A lean view of lean

Abstract

We would like to express our deepest gratitude to Wally Hopp, Mark Spearman, and the eight commentators—all luminaries in Lean—for their excellent Forum articles (Hopp & Spearman 2021, hereinafter, source “A”; Cusumano et al. 2021, hereinafter, source “B”). These articles have stimulated thinking and encouraged rich debate in our community. Over the past 40 years, Lean and the Toyota Production System (TPS) from which it originated have sourced much of the research on how to manage, improve, and connect operations. What began as a term chosen to highlight some salient characteristics of the TPS has evolved into a broad description of everything that improves operations management (OM). Yet, despite decades of effort, our two Forum articles show that scholars still do not agree on what Lean is. Should there be a difference between operational excellence (a general ideal) and Lean (a particular way to achieve it under particular circumstances)? When anything that improves OM and inter-operational connections is classified as Lean, do we lose sight of key tradeoffs that underlie competitive advantage? How do we reconcile Lean as operational excellence with the new appreciation of the value of “fat”—capacity or inventory buffers held to protect against variability or extreme events—under the current increase in recognition of the value of resilience in the OM community? Proponents of Lean claim that it has evolved much over the past 40 years, even apart from the TPS—but, in its evolution, when does Lean outgrow its name? In this editorial our aim is to be provocative and “stir the waters,” seeking to direct and deploy the energy generated by the exciting conversation started by these two Forum articles into a research agenda. Both of us have worked closely with the TPS and Lean for most of our careers. Suzanne began her doctorate at the Harvard Business School (HBS) in 1982 with the intention of studying decision theory. She took a course by Professor Bob Hayes just as the shock of the success of the TPS hit the academic world. The excitement of trying to understand what about the TPS led to its highly counterintuitive and surprising global competitive advantage led Suzanne to switch her field to OM. At this time, the key insights discussed included the new understanding of the importance of shop-floor operators and quality management to successful OM—and the essential role that OM plays in competitiveness. Hall's Zero Inventories (Hall, 1983) had not yet been published, but a photocopy of an early version was being actively discussed by the entire HBS operations department. Schonberger's Japanese Manufacturing Techniques (1982) had just been published. Hayes had recently published “Why Japanese Factories Work” (Hayes, 1981). There was a keen awareness that what was being observed could not be explained by strategy theory, which led to urgent sensemaking organized around the idea of producing just in time (JIT), usually considered in conjunction with Total Quality Management (TQM). Much of the discussion centered around culture, and whether the ability to produce JIT with close-to-perfect quality could be achieved outside of Japan. Schonberger (1982), Hall (1983), and Monden (1983) prioritized the reduction of in-process inventory that causes one workstation to block or starve another (see Schonberger's figures 2 and 3 for a complete explanation of the proposed causal mechanisms, p. 26). Suzanne's research focused on this exploratory stress: When is such a line stoppage expected to lead to process improvement, and when does it just cause lost production—or worse, discouragement and demotivation? Proposing theory about the relationship between line stoppages and learning required Suzanne to propose a definition of a JIT system that permitted determination of whether a given system was JIT or not. While all systems seek to deliver on time (not be late), what is unique to JIT is that it sets the objective of being neither late nor early. A JIT system was thus defined (de Treville, 1987) as one that has a “flow-control” mechanism that prevents production in the absence of a downstream signal. Later that decade, sensemaking about the TPS had evolved into Lean production, with the publication of The Machine that Changed the World (Womack et al., 1990) stemming from the International Motor Vehicle Program (IMVP) at the Massachusetts Institute of Technology (MIT). A relative of the IMVP at MIT, the Lean Aircraft (later, Aerospace) Initiative (LAI), launched Tyson's academic career. Like the IMVP, the LAI also led to a book (Murman et al., 2002). LAI supported Tyson throughout his graduate studies in the mid-1990s and provided him with access to its sponsor companies for the data foundational to both his master's and doctoral theses. Tyson thus could observe how Lean was being applied at companies such as Boeing, General Electric, Lockheed Martin, Northrop Grumman, Raytheon, and