The cluster concept in context

Are clusters purely theoretical constructs created for analytical purposes or is it possible to identify them empirically according to recognizable dimensions such as key actors, product specialization, company size, target customers, and the like? Porter’s definition, which is cited in several of his works, is of limited use in this regard, as it explicitly aims to remain as broad as possible. In more recent contributions, Porter has even added that “there is no model for clusters, but a multitude of configurations reflecting the particular circumstances of a location and a set of industries” (M. Porter & Ketels, 2009, 174). And yet Porter does point at three major dimensions that can be used to identify specific clusters: (i) the geographical focus, or the spatial scope of externalities produced by proximity; (ii) the business environment, namely the conditions created by the actions of companies or governments and by the existing institutional sphere; and (iii) specific activities, namely the product specialization or the market of reference common to the companies constituting the cluster.

95

The case of palm oil shows how this definition is at once called into question and enriched when it is applied to a specific context. First, rather surprisingly, the rubber or palm oil businesses are almost never conceptualized as clusters (or other forms of industrial concentration) either in historical sources and historiography of the industry or in the cluster literature. The notable exceptions to this are a technical contribution on palm oil in Indonesia in the last decade (Pahan, Gumbira-Sa'id, & Tambunan, 2012) and a market report on palm oil in Malaysia (Belai, Boakye, Vrakas, & Wasswa, 2011). However, the two countries are never considered as part of a single cluster. Second, when Porter’s categories are applied to the palm oil cluster, they immediately reveal some operational limitations.

Identifying Porter’s first dimension, namely the geography of the cluster, is a relatively easy task in the case of palm oil, because the crop can be grown only at specific latitudes and in special climatic and soil conditions (see Figure 2). This means that the geographical scope of the cluster can be defined by the locations where the crop can be grown, that is to say in the territories lying between ten degrees of latitude north and south of the equator. The thesis identifies two major cluster locations: Southeast Asia and West Africa. In both regions, externalities due to proximity can be found for the production of palm oil, as Porter’s cluster theory would predict.

However, both clusters are located across different countries: the Southeast Asian cluster includes territories belonging to the Malay Peninsula and DEI, subsequently Malaysia, Singapore, and Indonesia. The West African cluster is even more diverse, comprising several former British colonial possessions in West Africa, such as Nigeria, Cameroon, and Sierra Leone, as well as the former Belgian colony of Congo. This makes it potentially difficult to identify the business environment (in Porter’s terms) as these locations were under different colonial administrations and differed in their institutional structure. At least from a formal perspective, within each territory, companies faced different labor markets, land regulation, and administrative arrangements.

96

Nevertheless, my sources indicate that foreign investors perceived the business environments as homogenous within each of the two cluster locations. At the same time, they recognized the differences between the two set-ups: wild palm groves in Africa versus plantations in Southeast Asia, the sticky labor markets in Africa versus the availability of cheap labor, and the complex African farmers’ land tenure system versus easier access to land grants in Asia. As mentioned above, the correspondence between different palm oil stakeholders (researchers, government officials, and businessmen) and the reports on the industry always refer to the two major macro locations of West Africa and Southeast Asia, each stretching across several neighboring countries. Hence, the cluster structure and related business environment created a perceived institutional homogeneity that spanned different colonial institutional frameworks.

Finally, the palm oil cluster problematizes Porter’s activity dimension. While this study explicitly focuses on foreign companies producing palm oil, in Southeast Asia all palm oil players started off as rubber plantation companies before introducing palm oil to their estates and continued to produce rubber even when palm oil became increasingly dominant.

The same is true of the smallholding sector, which developed out of rubber production and adopted palm oil only after the crop had established itself in the estate sector. As a consequence, the palm oil cluster can be considered either a subsection of the larger plantation cluster, involving companies and smallholders devoted to growing agricultural commodities for the international market, or, from a narrower perspective, a “spin-off” of the dominant rubber cluster, which accrued its own rationales separate from the rubber business. My analysis shows that both conceptualizations are valid but that they refer to different moments in time. The former, interpreting the palm oil cluster as a subsection of the plantation cluster, is generally applicable to the first period of development during colonial times, as agency houses and plantation companies normally had only a minor interest in palm oil. The second interpretation of palm oil, according to which it is a spin-off, better describes the cluster from the 1940s on and, to an even greater extent, beginning

97

in the 1950s, when the crop became an increasingly important player in the vegetable oil market and the agency houses started to invest significantly in its production, financing bulking facilities and research programs and organizing their interest through formal institutions such as the Palm Oil Selling Pool.

Therefore, while remaining deeply connected and often influenced by the developments of the rubber business, especially in Southeast Asia, increased specialization implied the emergence of a distinct cluster with its own logic and internal dynamics. As illustrated in particular in my third paper, a comparison with the African palm oil production strengthens the conception of palm oil as being autonomous from rubber. First, the rivalry with competing West African locations fostered a shift in perception towards palm oil as an independent cluster. The fact that, during the interwar period, West African experts identified the Eastern colonies as a threat to the profitability of the local producers, sending their own experts to study the plantation model, contributed to the idea of palm oil as a promising new line of business in the minds of the agency houses. Second, the development of palm oil was characterized by a continuous interaction with its native location (Africa) for research purposes, which did not happen as extensively in the case of rubber. Finally, there is good reason to see palm oil and natural rubber productions conceptually as two different clusters because they belonged to two differently shaped global value chains.

While the rubber value chain was more fragmented at the buyer level and hence more producer-driven, the palm oil cluster was dominated by one major buyer: the multinational Unilever.