Evolutionary systems: modelling organisational innovation.

SUMMARY

The aim of this paper is to present the results of combining two complementary areas of research. The first area is manufacturing cladistics, a classification system where best practice can be mapped allowing organisations to locate their position in evolution, the position of competitors and the chance to re-engineer their organisation using the classification as a guide. The second area is evolutionary systems modelling, which has successfully been applied to ecosystems, urban systems, industrial networks, economics and financial markets. Using an evolutionary framework, designed to model through simulation the evolution of manufacturing form, new structural organisations may be explored. In addition, organisational innovation and the consequences of particular decisions in the context of introducing new technologies and practices to the existing structure may also be investigated.

KEY WORDS

evolutionary systems modelling; evolutionary systems simulation; manufacturing cladistics; organisational structures; organisational innovation; management commitment

INTRODUCTION

The results of a recent project are now beginning to show promise in at least providing an interesting insight into the processes involved in organisational innovation. This research successfully attempted to combine two, previously unrelated, approaches both of which share the theme of evolution. One approach is manufacturing cladistics an evolutionary classification scheme from the biological sciences. The other is evolutionary systems modelling, a quantitative approach from the physical sciences. One set of results, investigating organisational transformations and change initiatives, has been reported in a recent issue of this journal (see Baldwin et al., 2003). Therefore, as the background for this research has already been given in that paper, only a summary will be given here.

Manufacturing cladistics was developed by McCarthy and colleagues (McCarthy et al., 1997; McCarthy & Ridgway, 2000) with a two-fold purpose:

1. As an objective and unique classification scheme as evolution is the reference point and external to the observer; and,

2. As a decision-support tool to aid manufacturers in organisational change initiatives and transformations.

Several cladistic classifications for manufacturing are now available (e.g. Goh, 2000; Leseure, 2000; McCarthy et al., 2000) and the application of this approach legitimised. Figure 1 and Table 1 represent an example of one classification scheme of the evolution of automotive assembly plants based on data from the International Motor Vehicle Program (documented by Womack, Jones & Roos, 1990), surveys completed by car manufacturers and historical records (for a more in depth discussion on the characteristics listed in Table 1 and the organisational forms and their significance, see Baldwin et al., 2005; Bessant et al., 2001; Brown & Bessant, 2003; McCarthy et al., ; Womack et al., 1990).

[FIGURE 1 OMITTED]

However, as this is essentially a post-hoc analysis, limitations with this approach become apparent when there is a need to look to future scenarios so that risk and uncertainty can be reduced. This is true for explorations of unknown evolutionary trajectories particularly in terms of the implementation of innovative technologies and practices. Although still useful for manufacturers playing catch-up, its use for more advanced manufacturing systems, looking for novel organisational structures for competitive advantage, is less appealing. Nonetheless, the solution to this problem is arguably derived when applying evolutionary systems modelling as both share common principles. Furthermore, as evolutionary systems modelling has suffered in the past from a lack of simplified and descriptive visualisations, both approaches are highly complimentary and beneficial.

Evolutionary systems theory (Allen, 1976, 1997, 2001; Allen & McGlade, 1986) is a school of thought, accompanied with modelling methodology, from the science of complex systems and originates from Prigogine's (1973) Nobel Prize winning research. This school asserts that all models of systems have certain underlying assumptions creating a hierarchy of models from prediction and certainty through to exploration and potentialities (Allen, 1992, 1998a, 1998b). Two assumptions are essential for all system models--that there is a boundary between the system of interest and its environment, and that the system's components can be classified resulting in a taxonomy. In addition to these, assumptions are made concerning components and their interactions. When components and interactions are assumed to be of an average type, i.e. normally distributed around the mean, as with system dynamic models, there is only one future path, which is the most probable. The models appear to have complete understanding and knowledge and can therefore give perfect predictions.

However, when interactions are assumed to be of a non-average type, the model begins to change from prediction to exploration. By introducing diversity, in that all potential types of interactions are accounted for, many potential future states may be reached through the exploration of self-organisational processes. Nonetheless, there are criticisms of this approach, particularly when applied to social systems. The first is that the diversity of interactions is usually represented by 'noise', a stochastic mechanism in the equations, and as such are somewhat removed from reality. The second criticism is that although the interactions are treated as non-average, the components are not.

It is this treatment of the components that separates self-organisational models from evolutionary models (Allen, Strathern & Baldwin, 2004). By also treating the components as non-average, which introduces internal or micro-diversity, a more accurate representation is given of real evolutionary change. Whereas the former represents blind adaptation, the latter, by introducing micro-diversity, represents co-evolution through experience and learning. The 'means' and the 'end' are transitory and in continual paradoxical dialogue through feedback (Allen & Ebeling, 1983). With this treatment there is also devolution of control from the global (system) level to the local situation and is a manifestation of the relative performance of individual behaviours. Furthermore, as evolution is an open process, involving both chance and determinism, there is an inevitable degree of 'error making' or the imperfect transfer of information (Allen & McGlade, 1986). This is a necessary requirement, however, and creates the forum for learning through the continual exploration of behaviour space (for a more in-depth discussion of the hierarchy of models with examples refer to Allen, 1998a; Baldwin et al., 2005; Baldwin et al., 2004).

To date, two aspects of the results highlighting the benefits of this research synthesis have been reported. Organisational transformations were investigated in an earlier issue of this journal (i.e. Baldwin et al., 2003). The Modern Mass Producer (see Figure 1) was taken as the case study to which bundles of practices (from Table 1) were introduced and the consequences of such introductions explored. The second set of results (Baldwin et al., 2005) simulated the full evolution of the automotive industry, from the Modern Craft System to the Agile Producer, according to the cladogram in Figure 1. Issues such as organisational culture, psychological barriers, risk and uncertainty in decision-making, unpredictability and the limits to foresight and planning were highlighted in both these studies. The set of results presented here have an altogether emphasis and were not expected at the outset of the study. Altogether different issues, such as organisational innovation, management commitment, organisational maturation and stagnancy, the impact of management fads and fashions, and internal synergies and symbioses, come to the forefront.

[FIGURE 1 OMITTED]

METHOD

The full details of the methodology have been described elsewhere (Baldwin, 2004; Baldwin et al., 2003), so will only be summarised here. The aim of the study was to use the evolutionary systems model, with the least assumptions, to model the cladistic evolution of the automotive industry (see Figure 1) as described by McCarthy, et al. (i.e. 1997, 2000; McCarthy & Ridgway, 2000). The synthesis was based on the automotive manufacturing cladistics research due to several reasons. The first was because the history of the automotive industry is well known and publicised through both academic and trade publications. The factory layout, worker organisation and the practices employed are also well known with many pervasive in all manufacturing. The second reason was that the actual cladogram was attractive and logical and with 53 characteristics (see Table 1) and 16 organisational forms (see Figure 1), the data collected would prove to be rich. The third reason is that the automotive cladogram has been published several times (McCarthy et al., 1997, 2000; McCarthy & Ridgway, 2000) and is easily accessible.

Seventy-three completed questionnaires, the total returned from 1565 manufacturing organisations, gauging the positive, neutral or negative interactions of the 53 characteristics detailed in the automotive cladistics research, constituted the dataset (see Table 1). The simulation model, based on the equations given in Allen (1976, 1984) and Allen, et al. (1985), was developed in Turbo Basic and run in the Microsoft Dos operating system.