic] papers pursue an optimizing approach given the assumption of a completely rational decision maker.
In a supply chain, however, these activities are usually spread over multiple functions or organizations and sometimes over lengthy time horizons. Therefore, on top of this categorization of activities, it is necessary to overlay a coordination system: an explicit definition of processes, responsibilities, and structures to bring together multiple functions and organizations. The design of the coordinating system expands the supply chain problem space to include, among other things, the assignment of roles and decision rights among the coordinating partners, the selection of partners, the design of incentives, and the design of processes to monitor performance, set goals and solve problems. The particular approach taken in the design of the coordination system determines the complexity of each partner's role.
The strategic and operational activities and the design of the coordinating system provide much of the subject matter for over 10,000 peer-reviewed academic papers on supply chain management. Seeking to provide prescriptive recommendations, many of these papers pursue an optimizing approach given the assumption of a completely rational decision maker. Testament to the complexity of these activities is the high level of mathematical sophistication used to derive the recommendations.
In reality, managers do not act as completely rational agents. (We imply no offense in this). In response to the complexity of the supply chains, managers limit the problem space, whether consciously or unconsciously, by selecting a limited set of inputs to inform their decisions, and using simplifying heuristics (shortcuts) when making these decisions. This behavioral approach (to distinguish it from a completely rational agent approach), however, introduces biases in the supply chain management decisions. We use the term bias here to represent the quality of having a systematic inclination in a distinguishable direction, typically implying suboptimal results.
In principle, the coordinating system should be designed to account and compensate for the individual and functional biases that the supply chain partners could have. However, since the approach taken in the design of the coordinating system determines the complexity of each partner's role, the design is also in part responsible for the biases that a function or an individual might have. That is, the coordination system design choices also predispose individual partners to certain problem space simplifications and heuristics. This dual realization—that the coordinating system needs to account for individual biases and that the coordinating system is in part responsible for an individual's biases—creates a level of design complexity not currently explored by the supply chain literature.
The potential impact of these dynamics on a company is that its supply chain limps along.
We believe the appreciation of these behavioral dynamics on supply chain management can provide the key to understanding and removing the many and oft-repeated examples of inefficiency that plague it. Examples of these inefficiencies include the bull-whip effect [when a fluctuation becomes exaggerated along the supply chain -ed.] and its related stretches of wasted inventory and unmet demand.
Q: What are some of the behavioral dynamics that supply chain managers should be aware of? Did anything in your research catch you by surprise?
A: We began our examination of behavioral dynamics in supply chains examining demand and supply planning processes. In particular we were interested in how behavioral biases, e.g., psychological, incentive related, etc., affect both the making of forecasts and approving/modifying inventory policy decisions. Most research in these areas in the academic supply chain literature is based on assumptions that are (a) derived from experiments (e.g., students playing the beer game), (b) derived from surveys asking managers how they make the order decision, or (c) made to ensure tractability in analytical models.
We were interested in a more direct analysis of managers' decision making for these activities to gain a specifically context-based understanding of how these decisions are made as opposed to how managers say they are being made. A more systematic understanding of how and when these biases affect decision making along with their consequences would benefit companies as they seek to improve their inventory decision-making capability.
We are still in the preliminary stages of this research, but our work so far is informed by much research in the forecasting literature about the different biases that prevail. As a result, we have not been surprised by the biases that we observe in demand planning. We've found evidence of individual biases such as optimism, recency, search for supportive evidence, and illusory correlations—all well-documented biases in the forecasting literature. In appreciating the problem of demand planning as a coordination problem across functions, the behavioral approach to the motivation and incentives component of coordination seems a common foil. Natural or more directly instituted functional incentives play havoc with motivating higher aims of efficient demand planning. Measuring and interpreting efficiency along these higher aims can also be behaviorally compromised when measures have to be simplified or aggregated.
Q: What is the potential impact of these behaviors on a company's supply chain activities? What steps do you expect managers will need to take to identify and overcome these issues?
A: The potential impact of these dynamics on a company is that its supply chain limps along, never quite achieving its full potential as a source of competitive advantage or of the reduction of the entry barriers to other competitive settings.
It somewhat goes without saying that being aware of your problem is a prerequisite to its solution. However, because individual, and probably by extension organizational, biases are often difficult to self-diagnose but easy to diagnose in others, explicitly stating biases is an important step in improving performance of these types of process. Unfortunately, just identifying biases is not enough to get rid of them. We humans are resilient in our beliefs and assumptions. Intervention needs to be at the processes(roles, activities) and structural levels (components of coordination) to explicitly address the effect of individual and functional biases at work.
These kinds of problems require more than IT implementation or optimization algorithms. In some cases, ever-present bias must be fought with bias; behavioral problems need behavioral solutions. From our observations there seems to a palpable orientation in managers and executives who purposefully address the behavioral issues in demand planning. They see the problem in demand planning in terms of the behavioral dynamics involved rather than, say, as an optimization problem with objective to minimize forecasting accuracy.
The ubiquitous and resistant nature of these behavioral impediments to effective supply chain management implies that its improvement is difficult to replicate, further strengthening its ability to provide competitive advantage.