Organizational Learning

Contributed by Olav Sorenson

Organizational learning refers to the idea that organizations improve in their ability to perform some operation with experience. For example, over time, firms generally see a steady decline in their costs when they continue to produce the same product. Underlying this macro-level pattern is a micro-level foundation of variation, selection and retention. Firms experiment with new ways of doing things, either intentionally or unintentionally. Most of these experiments fail, but a few of them prove useful. Firms attempt to identify these better ways of doing things and incorporate them into their standard operating procedures. Because firms improve their performance with experience, these dynamics create a kind of positive feedback loop that endows early movers with an advantage with respect to performance.

The literature on organizational learning covers a diverse array of ideas that go well beyond even the optional readings listed above. However, the two most prominent lines of thought are probably learning curves (economies of experience) and the choice of exploration versus exploitation.

The literature on learning curves has been most active in economics and reflects the idea that firms improve their performance on some task as they do more and more of it, so-called learning-by-doing. In this literature, the researchers do not observe variation, selection and retention; they simply assume that it produces cost curves that decline at a declining rate with cumulative production (or performance curves that increase at a declining rate with production).

Much of the early research focused on the production of military hardware during World War II because the U.S. government had all manufacturers use the same production plans and paid cost-plus rates for their output. As a result, researchers did not need to worry about how differentiation or pricing might influence performance. Rapping (1965), for example, estimated the slopes of these experience curves for the production of aircraft.

More recently, researchers have been exploring a number of interesting extensions to this baseline expectation, extensions that allow different firms to experience different curves depending on their organizational structure, their portfolio of products and other dimensions of firm heterogeneity. From the perspective of strategy, these extensions obviously have greater relevance to the choices that firms face because they relate the steepness of the learning curve to organizational design, product positioning and geographic location choices.

With respect to geographic location choice, Darr, Argote and Epple (1995), for example, studied pizza restaurants and demonstrated that learning-by-doing appears to spill over to other restaurants in the region, particularly those with connections to each other (e.g., through common ownership). Presumably, firms communicate the information that they gain through experimentation to one another, either actively or perhaps passively through the movement of employees across firms. Several studies since then have found similar results in the operation of hotels, in the production of footwear and Liberty ships and in the provision of other goods and services (e.g., Baum and Ingram, 1998; Schilling, et al, 2003, examine similar issues through an experimental design). Learning-by-doing, therefore, may contribute to agglomeration externalities (see Alcacer and Oxley reading list).

Another topic has been how the strategy and structure of the organization influences the efficiency with which it can learn. Sorenson (2003), for instance, discusses how the degree of interdependence between the processes involved in production can inhibit the firm’s ability to isolate and implement more effective routines. Consistent with this perspective, he finds that vertically integrated firms learn more slowly, particularly in stable environments. Following a similar logic, one might also expect multi-unit and multi-product firms to learn faster because they can engage in parallel experimentation (for evidence, see Baum and Ingram, 1998, and Sorenson, 2000). Firms might also vary in systematic ways in their ability to retain knowledge (Thompson, 2007). Managers, therefore, must consider not just the static implications of their choices but also their implications for learning and therefore for the longer run.

Despite the wealth of empirical evidence for these learning curves, these studies are not without their problems. Other unobserved factors that vary over time potentially threaten the validity of interpreting these curves as learning. Thompson (2001), for example, gathered data on the capital investments of Liberty shipbuilders, which also increases monotonically over a firm’s life, and demonstrated that investment can account for nearly half of the apparent “learning” of the shipbuilders.

The other main line of research considers the opportunity costs to learning-by-doing. Yes, firms can continue to get better and better at some area of expertise, but over time this inertia tends to result in a mismatch between the firm and the environment. A firm could become very good at doing something that the market no longer values, something that March (1991) describes as a “competency trap” (see also Denrell and March, 2001). A fundamental tension therefore exists between whether the firm builds on its strengths versus whether it explores the environment for alternatives that might have greater appeal. Much research has sought to relate this tension to policy choices at the level of the firm (e.g., Sorenson, 2000, connects it to the breadth of a firm’s product line).