|
Of Saviours and Punks:
CDR Working Paper 97.3, June, 1997 Email your order for a paper copy to DIIS order form
The paper is a fieldwork-based case study of the 'commodity chain' for the Nile Perch fish from Lake Victoria, Tanzania. This fish first began apperaring in significant numbers in the lake in the early 1980s and within a few years a large 'artisanal' fishery developed around it. 'Commodity chain' analysis focuses on how a commodity is produced, marketed, distributed and consumed, which groups are involved in each of these stages, how they are organised and how they interrelate. The aim is to identify principles of market structure and organisation and the basic pattern of distribution of earnings and profits.
The paper traces the restructuring of the market for Nile Perch especially after the opening of the first industrial processing plants on the lake in 1990. It links this analysis to issues concerning the social and economic characteristics of marketing chains in marginal economies, and to ones concerning the sustainability of economic growth based on unregulated natural resource extraction.
The Nile Perch is a fatty white fish found apparently exclusively in freshwater lakes in central and eastern Africa, overwhelmingly in Lake Victoria. Although neither especially tasty nor accessible to international markets, probably 40,000 ton of Nile Perch frozen or chilled fillet was being processed at upwards of 30 factories around the lake and exported to developed country markets by the mid-1990s. Some thousands of tons of salt dried Nile Perch were also being processed 'artisanally' and exported to neighbouring countries, principally Zaire, while within the three countries bordering the lake there were further significant domestic trades in chilled, whole frozen, salt dried, smoked, and fried Nile Perch and in the remains of the fish. In Tanzania since the second half of the 1980s, when a significant 'artisanal' fishery for Nile Perch first emerged, the fish has often been nicknamed mkombozi (saviour). The fish's emergence and capacity to penetrate ever wider markets reversed the apparently intractable decline of the fishing industry in a region already suffering from falling production of its other staple, cotton. By the mid-1990s, thousands of young men had found employment as fishermen and, while the main beneficiaries were obviously the owners of the processing plants, some local vessel-owners were accumulating on a previously undreamt-of scale. This whole process occurred overwhelm ingly as a result of the spontaneous working out of different types of 'market forces', and with only minimal state involvement. At the same time, important questions began to be raised locally, both about the outcomes and the sustainability of this chain of development. In particular, public discussion was being broached about its relation to local diet and food security, about the relative distribution of returns between the main groups of actors (including the Tanzanian state itself) and about its physical sustainability. The number of processing plants in Kenya had already fallen from a maximum of 15 to 11 in the wake of declining catches (E. Jansen, personal communication), and the first reports of declining unit catches in the Tanzanian section or the lake date from 1994. Most of these questions feed into international debates on the same subjects, as well as into more abstract debates concerning the long-term sustainability of economic growth based on 'free' markets, especially when economic growth rests on a boom in the extraction of natural resources. This paper examines these questions, but in the course of addressing a somewhat different general focus, namely what the Nile Perch production and marketing systems in Tanzania, and their international extensions, tell us about the economic and social characteristics of spontaneously emerging marketing chains and private enterprise generally in deregulated but still predominantly marginalised economies. The main conceptual framework adopted here is derived from recent studies of global and national commodity chains by, amongst others, Gereffi (1994), Harriss-White (1995) and Bernstein (1996). In these studies, commodity chains are analysed with respect to sequences of production and circulation in 'core' and related chains, the divisions of labour accompanying them, the distribution of earnings and profits they entail, and the forms of social and economic power through which market power is organised to perpetuate or subvert their distribution. In this paper, the latter are approached as an outcome of the interaction between a given set of wider social and economic structures on the one hand, and strategies employed by specific groups of actors and actresses in the marketing chain on the other. A secondary framework will also be used for discussing the relation of the Nile Perch chain to more general processes in the current world economy. This framework arises out of elements of the recent critiques of 'globalisation theory' by Ruigrok and van Tulder (1995) and Hirst and Thompson (1996). The study of the Nile Perch chain is part of a wider project of the author to examine emergent marketing chains in port-liberalisation Tanzania. Besides Nile Perch, studies were also undertaken of the prawn and dagaa fishery chains during a 6-month period of fieldwork between November 1995 and April 1996. A study of the cotton sector, which unlike fisheries was very highly regulated, and where also investment pay back periods are likely to be much longer, is planned for July-October 1997. The Nile Perch study involved exactly 100 extended interviews carried out around Lake Victoria and at markets in Tabora, Sivgida, Dodoma, Dar es Salaam and London. Interviewees comprisedfishermen, traders of various kinds, artisanal processors, matajiri (proprietors of boats and other fishing equipment), transporters, trawler operators and crew, factory owners, managers and workers, and government staff and offices. With exception of the last of these categories, interviews followed the lines of individual or corporate 'business histories', using check-lists inspired by that found in the appendix to Harriss-White (1995). Over a dozen different check-lists were used, tailored to different roles in the chain. Interviewees were chosen on the basis of membership of a given division of labour category or sub-category in the chain and of availability, in collaboration with research assistants with extensive prior knowledge of the local fisheries. Interviews were usually held in Kiswahili with questions and answers translated from and back into English. Interviewees were told that I was working on a background study of private business in Tanzania, sponsored by DANIDA as part of its preparations for redirecting some of its assistance from government to the private sector. They were told also that the study did not involve making recommendations for support either to the fisheries sector in general nor to any individual or company within it. [ The main research assistant for this study was R Mhekele. Additional assistance was provided by F. Matimbo (Dar es Salaam), L Mkwizu (Dodoma, Singida, Tabora) and M Medad (Mwanza). The author would also like to acknowledge the assistance of E Jansen and C Sørensen in sharing with me information from related studies in progress. Lis Bluhme (CDR) typed the first draft of this paper. Many thanks to all those mentioned.] The paper opens with a sketch of the local physical and socio-economic settings which form backdrops to the development of the chain. This is followed by a discussion of the trade's macro-economic significance. Subsequently the paper turns to a description of the division of labour within the chain, and for the commodities intersecting with it either as inputs, services or goods produced and sold together with Nile Perch. Next comes a discussion of the social character of the actors and actresses in the chain and of the strategies and structures through which they interact. This is supplemented by information on earnings and profits in some branches of the chain. In conclusion, the paper turns to general questions concerning the production and contract form in African fisheries, market power and development, enterprise development and accumulation, and the emergent relation of Tanzania to the 'new' international division of labour.
2.1The International Economy In the 1970s several European commentators (notably Frobel et al., 1977) developed the concept of 'New International Division of Labour': This basically denoted a postulated tendency for large-scale firms in the developed countries to export the greater part of their production to less developed countries, principally in order to take advantage of lower local labour costs. On theother hand, the home countries of these large firms were to remain centres for the now-international enterprises' strategic direction (facilitated by advances in communication and transport technology), research and development and global marketing functions. The actual degree of realisation of this postulated tendency even during the 1970s itself is unclear. But by the end of the 1980s, discussion of it had largely been displaced by two other conceptions of the emerging international division of labour. In the first, influenced by the efforts of companies like Ford to create a 'world car', large firms were seen as becoming 'globalised', i.e. as exporting vertically-integrated production and sourcing functions to a wide variety of countries in order to produce a single basic model corresponding to a single basic type of 'mass' international demand. In the second, influenced by one of the most important emerging competitors to Ford, Toyota, large firms were seen as being 'glocalised' rather than globalised. This implied a relocation of production facilities to a relatively small number of sites, all created in the 'triad' of developed regions (North America, Europe and East Asia), relying on creating greater efficiencies in sourcing from independent local suppliers rather than a vertical integration, and producing high value-added products in forms significantly influenced by local tastes. According to Ruigrok and van Tulder (1995) and Hirst and Thompson (1996) each of these concepts more closely resembles a programme of desired outcomes for particular enterprises than an empirically-supported understanding of more general trends with regard to the activity of multinational companies, foreign direct investment and international trade. It is true that the level of total economic activity accounted for by multinational companies is rising, as is the level of total production and consumption accounted for by international trade and the level of total investment accounted for by foreign direct investment (FDI). But an increasing share of FDI and international trade occurs between developed countries themselves and the overwhelming majority of multinational cooperations are neither global in any meaningful sense, nor particularly systematically 'localised' within their external areas. Hirst and Thompson go little beyond this critique, but Ruigrok and van Tulder go on to suggest a framework for analysing different forms of firm internationalisation processes, on a continuum which includes all three of the strategies described above as well as others. This framework will not be described or criticised in detail. Suffice it to say that it does not provide grounds for anticipating any role in the global economy for less developed countries than increasing marginalisation, mitigated only by being sites for greater multinational corporation import penetration and occasional investments by giant oil companies. Indeed, these countries are barely mentioned in the analysis at all. If the framework advanced by these authors is broadly correct, but a more explicit and less unidimensional elaboration of the position of less developed countries in the current international economy is desired, then an obvious strategy is to try to derive various secondary processes atwork in the latter, which might impose limitations on other marginalisation or set off tendencies countervailing it. In this respect can be distinguished particular branches of international capital which are more likely to retain a strong interest in investment in less developed countries on the one hand, and countervailing