Texas Instruments. During most of this time, he was part of the Product Development Focus Group—one of five within LAI—exploring research questions about how Lean might apply in that context. Later on, as an assistant professor, Tyson had a unique opportunity to study Lean implementation in the F-22 program at Lockheed Martin, which led to a paper in this journal (Browning & Heath, 2009). Thus, we have both witnessed firsthand the power of Lean practices and thinking to marshal resources around important research and applications. That said, our industry experiences and other research over the past couple of decades have given us a clear picture of how a failure to study Lean as a set of formal concepts that are precisely defined can impede understanding and communication. We have also observed weaknesses in our ability as a field to do effective research and sensemaking on this key topic. Our intent in this editorial is to map out a path to a formal conceptual definition of elements commonly ascribed to Lean, following the eight rules proposed by Wacker (2004). We then revisit one of the most highly cited articles about Lean—Shah and Ward (2003)—to explicate its contribution and explore the work done as a foundation for rigorous definition development. This analysis lets us propose a path toward (1) identifying, rigorously defining, and exploring operational attributes that underlie Lean and operational excellence, (2) disambiguating Lean from operational excellence, and (3) distinguishing between theory that captures tradeoffs between (i) Lean attributes and (ii) conceptual exploration—pre-theory—that underlies Lean thinking and Lean as a philosophy. Hopp and Spearman (A, p. 612) stated that “Almost everyone would agree that Lean focuses primarily on efficiency” and proceeded to “equate Lean with efficiency management and regard anything that increases the efficiency of delivering products as a Lean practice.” Yet, most of the commentators (in B) took exception to equating Lean solely with efficiency. This apparent disagreement recalls the initial insight from JIT production that a simplistic equating of efficiency with maximizing output ignored losses from producing goods that were not needed or defective. Of course, the myopic pursuit of efficiency can be taken too far (Lawson, 2001), which is clearly problematic. Short-term efficiency can be improved at the expense of long-term efficiency. Indeed, efficiency has been characterized as an enigma (Kanigel, 1997). Therefore, the term “efficiency” requires disambiguation with respect to which resources are targeted and the time horizon in use. Cusumano (B, p. 628) and Netland (B, p. 636) noted Krafcik's (1988) decision to use the term “lean” rather than “fragile” to describe a key aspect of the TPS phenomenon. Dictionary definitions of the adjective “lean” invariably include terms such as “lacking” and “deficient”—terms with a negative connotation in many contexts. On the academic side, conceptions of Lean evolved toward “Lean thinking” and delivering customer value (Holweg, 2007; Womack & Jones, 2003). But when waste is defined as anything that does not add value, then the focus often still reverts to waste reduction (A, p. 612), or the removal of “non-value-adding activities.” Hence, a review of the literature (Browning & Heath, 2009, p. 25) concluded that Lean's hallmark characteristic is waste minimization, a conclusion in agreement with Cusumano (B, p. 630). Therefore, a disambiguation of the term “waste” is needed too. Despite the connotations and denotations of the term itself, many proponents of Lean do not want it to be primarily about waste minimization or efficiency but rather “value maximization,” where getting leaner may require adding rather than removing activities and buffers (Browning, 2003). The Covid-19 pandemic and other major events have stress-tested many operations and supply chains, exposing a lack of resiliency and excessive brittleness. It is not uncommon to hear Lean proponents state that adding more inventory buffers in such situations can be justified given a “true” understanding of “evolved” Lean. Justified? Probably. However, that adding buffers is offered as the way to “get Leaner” suggests that our terms and definitions have become misleading and unhelpful. The concept of value is a promising path to follow, because value is a function of both efficiency and effectiveness—that is, minimal sacrifices for maximal benefits (Browning, 2003). Ward et al. (B, p. 630) called Lean “a way of thinking about creating needed value with fewer resources and less waste”—that is, creating value more efficiently. However, efficiency is a component of value, not separate from it, because it allows stakeholders to make fewer sacrifices. The phrase “creating value more efficiently” keeps the focus on efficiency, which already contributes to determining value (along with effectiveness). The concept of value also moves us