Seeing trees and forests: a comparative evaluation of business clusters and national industry associations in the New Zealand forest sector.

In contrast, clusters can be presented as inclusive groupings. A 'marshallian' conception of clusters envisages that cluster membership is purely a product of geographical location within the cluster 'ecosystem' (McCann, 2001). This implies that clusters exist without any formal membership. Some go further and suggest that clusters are self-organizing entities that do not require and are largely unresponsive to efforts to deliberately mould the behaviour of individual participants such as through a membership association (Porter, 2000). Similarly, the untraded dependencies that promote localized learning through informal channels and the operation of external labour markets have been thought of as the basis for cluster 'membership' (Malmberg et al., 2000; Leamer and Storper, 2001; Pinch et al., 2003). On the other hand, long term development of a cluster is generally thought to involve more than geographical proximity. Porter (1998) has, for example, identified a cluster of 400 medical device companies in Massachusetts, employing close to 40,000 workers that lay dormant until revealed through a search for potential clusters. Once revealed, he reports that business executives came together to consciously exploit the advantage of belonging to a cluster. The Schmitz (1995) model of business cluster development proposes that deliberate planned action by selective groups of cluster firms is necessary to sustain the advantages of a cluster. One important form of deliberate planned action is membership of a trade association that assists technology transfer or collective investment in shared infrastructure (Schmitz and Nadvi, 1999).

Consequently, industry networks and business clusters have been viewed as variations on a theme rather than being wholly different species (Rosenfeld, 2005). Both have been linked to the 'associative economy' in which businesses are thought to gain by making more use of the resources of other businesses and support agencies than in the past. In this context, clusters and industry associations might be mutually supportive. A cluster may not require membership of a formal association but it might be an outcome that enables a geographic concentration of enterprises to improve the services obtained from a national association. Indeed one survey of business clusters around the world found that 89 percent of all cluster initiatives have a facilitator to manage activity, most of which do this at least part time from an office (Solvell et al., 2003). Once organised into some form of membership association, pressures to balance activity and participation will exist for national associations and cluster groups.

The capacity to incorporate value chain representation may be identified as a point of difference between clusters and industry associations. The ability to induce co-location of different components of a value chain is sometimes identified as the particular strength of an enterprise cluster. Indeed the incentive for specialist suppliers to join a cluster was identified among Marshall's original advantages accruing to a cluster (McCann, 2001: 57). Modern day cluster mapping uses input-output linkages to delineate clusters as well as the concentration of same industry activity (Feser and Luger, 2003; Porter, 2003). In contrast, industry associations can be limited to a single activity and may exist expressly to support the interests of one stage of the industry against another.

Value chain integration links to potential differences in origin between clusters and industry associations. Geographical concentration can be associated with the centrality to a focal organisation such as a skill development agency, research centre or large enterprise that 'spins out' new start up companies (Colombo and Delmastro, 2003). These processes can lead to the concentration of industry know how that becomes a stimulus for innovation and market growth. Such a role of clusters is potentially important as most innovation relies on interactions among various actors rather than the creativity of lone inventors (von Hippel, 1988; Lundvall, 1992; Baptista, 1996). Similarly, long association with a particular industry can give clusters a marketing brand that gives an edge in international markets and sustains cooperation to maintain the cluster's advantage (Merrilees et al., 2006). Such business development advantages seem less likely to explain the origin of an industry association although they can play an important role in promoting industry standards around a source of national technological advantage (Bennett, 1998; Bresnahan et al., 2001). As well, not all clusters are associated with learning communities or a distinct marketing brand. In fact, it has been argued that 'physical clustering' (where businesses locate in proximity to each other without functional connections or any special advantage being obtained) is particularly prevalent among clusters of advanced technology activity (Hendry et al., 2000; Oakey et al., 2001).

Many enterprise clusters are based on final assembly rather than complete value chains. For example, industry districts in Italy frequently comprise concentrations of near identical types of enterprise. Those districts based on an industry specialisation and associated ancillary activities are a sub category rather than the norm (Paniccia, 2002). In the USA, it has been shown that concentrations of footwear production are independent of the location of raw material or equipment suppliers and of the distribution of buyers (Sorenson and Audia, 2000). At the same time, industry associations may facilitate participation from supporting sectors through mechanisms such as associate memberships or by setting up pan industry groups. For example, in New Zealand a recent development has been the establishment of such a pan industry group in the forestry sector ('WoodCo') to integrate the marketing strategies of individual national associations.

There is much, therefore, to support the contention that clusters and national industry associations are variants rather than wholly different species. Nonetheless, differences may develop from the national versus local focus (Table 1). To the extent that government actions tend to have nationwide impact and are potentially more susceptible to influence by a national than local group, lobbying may be a more significant function of national rather than local groups. Countervailing influences may be that national regulation has a regional dimension and that agreement over preferred actions is easier to obtain among a local than national group. Assuming that lobbying is more a function of national associations than regional clusters, this may mean that the public good nature of collective activity is more of a constraint on national associations than cluster groups.

As well as lobbying, national associations may be better placed to supply specialised business services to members than is a cluster group. National associations can collect industry wide data and use this as the basis for informed advice or other services to individual members (Bennett, 1998). Similarly, the development of product standards, codes of conduct and industry marketing programmes that target all industry members are tasks that are potentially more suited to a national than local group. As well, to the extent that a national association is larger and better resourced than a local group, it may have capacity to employ full time business advisers and project officers. On the other hand, the possibility of addressing industry wide issues may bring greater difficulty in obtaining agreement over priorities. In this sense, the more restricted agenda open to a local cluster group may be an advantage in focussing attention on issues where there is broad agreement among enterprises.

The ability to sustain member participation can offset some of the opportunities existing for national associations. Partly for practical reasons, a locally based group may have a more actively engaged membership than a national association. Given that participation is generally a voluntary commitment on top of full time management responsibilities, savings in time and cost attending local meetings over this required to participate in a national group may of itself encourage involvement. Indeed the potential for intensified face-to-face interactions, short cognitive distance, common language, trustful relations, easy observations and immediate comparisons have been held to be the unique features of clusters (Malmberg and Maskell, 2002). In contrast, national associations may operate with one-way communication from association staff and little member input into the direction that the association chooses to take in its activities (Enderwick and Wilson, 1992). Alternatively, involvement can be limited to large, well resourced organisations motivated by an agenda that differs from the priorities of small enterprise (Semlinger, 1995: 23-24). Countervailing possibilities may complicate whether national associations are distinguished by their relative lack of participation. Having a restricted range of active participants may bring greater agreement and commitment among those who do participate. Consequently, while local groups may in theory have a more diversified membership than a national group this can be offset by the difficulty in obtaining agreement around priorities among a diverse group of enterprises. In New Zealand, this has resulted in some cluster groups setting membership rules to limit participation to a like group of organisations (Perry, 2004).