tendencies released within certain more important less developed countries themselves by, e.g., greater multinational import penetration, on the other. In practice these processes may overlap. Several branches of international capital can be considered likely to retain strong interest in maintaining or even extending an investment base in less developed countries even in the context of the trends described above. The first and most obvious are branches depending on non-substitutable inputs geographically located only or mainly in such countries: the oil companies mentioned by Ruigrok and van Tulder, but also mining companies, tobacco and tropical food companies, oil seed manufacturers, seafood corporations and tropical tourism hotel chains come to mind. A second are branches producing public goods, like electricity generation equipment for whom national governments in less developed countries remain very important customers and in relation to which setting up local or regional production capacity is often a correlate of supplying such goods. A third is branches producing and distributing goods such as cement, beer and petroleum, for which there is high effective demand even in poor countries and strong 'natural' barriers (for example transport costs) to competition. Here, relatively small local investments can easily guarantee local or regional market domination - as well as can be very high profits - at least where the investments in question are in discounted former state assets made available in privatisation 'fire sales'. Finally one can consider capital earned in illicit branches of international trade like drug or unofficial weapons dealing (both of which are very high in the rank order of internationally traded commodities by value) and which for legal reasons are hard to invest openly in the core triad of regions. Each of these kinds of external investment - sometimes on a significant scale - have been evident in Tanzania since 1992-93, against a general background of severe industrial decline. Greater multinational import penetration in less developed countries will hit hardest at capitals in the latter which are already oriented toward main production for domestic markets. This implies capitals located in CDCs which while relatively poor on an international scale nevertheless possess large markets with a certain degree of purchasing power. The latter tend to be found in the regionally dominant states, such as South Africa, Nigeria and Kenya in Africa. It is unrealistic to see all of these capitals being outcompeted or absorbed by multinationals. A more likely scenario is that they respond to intensified competition by seeking to expand into neighbouring country markets (possibly by establishing neighbouring country production basis - particularly, if these are available cheaply via privatisation) or by shifting into low technology locally-based expert branches where knowledge about local factor supply conditions may provide a comparative advantage. Since 1994, in Tanzania there has been a shift of large South Africancompanies into (inter alia) the beer and petroleum distribution sectors and of generally rather smaller Kenyan Asian businesses into a wide variety of others. 2.2The International Economy of Food Production of food commodities is subject to broadly the same general tendencies evident in the international economy generally, with the reservations that certain food commodities have always had a special political significance in all countries, and that there are objective climatic restrictions on what commodities can be grown where. The commodity of greatest and widest political significance has always been grains and regulation of food production and trade on a world scale has always centred on regulation of grains. Largely because of this fact, the developed world has for a long period been more than self-sufficient in grains, and through extensive state support has had the opportunity to develop a fully industrialised grains sector. The combination of physical surpluses and technical advances which followed from this industrialisation led to the parallel industrialisation of the livestock sector through the animal feeds connection and to the increasing substitution of tropical crops (e.g., vegetable oils, sugar) by grain-based alternatives (soya oils, corn syrup sweeteners) (Friedmann, 1993). It was also associated with the emergence of several very large integrated multinational, but usually US-based, food companies. Japan and the Soviet Union represented exceptions to the rule of developed country grain self-sufficiency. The Soviet Union imported large quantities of grains from North America, at least until its collapse in 1989-90. According to Friedmann (ibid.), Japan by contrast developed a policy of 'strategic multi-sourcing', helping through a programme of diversified small-scale investment to create as many exporters as possible to allow importers to pick and chose and, presumably, keep prices down. Increased affluence in developed countries generally has meanwhile created new markets for certain specialized foodstuffs, including a number which can only be produced in tropical or semi-tropical countries. According to Friedmann (ibid.) these markets have largely been supplied by a group of 'New Agricultural Countries' including Brazil, China and India which have combined the promotion of high value added exports with their own domestically-led industrialisation of parts of the grain and livestock sectors. It is not part of Friedmann's argument that the food multinationals are involved in the trade in high value added fruits and vegetables, although some are now using tropically-grown materials again in the production of feedstuffs. In a recent contribution Watts (1994) has gone well beyond Friedmann's observations to claim that there are renewed systematic tendencies toward the internationalisation of food products and trade, centred around high value-added fruits and vegetables from tropical and semi-tropicalcountries. This is said to be based on a continuation of the 'growing transnationalisation of agro-food capitals' on the one hand and the application of Japanese-style 'pick and choosing' strategies on the