closer to the basic concept of productivity, which has some ambiguity of its own (e.g., Adler et al., 2009). Shah and Holweg (B, p. 633) quoted Ohno's expression of providing “maximum value output for minimum input,” but this is essentially the definition of maximizing productivity, a goal hardly novel or unique to the TPS. Excellent operations have always been about providing productivity and value (Browning, 2020), so these cannot be claimed as unique, distinguishing characteristics of Lean—unless, as the term “lean” connotes and denotes, the emphasis in increasing value is on improvements to efficiency. Thus, we must add “value” to the list of key terms in the vicinity of Lean that need disambiguation. Hopp and Spearman (A, p. 612) equated waste with variability and the buffers required to protect against it. In various processes and situations, some variabilities are more consequential to expected outcomes than others, which leads us to the concept of risk. In the context of product-development projects, Tyson has explored risk as a type of “anti-value”—for example, the portion of a project's value being put at risk by threatening uncertainties (Browning, 2019; Browning et al., 2002). This perspective illustrates that many types of supposedly non-value-adding activities, such as product design testing, do in fact add value by creating useful information that reduces the risk of a project not meeting its goals. Of course, whenever we incorporate uncertainty and risk into our methods, we should eventually confront the behavioral phenomena of risk attitudes and biases (some of which Hopp and Spearman mentioned in discussing their fourth lens, A, p. 620) as well as the subjective aspects of value. We also ask, “Value to whom?” The literature on Lean has tended to take a customer-centric view of value. Ideally, customers would like to receive a satisfactory product or service, immediately, for free. Firms that have approached that ideal (e.g., internet firms allowing free downloads of their software) have often gone out of business, because their lack of revenues could not cover the costs of employees, supplies, and investors—all while operating in ways that meet social and governmental expectations. A modest subset of Lean literature has explored the employee-oriented aspects of the TPS's “respect for people” (Sugimori et al., 1977). Yet, beyond customers and/or employees, enterprises must provide an appropriate balance of value to all of their stakeholders, over the long term (Browning, 2003; Browning & Honour, 2008; Martin, 2019). Given stakeholders' competing value propositions, this is much easier said than done. Yet again, this is a challenge for all enterprises and operations, Lean or otherwise.11 “In many supply chains, it may be the case that not providing protection for low probability events is optimal for profit-maximizing firms but sub-optimal for society. In such settings we have a very nuanced problem of how [to] moderate the relative weight on lean/efficiency so that supply chains serve society. Doing this in a way that preserves the benefits of competitive markets is a big challenge. But it's one we must face if we are to be better prepared for the next pandemic than we were for this one.” (Hopp 2021) Many OM textbooks discuss the concept of value well before they discuss Lean. Providing value to all stakeholders is again a general OM concept, not one that distinguishes Lean, and not even one that Lean has addressed broadly enough. Beyond production contexts, in projects, during his time working in industry, Tyson witnessed problems caused by a reductionist view of Lean (vs. a system view, as discussed below), including passionate debates about which activities were “value-adding.” This led him to realize that value is not an intrinsic property of individual activities but rather an emergent property of an overall system of actions and interactions (i.e., process): A value-adding activity fed bad inputs will produce bad outputs. This situation is especially acute in projects, where many activities require only information as inputs—information that may be proxied by assumptions (albeit with the risk that they are imperfect). If these inputs are later invalidated, what happens to the activity's value added? Tyson concluded: “In many cases, lack of value stems less from doing unnecessary activities and more from doing necessary activities with the wrong information (and then having to redo them)” (Browning, 2003, p. 51, emphasis in original). The actual value is provided by an activity's output—products, services, deliverables, or information—not just its execution, and it is a function of the quality of the inputs—products, services, deliverables, information, or assumptions—used to create it. Thus, value depends on not only value-adding actions but also their network of interactions (Spear & Bowen, 1999)—that is, the overall process. Viewing operations as processes and systems