A Behavioral Perspective of Replication in Complex Environments

Replicating organizational routines, business models, and best practices is an important strategy to spur organizational growth and performance improvement (Nelson and Winter 1982; Teece et al. 1997; Rivkin 2000; Winter and Szulanski 2001; Brusoni et al. 2009, 2011; Guettel and Kronlechner 2009, Guettel et al. 2012; Winter et al. 2012). Yet, in replicating successful practices, firms face three major challenges. First, complete and accurate replication is often hard if not impossible to achieve (Szulanski and Winter 2002, Brusoni et al. 2009). This raises the question: which elements of a successful practice should firms seek to replicate as accurate and complete as possible? Second, identifying which practices are actually “best practices” and are worth to replicated is far from trivial (Denrell et al. 2004). This raises the question: what are the consequences of inaccurately replicating practices that are not literally best practices, i.e. practices that cannot be improved anymore? Third, in many cases, the value of practice is highly context dependent (Williams 2007, Guettel et al. 2012) and a replication template may have to be adapted to the local environment. This raises the question: If a template is applied to a different context, how much of it should be transferred? Building on Rivkin (2000, 2001), we will seek to answer these three research questions in a series of experiments with an NK simulation model (Levinthal 1997). Thereby, we can draw on our own substantial experience with this model (e.g. Posen, Martignoni, and Lang, 2012; Dietl et al. 2012a,b; Posen, Martignoni, Lang 2013) as well as our extensive research network that also includes, for example, Dan Levinthal from the University of Pennsylvania (Posen, Martignoni, and Levinthal 2012), Hart Posen from the University of Wisconsin and Nicolaj Siggelkow from the University of Pennsylvania (Martignoni and Siggelkow 2012). For the three subprojects, we have already run some first experiments and already discovered some first very interesting results. For example, in Subproject 2, we find that if replication is imperfect, firms are – on average – better off if they seek to replicate non-optimal practices (local peak) rather than literally a best practice (global peak). Like our own related prior research, we intend to publish our findings in top academic management journals, such as Management Science, Administrative Science Quarterly, Organization Science, or Strategic Management Journal. Given the importance of this topic, it is not surprising that there are already some important prior modeling efforts on imperfect imitation and replication using the NK performance landscape model (e.g. Rivkin 2000, 2001, Csaszar and Siggelkow 2010). In our project, we seek to further extend this line of inquiry, in particular on the mechanisms through which inaccurate and incomplete replication processes may affect organizational performance. Understanding these mechanisms is particularly important. While there is strong consensus that perfect replication is hard if not if not impossible to achieve (Szulanski and Winter 2002), there is much less agreement on the performance implications of inaccuracies in the replication process (Winter and Szulanski 2001; Szulanski and Jensen 2004). The implications of this project extend beyond replication processes. While we frame our discussion around the intra-firm transfer of practices (=replication), our insights are also relevant to inter-firm transfers of practices (=imitation). With imitation, similar questions arise: How much should a firm imitate? Whom should it imitate? Which elements should it imitate?