other. Because the crops involved are only of interest to affluent northern hemisphere consumers if they correspond to specific norms of quality and presentation, and because they also typically do not lend themselves to mechanisation, their growing is normally contracted out rather than being undertaken by the multinationals directly. The result is 'flexible accumulation in the fields'. Watts provided little direct evidence to support his arguments, either with respect to the global expansion of demand and supply of high value added fruits and vegetables, nor for the systematic involvement of transnational corporations in this process. In fact it seems likely that the bulk of transnational corporation interest in tropical food crops remains in tea, coffee, cocoa, bananas and pineapples as it did a half century ago (or in most of these cases, even earlier). Moreover, it seems further likely that because of the high capital content and level of infrastructural dependence of high value fruits and vegetables, their production in tropical and semi-tropical countries will tend to be highly physically concentrated around the already developed poles. 2.3The International Economy of Fish The analysis developed so far suggests that there is likely to be little direct international interest in promoting production or trade in 'new' tropical foods in countries like Tanzania, except perhaps on a relatively small scale by capitals from the more developed parts of the same region. However, developments in the specific international economy of fish mean that in the case of this type of food, certain modifications need to be introduced to this picture. The obvious difference between fish and most other foodstuffs is that fish are a non-renewable resource: the story of the industrialisation of the fishing industry is therefore simultaneously the story of the systematic depletion of international fish stocks. This tendency and industry responses to it lay at the heart of understanding the current international economy of fish. The industrialisation of the north Atlantic and north west Pacific table fish fisheries began in earnest in the 1950s and 60s, associated with major increases in industrial country demand, the formation of vertically integrated seafood corporations in the US, the Soviet-Union, Japan, Korea, and Taiwan, and the associated construction of a number of factory freezing fleets. Evidence of overfishing of Atlantic cod already emerged by the end of the 1960s and by the early 1970s it also became clear that North-east Asian stocks had been heavily depleted. Producer responses to these trends have included movement to different species and different fishing grounds and have coincided with an industrialisation of a large number of new fisheries, even as the decline of Atlantic cod was making redundant the first generation of factory freezer vessels. While the traditional North Atlantic and North-east Asian fisheries remain active, though on a reduced scale, the last two decades have seen a growth of international fishing effort in four main directions. One has involved the development of a very large-scale international fish meal complex, based on the serial targeting of various small surface-based species for industrial processing for feed. The target species concerned have been anchovy, pilchard and sardine. Fishing for millfeed possibly accounts for about a third of all consumption from industrial fisheries today. A second direction has been a concentration at the other end of the market on capture of certain very high value species such as tuna, squid and prawn. In the case of tuna this has involved the development of very high technology fishing systems incorporating satellite data analysis and air-based reconnaissance. In the case of prawn (and salmon) it has mainly involved movement in a third direction- the promotion of acquaculture, some of it very highly intensive. [This is discussed in a parallel paper on prawn (Gibbon, 1997a).] A fourth direction has already been indicated - the serial exploitation of substitutes for the main traditional developed country mass table fish markets. One recent example of this trend was the 'discovery', incorporation, exploitation (and then exhaustion) of 'Southern Hake' stocks off Chile during the 1980s (Schurman, 1996). Besides a general intensification of technology (larger boats, larger-scale gear, small size meshes, etc.) each of these trends has been associated with shifts in fishing grounds, to steadily more remote regions. By the 1990s almost all the world's largest marine expanses, and most of its freshwater ones, had been overfished. Significant parts of the North and West African seaboards were overfished by EU (mainly Spanish) vessels over a very short period in the 1980s, for example. In the face of increasing regulation by governments in most regions, mainly taking the form of extensions of territorial waters, there was also a tendency for changes in ownership patterns to occur, with large internationally-mobile fleets declining and with more local flagging joint ventures and charter assignments, as the large international players found themselves obliged to 'localise' their operations. In a context of continuing very strong demand in developed country markets, promoted both by rising average incomes and by the gradual weakening of the market image of meat, the underlying world price for table fish has risen steadily. So too have price fluctuations occasioned by temporary blips in supply. Meanwhile there have been developments in international transport technology, chiefly the generalisation of refrigerated container transport on a world scale, which have relatively cheapened secondary costs. All of this has created opportunities for windfall profits by producers of fish such as Nile Perch which only a few years earlier would have been considered born insufficiently attractive and insufficiently profitable to be internationally tradeable. On the other hand, such species are favoured by northern hemisphere importersprecisely because of their ability to dampen international price increases generally. If their own prices rise to that of staples like cod, they cease to be of interest.