is therefore essential, especially at the higher levels of organizations (Browning, 2020; Browning et al., 2006), if we are truly to move toward operational excellence. In many practical situations, however, the path to process improvement illuminated by Lean seems to orient around the question, “What current activities can we stop doing?” (Browning, 2003). Taking an activity-level view of value has prompted short-sighted, non-systems-thinking organizations to take shortcuts to waste reduction—what Tyson has called “liposuction” (i.e., “cutting out the fat,” the non-value-adding activities), resulting in “emaciation” rather than Lean—but nonetheless prompted and driven by Lean's connotations (Browning, 2003; Browning & Sanders, 2012). Several studies have identified occasions where organizations' use of Lean as a unifying term for its operational improvements has pushed those efforts in unhelpful directions by overemphasizing certain dogmas, heuristics, or practices. Taken too far, even Lean itself can become wasteful (e.g., Browning & Heath, 2009; Cusumano, 1994; Cusumano & Nobeoka, 1998). Health comes rather from the proper use of diet and exercise to balance muscle with healthy fat. Hopp and Spearman's lenses of Lean provide useful perspective on what it means to manage an operation effectively, regardless of whether the term Lean applies. Their Process Lens could be renamed the “Activity Lens,” however, because a process is already defined as a network or system of activities. Hopp and Spearman essentially admit as much when stating, “this lens is less helpful in identifying waste that propagates from other parts of the process” (A, p. 614)—that is, that the Process Lens itself does not even account for the entire process. Then, their Network Lens might just as well be called the “Process Lens,” because it emphasizes the importance of systems thinking, thereby connecting to the works of Goldratt (e.g., 2004), where eliminating waste (e.g., excess capacity) at non-bottleneck activities actually makes the overall process less productive. It is just these types of counter-intuitive results that justify the system/process perspective rather than the reductionist, individual activity-by-activity perspective—but these concepts predate the term Lean. In production, whether the term Lean has a positive or negative connotation depends on the context: Variability needs to be buffered, either with inventory or capacity. The insight that underlies Lean is that by eliminating variability, one can eliminate the need for buffers. This is exemplified by our discussion in Section 3.1 comparing the use of inventory reduction under JIT to identify a bottleneck through the starvation or blocking that results, to the use of inventory to minimize a bottleneck's being blocked or starved under Goldratt's theory of constraints (TOC) (Goldratt & Cox, 2004). Under TOC, bottlenecks are typically identified not by first reducing in-process inventories but rather by observing a tell-tale pile of inventory awaiting processing. Thus, the buffer recommended by the TOC approach to avoid a loss of bottleneck capacity might well be (mis)categorized as waste through a JIT lens. Initially, Lean was “explicitly derived from” and “equated … with” the TPS (A, p. 610). Shah and Holweg (B, p. 633) called “the underlying philosophy of the TPS” the “True North” of Lean—the guidance for defining all truly Lean concepts. Netland (B, p. 635) described Lean as once “synonymous with the TPS” and derived by observing it as a phenomenon. The IMVP saw the TPS more generally as a “multifaceted phenomenon” (B, p. 630), expanding well beyond Krafcik's (1988) use of the term Lean. Womack et al. (1990) described five facets of the TPS (fulfillment, supplier management, customer management, product development, and management) that went beyond the management of the buffering-variability tradeoff in production. Other early work on the TPS highlighted other aspects such as “respect for people” (Sugimori et al., 1977), although this aspect did not initially receive the same emphasis outside Toyota. Yet, even though it captures only one aspect of the TPS, the term Lean became the descriptor for its entirety and all that has evolved from it in the ensuing decades. These rules typify the TPS but not post-TPS Lean. Hence, at what point are variants best understood away from a conceptual attachment to Lean, rather as stand-alone theories, principles, practices, and/or tools for better OM at particular times? The term Lean was originally deployed to capture a reduction in variability that permits reduced buffering. This objective and the practices deployed to achieve it are described by Spear and Bowen (1999, p. 104) as temporary “countermeasures.” Lean used in this original sense may well be inappropriate for a situation in which variability is a source of competitive advantage or profit, or when resilience is a priority. Productivity has always been