3.1The Economy of the Lake Victoria Region The Nile Perch fishing grounds are located around the entire shoreline of Lake Victoria although they are concentrated in certain areas more than others. By 1996, when this study was undertaken, the busiest areas on the Tanzanian shoreline were in the south west corner of the lake, on the islands in Muleba, Geita and Sengerema districts. A second area of concentration was Ukerewe, a very large island in the south east corner of the lake. Besides fishing, most artisanal processing of the fish was also being carried out on these islands. By contrast, industrial processing of the fish was overwhelmingly concentrated in Mwanza town, Tanzania's second largest settlement and located more or less at the mid-point on the lake's southern shore. Factory processing there had also spawned a second branch of artisanal processing, based on fish waste products. This was scattered between one large and three smaller sites around the town. Lake Victoria is the largest inland water surface in Africa. It is difficult to properly group its size. The lake is large enough to have its own 'sounds', 'gulfs', channels and groups of islands. Modern passenger ferries circumnavigate it rather than sail across. The trip from Mwanza to the two other largest ports in the Tanzanian part of the lake (Musoma and Bukoba) takes around nine hours each. If the current generation of industrial trawlers wished to fish in the centre of the lake, the sailing time for Mwanza would be close to 18 hours. By 1996 much of the shoreline, especially on the Ugandan site was being choked by water hyacinth (magugu maji). On the Tanzanian shore some inlets tended to be blocked for a few days, but in general the Nile Perch fishery had not been badly impeded. [ The hyacinth, and another secondary plant which tends to accompany it, was in 1996 particularly prevalent in Smith Sound. Fish trapped by it are vulnerable to asphyxiation and nets are subject to clogging. Water hyacinth probably presents a more severe problem for the dagga fishing than the Nile Perch one however, as the reduced water temperature and darker water body generated by it impedes the development of this fish ’s main foodstuff, plankton.] The lake lies in the middle of a flat and gently rolling plateau, at a height of around 1100m. Throughout the plateau are spectacular granite outcrops and close to Mwanza these form cliffs around the shore. Offshore are more granite outcorps, including the one which drained the MV Bukoba with the loss of several hundred lives in mid-1996. With the advent of colonianism the lake and its shoreline was divided between the three East African territories, and efforts were made by the colonial government to integrate each of the lake regions into their respective 'national economies'. Around what became the Tanzanian shoreline the bulk of the populationbelonged to the Wasukuma people. These practised farming (maize, cassava, groundnuts, sweet potatoes and chick peas) in the high rainfall area close to the lake and herding in the less densely populated and more arid areas at a distance from it (McLoughlin, 1969). After 1945, when British economic and political priorities dictated shift to an explicitly 'developmentalist' agenda, Sukamaland became the subject of probably the most comprehensive set of state interventions in the history of the country. Under the 'Sukumaland Development Scheme' the area was designated for peasant cotton cultivation. Huge areas of bush were cleared particularly in Geita and Biharamulo districts, local water supplies developed, cattle off-take quotas introduced, cotton production promoted and land use and cultivation methods carefully monitored. To police the entire operation an overarching 'native authority', the Sukumaland Federation, was set up. Most of the objectives of the Sukumaland Development Scheme failed miserably, but especially in the newly opened-up areas cotton production grew vigorously, buoyed by a rising world price in the post-Korean War period. Reliable data on acreages are lacking, but seed cotton output increased from 50,000 (180 kg) bales/year in 1950-52 to 227,750 bales/year in 1962-64. After independence, cotton continued to increase in importance until the early 1970s, whereupon its cultivation stagnated and then contracted. Farming systems studies conducted during the region in the 1980s demonstrate an increasing substitution of maize and rice for culture on better soils and in higher rainfall areas. The bulk of these crops were also traded, but in the developing network of regionally-based private markets around the lake rather than in the nationally-based and state controlled ones centred on Dar es Salaam, as cotton had been (Meertens et al. 1995). Increased local marketing of maize and rice was part of a general restructuring of the economy of the southern lake area during the 1980s. Two other important elements were a revival of the local gold mining industry (dominant since the 1940s), and an apparently huge increase in smuggling with Kenya following the reopening of the common border in 1983. The story of 'artisanal' gold mining industry in Geita district is told in Chachage (1995). Gold 'rushes' by artisanal miners, some of them laid off by the formal sector