easier to improve under conditions of stability, predictability, and certainty. Firms that can achieve these conditions inside of protected bubbles (such as the time horizons of heijunka) benefit the most from Lean. But the majority of the world includes instability, unpredictability, and uncertainty—in supply, demand, and how best to connect them, both currently and in the longer term. Variability of various types that cannot be removed must be buffered. One way or another, variability adds costs. Those costs may show up in different contexts as buffers of capacity or inventory; the purchasing of options or insurance, or hedging; the transfer of costs (deliberately or unwittingly) to other stakeholders; and/or the costs of worker and managerial attention. Are such costs worth their benefits? That is, do such actions and buffers add value? Buffering that appropriately accommodates strategic variability and provides resilience can be considered as “healthy fat” that allows the operation to satisfy customers and achieve other strategic objectives. However, if we allow Lean to evolve to the point that any measure that improves operational performance makes it Leaner—where even adding fat increases leanness—we arrive at an obvious contradiction in terms (at least to most people, though not to all academicians). Thus, we must add Lean to the list of terms requiring disambiguation. The more clearly a definition delineates the concept from seemingly similar concepts, the better the definition is. In operations management, the differences between many of the strategic programs are not readily apparent. For example, TQM, JIT, lean manufacturing, and continuous improvement share many concepts. The testing of how much each concept adds to the explanation depends upon exactly how each term is uniquely conceptually defined. Overlap in the definitions causes any theory using these terms to be vague and/or ambiguous causing all empirical results to be highly questionable. Put more formally, the connotations of the defined terms should match with their denotations (Bunge, 1967). For example, consider the term ‘philosophy’ sometimes used in production/operations management definitions (JIT philosophy, TQM philosophy, etc.). The term ‘philosophy’ broadens the definition's connotations to include unspecified properties, while leaving the denotation unchanged. … This broadening is called ‘concept stretching’ (Osigweh, 1989). In operations management, there are many examples of expansion of definitions. One such example is ‘big’ just-in-time and ‘little’ just-in-time. ‘Little’ just-in-time focuses ‘more narrowly on scheduling goods inventories and providing service response where and when needed’ (Chase et al., 1998, p. 324). On the other hand, ‘Big JIT (often called lean production) is the philosophy of production operations management that seeks to eliminate waste in all aspects of the firm's production activities, human relationship, vendor relationship, technology, and management of materials and inventories’ (Chase et al., 1998, p. 324). Although both definitions have some clarity difficulties, clearly the second definition does not prohibit any non-value adding activities. Consequently, if any organizational activity reduces waste regardless of the specific cause, it is by definition within the ‘philosophy of big JIT’. With big JIT, anything ‘good’ that reduces waste is big JIT and anything ‘bad’ that increases waste is not in the philosophy of big JIT. Consequently, big JIT claims all that is ‘good’ and claims all that is ‘bad’ is not big JIT without giving the specific relationships that caused the reductions or increases in waste. It cannot be considered a ‘good’ definition for theory since it violates both the uniqueness and the falsifiability virtues of ‘good’ theory. Additionally, it violates the internal consistency virtue since it does not specify ‘how’ and ‘why’ big JIT causes waste reduction. Our intent is to put Wacker's recommendations into practice. We first ask: What is Lean? A phenomenon? An ideal? A philosophy? A way of thinking? A collection of practices? A strategy? Widespread agreement is lacking. Netland called Lean a “broader strategy of how to manage operations” (B, p. 636). Ward et al. (B, p. 631) noted that Lean enterprises should be “highly adaptive,” and that the application of Lean practices and tools without “Lean thinking” is a likely explanation for failed cases. Probably so… but Lean is an unnecessary term here. All enterprises should be appropriately adaptive, and the misapplication of any OM concepts, methods, or tools is a likely explanation for failed cases. Shah and Holweg (B, p. 634) noted the tendency of Lean itself to adapt, particularly to applications beyond manufacturing, while maintaining an underlying philosophy described essentially as productivity. But productivity, properly defined, has always been a focus of OM. And once we detach Lean from the TPS, as Shah and Holweg (B, p. 635) suggested (focusing instead on, e.g., Tesla, or any other popular phenomenon), must we still call it Lean? Is any increase in operational productivity attributable to Lean? Questions about the nature and limitations of Lean are important. While we appreciate their efforts to be pragmatic, we do not agree that “any definition of Lean that elevates practice or deepens our understanding of how to elevate practice is good” (A, p. 611), because, while elevating practice and deepening our understanding are obviously good, putting the Lean label on all such efforts merely causes confusion. Yes, Lean has evolved far beyond the TPS, but at what point in that evolution does Lean outgrow its name? We start our exploration by considering three alternatives to the TPS: TOC (Goldratt & Cox, 2004), seru (Yin et al., 2017), and Quick Response Manufacturing (Suri, 1998). These alternatives show that the pre-Lean TPS was not expected to cover the entire space of operational excellence. While the TPS was arriving to the West, other production systems also emerged, some of which were generally considered to be TPS competitors. Flexible manufacturing systems were receiving considerable attention. Cellular manufacturing was considered by companies such as Kone as a non-TPS way to achieve JIT objectives with respect to flow, given the general view that the TPS applied to assembly lines (de Treville, 1987). That Goldratt's TOC—as described in The Goal (2004), originally published in 1984—proposed an alternative to the TPS can be seen in comparing the use of exploratory stress. In contrast to JIT's systematic removal of in-process inventories to encourage line balancing and other types of process improvement, the TOC calls for a systematic reduction of batch sizes to reveal bottlenecks.22 Note as well that batch-size reduction under the TPS begins from system-wide reduction of setup times and has the objective of reducing in-process inventory, in contrast to batch-size reduction under TOC that highlights which setup times need to be reduced with the objective of reducing flow time. Both approaches are built on the same system dynamics but use different protocols and logic. Managers can choose to alleviate these bottlenecks by reducing setup times or adding capacity. For bottlenecks that have been identified, the objective is to use buffer inventories to prevent their being blocked and starved. The same system dynamics underlie the two uses of exploratory stress, but the actions taken differ. Note that a TOC implementer may well turn to TPS theory on how to reduce setup times on an identified bottleneck—even though TPS calls for reduction of all setup times, without prioritizing bottlenecks. More generally, TOC builds around bottlenecks, whereas TPS seeks to eliminate them as a step toward balancing capacity in the line. The Goal (2004), however, shows Jonah claiming that such increases in capacity balance bring a plant closer to bankruptcy, directly challenging TPS objectives. In the 1990s Sony and Canon sought advice from a TPS expert in Japan on how to adjust the TPS in the face of rapid product proliferation and highly volatile consumer demand. It was clear to these companies that the TPS would leave them without the required responsiveness. As described by Yin et al. (2017), the TPS expert advised them that the TPS was not the right system for their context. He instead created an alternative approach called seru (the Japanese word for biological cell). Under seru, assembly, testing, and packaging are carried out on small, general-purpose machines on wheels. Cells are configured as needed to produce orders. Yin et al. (2017) described the replacement of large, automated machines (highly efficient under high-volume production) with general-purpose machines, and how the increase in responsiveness caused by this apparent loss of efficiency led to a dramatic increase in profit while supporting innovation. This powerful intervention started from a clear departure from the TPS. Which approach is Leaner? Attempting to answer that question distracts from the more fundamental tradeoff to be explored between the increase in output for given capital investment provided by automation and responsiveness. Both TOC and seru offer advantages for products that are engineered to order, in contrast to the highly repetitive operations targeted by the TPS—and made more so through practices like heijunka that level demand. Suri (1998, pp. 227–243) argued that the production of engineered-to-order products fits poorly with a typical JIT system. He began by establishing that the “pull system” claimed by JIT (and foundational to the TPS) is more accurately described as a push system with a mechanism in place (kanban) to limit the buildup of in-process inventories. For example, suppose that there are five kanbans between two adjacent workstations. The upstream workstation is permitted to start production as soon a