mines in neighbouring Shinyanga, began at Nyaruguru in the late 1970s and by the time that claims were officially issued in 1984 around 50,000 people were present on the site. Later in 1980s further rushes occurred at Buziba-Rukarakata, at Rwamagaza and at Mgusu, all south-west of Geita town. Until around 1990 (and again since 1995) the great bulk of the gold produced was smuggled out to Kenya; some of the proceeds were used to smuggle consumer goods back into Tanzania, but the latter trade also had a life of its own with beer, cigarette, khanga and soap as the main unofficial imports. Considerable fortunes were made both in gold and in smuggling. One of Mwanza's two most eminent African capitalists today began his career as a local maize and rice trader, suffered imprisonment for smuggling in 1983 but resumed in 1984 as a semi-official importer of goods from Kisumu and Nairobi 'under the patronage of a prosperous Asian he knew in Uganda' (Truelsson, 1996). As the 1980s progressed he gradually formalised his activities, amongst other things going on into cotton seed oil production in 1990 and most other links of the cotton chain by 1996. By the mid-1990s his large storage-cum-office block cum Bureau de Change next to the main bus station, rebuilt in red Ugandan brick after being burnt down (allegedly by a rival) had come to literally tower over the town's African business district. Other 'big money from magenta' (smuggling) went into gold. This recent history has contributed to Mwanza town's rapid growth (with a population somewhere between 0.25 and 0.5 million it is Tanzania's second largest settlement) and to its contemporary 'wide-open' feel. Evidence of new money is not hard to find. Capri Point, directly overlooking the lake and the most spectacular granite outcrop in the town itself, is dotted with very expensive- looking new houses, which can also be found on the outskirts of town at Nyegezi, Nyakoto and Nyamanoro. Older local Asian capital together with even more recent foreign capital, mainly from the Canadian, Australian and South African formal sector mining houses who are now pushing out the artisanal producers, may be observed at play together at the recently-opened Mwanza Casino [ The casino is owned by a Dar es Salaam - based company, managed by a Frenchman and staffed roughly evenly between Bulgarian and African croupiers. The African croupiers are virtually the only Africans to be seen there. One of the Bulgarians told me she and her companions were ‘lured ’ to Mwanza by talk of a high - rolling international resort; they could not leave the country because the owners had not paid their income tax.] and at Capri Point's Tilapia Hotel. 3.2The Lake Victoria Fisheries The promotion of fisheries on Lake Victoria never constituted an important element of the development agenda of any government in Tanzania, either colonial or independent, although it was a colonial government technical intervention of the early 1950s which laid the foundation for the current Nile Perch boom. There have been commercial fisheries on the lake almost the whole of the 20th century. Garrod (1961) dates the first adoption of factory-made nets by 'native fishermen' as 1908 and the first large scale commercial fishing boom as 1917-20. At the beginning of a second boom (in 1930) there were already 1.7 m pieces/year of the indigenous Tilapia O. esculent (ngege) being transshipped by rail from Kisumu into central Kenya, and at its end in 1940 there were 5 m pieces/year. The commercial fishery in Tanganyikan waters developed more slowly, but between 1923 and 1926 230 ton/year of dried ngege were also being transported by rail from Mwanza (Graham, 1929). In both Mwanza and Mara there were also more local trades. While the bulk of the lake's biomass this century seems likely to have primarily comprised the little-exported Haplochromis (furu), and while there have been commercial fisheries throughout for Bagrus (hongwe), Labeo (ningu), Clarius (mumi) and one or two other species, the main commercial fishery until the 1980s was always for Tilapia, to be traded in a dried or smoked form after artisanal processing. Fishing for it from the beginning of the century seems to have been by methods not far removed from the dominant ones today- mainly setting gillnets overnight in long stationary chains, and processing methods (gutting, then sun drying on rocks or smoking on earth kilns) also remained largely unchanged. However, until the mid-1980s there was still a clear annual fishing season (wet season and early dry season) and even the commercial fishery was largely part-time in nature (Bituro, 1974). Ngege was never a prolific breeder and its main feeding grounds (waters at a depth of 4-20 m with a grassy floor) were easily accessible by canoe. It is therefore not surprising that from a very early date concerns were raised about its overfishing. Graham's (1929) report stemmed from these concerns. He recommended a ban on gillnets with meshes smaller than five inches and indicated that discussion was already occurring about the possible introduction to Lake Victoria from the nearby freshwater sources of two exotic species: the Nile Perch or Lates niloticus (sangara), and the Tilapia O. niloticus (sato). According to Garrod (1960, 1961) a third phase of rapid expansion of the commercial fishery occurred in 1949-53, especially in Kenyan waters, which towards the end was accompanied by the introduction of nylon nets and of outboard motors. In inshore waters unit catches fell rapidly after 1945. An increasing movement of fisheries to camps in the islands was reported and in 1956 Graham's net size regulation was abandoned as unenforceable. It was around this time that the Nile Perch and O. niloticus were introduced to the Ugandan part of the lake as part of a fish replenishment programme. According to Reynolds and Grebovval (1988), Nile Perch first began appearing in significant numbers in catches in the Tanzanian part of the lake in 1982. Nobody I interviewed had fished it commercially prior to 1984, but by late 1986 a Norwegian survey reported in Reynolds and Greboval (ibid.) found that almost half of all gillnetters (themselves 84 per cent of all fishermen using vessels) were using meshes of more than six inches - more or less specific to Nile Perch. This was partly an effect of the rapid development of a market for Nile Perch. Local markets for fresh Nile Perch and markets which were both local and regional for fried and smoked Nile Perch were established already by early 1986. But it was also an effect of the general impact of rising stocks of Nile Perch on the population of other favoured lake species. Nile Perch was carnivorous and its targets included the indigenous Tilapia as well as hogwe, mumi and furu. Amongst remaining fish in the lake the impact on the exotic Tilapia sato appeared to be roughly neutral,while the main beneficiary was the sardine-like R. argentea (dagaa Mwanza) which competed for food with furu but which was generally found at different depths to the Nile Perch. In 1987 the first frozen Nile Perch fillet was exported from Kenya and by the end of the year no less than ten Kenyan factories, in Kisumu, Nairobi and Mombasa, were exporting the fish - primarily at this time to Israel, where the main customer was the Israeli armed forces. Production was at first on a small-scale (4000 tons/year) and it seems likely that even the bulk of Kenyan catches were at this time being consumed domestically. This was also true of Tanzania until 1990-81. Interviews conducted during the study suggest that during the late 1980s the Nile Perch trade was primarily local, not least through the take-off of a major market for fried sangara in Mwanza town, and inter-regional, with the opening up of a national market for smoked Nile Perch (sangara moto/sangara moshi) and a market in Dar es Salaam for frozen and semi-processed wholefish. The main export markets at this time were initially dominated by Tanzanian traders taking the fish as far as Ngare, but from 1988-89 slowly taken over by Zairean traders buying in Mwanza. During 1988 the long-distance trade in artisanally processed fish became more extensive and better organised. The landing at Kirumba just outside Mwanza town (the main local centre for informal transport around the lake) became its main entrepot, as well as taking over from Mwanza central market as the main site at which fresh Nile Perch was locally traded. According both to Reynolds and Greboval (op.cit.) and to my own interviewees, virtually all the Nile Perch caught at this time was fished between the Kenyan waters and just west of the Mwanza Gulf. There was little fishing in Sengerema district and virtually none in Geita district or along the Bukoba shoreline: 'the Wahaya were hardly involved and even in Geita people thought of fish like they thought of frogs'. [ Statements in single quote marks are either verbatim translations by my research assistants from direct speech in Kiswahili by interviewees, or verbatim quotations in English.] In 1990-91 market conditions began to change fundamentally. As more and bigger processing plants opened in Kenya, their owners started buying fresh Nile Perch in Tanzania. They worked mainly through Tanzanian agents, who were provided with motorised collector boats, vans and trucks. The prices offered by the Kenyans easily exceeded local prices 'and we all saw scales for the first time on a beach'. Simultaneously, both in Zaire and in southern Tanzania, demand - and hence competition - for fish was further stimulated by the local adoption of salt drying, which increased shelf life (and cheapened processing costs by eliminating the need to purchase fuelwood). 3.3The Sites of Production and Trade Since 1990 the Nile Perch fishery has spread to the whole of the Tanzanian Lake shoreline. The most intensive fishing is conducted from makambi (camps) rather than villages. A majority of camps are situated on the islands (mainly in the Lake's south-west corner) as fish stocks in the area reachable from the shore by canoe or sail-powered vessel have been depleted. Makambi may be more or less permanent and more or less organised and have populations ranging from a dozen to over a thousand people. A handful of vessel owners (matajiri) are powerful enough to be able to set up and run large camps on their own, but most makambi comprise a series of sub-camps organized by different matajiri. Besides fishermen, the camps are populated by would-be fishermen trying to find employment, by specialised fish processors, netmenders and other tradesmen (fundi), cooks (all women), and sometimes the wives and children of a tajiri. On the periphery of some of the larger camps are kiosks, bars, tailors' premises and even video halls. The population is normally a floating one, in more than one sense: 'all kinds of people are attracted here', one camp chairman told me, 'youths who have been chased away from their houses or their villages, women who have run away from their husbands, men who have killed'. Law and order is rough and ready, and basically vested in the matajiris and their meja (supervisors, literally majors). Most camps are visited by agents from the factories in Mwanza on a daily basis or as part of 'rounds' thrice or twice weekly. At all the larger landings there are permanent wooden structures to hang the agents' scales and provide them with ofisi (offices). In 1996 factory purchases of fresh Nile Perch accounted for probably three quarters of all Nile Perch caught. Fish that the factories were unable to collect, because of delays in arrival at camps or arrival without adequate supplies of ice, as well as fish which the factories rejected as too small or as 'stale', were processed artisanally in the camps themselves. The largest physical site for the bulking and resale of artisanally processed fish in Tanzania was still Kirumba mwalowi. Between two and three kilometres north of the centre of Mwanza town, the landing had become a walled compound covering an area of about 250 x 800 m, on a sandy bank rising evenly from the lake like a terrace at a football ground. It was divided into areas specialising in different commodities including, besides processed fish, bananas, building poles, firewood and charcoal. In each of the areas except the ones where wood was traded, goods were stored on stands built of rough-sawn thick cypress planks resting on granite blocks. The stands were mostly 2-3 m long and 0.5 m off the ground and some were stacked to a height of 2-3 m. Most were covered in blue tarpaulin sheets stencilled with the letters 'UNHCR', presumably emanating from the camps in neighbouring Kagera region at that time housing Rwandan refugees. Kirumba mwaloni bore certain signs of recent formalisation. Besides the boundary wall and gates there were a couple of water taps, a public toilet and the offices of four or five cooperative societies, three of them dealing with fish. These were solid-looking wooden huts with desks, chairs, filing cabinets and telephones, and walls covered with direct telephone dialling codes forTanzania and neighbouring countries. Next to the mwaloni was a retail market and beyond it a relatively prosperous African suburb. The road leading to the mwalo was lined by bars and 'nightclubs', eating places, guest houses and hair saloons and not far behind it lay at least one respectable modern hotel and an international telecommunications office. Kirumba was actually in decline as a centre for fish trading, though not as an informal passenger and cargo port. Its golden years were as a market for fresh fish, just before the generalisation of the presence of factory agents. The trade's centre of gravity in Tanzania is now firmly at the factories themselves. In April 1996 there were seven functioning plants in Mwanza town and one in Musoma, as well as a number in various stages of development, including a further four in Mwanza town alone. They were a mixed bunch, ranging in appearance from a small converted tyre godown (warehouse) without running water and with just one or two small plate freezers and a small flake ice machine, to a purpose-built plant of 1250 m2 currently under construction, which will have two automated lines with their own tunnel freezers, 96m2 of cold storage and 'everything imported from Finland except the cement'. The last important element of the local physical setting were the sites where the artisanal processing of factory fish waste occurred. In 1996 the largest of these was at Nyegezi, around four kilometres south of the main groups of factories in Mwanza town, along the Dar es Salaam road. Driving along this road in 1996 was never easy, not merely because of its appalling condition and steep gradient, but because of frequent encounters with groups of four or five youths pushing trolli (hand-carts with axles, wheels and tyres cannibalised from pick-ups). Each trolli was spilling over with between one and two tons of stinking fish carcasses, bought by traders (also known as matajiri) at the factory gates for processing at the c. one hectare Nyegezi site. Clouds of flies swarmed round this site, which could be smelt at some distance away, and above them swarmed quantities of swallows and a variety of hunting birds. Upwards of 200 people were using the site in some way or another, about half of whom were women- many with accompanying children. Although squalid, there were signs of formalisation even here, with running water, an enclosed toilet, a site office under construction and three or four kiosks. 3.4 The Nile Perch Boom and National Economic Policy The Nile Perch boom, like the boom in prawn (see Gibbon, 1997a), coincided with the adoption and elaboration of structural adjustment policies in Tanzania and owed a considerable amount to them. For a decade and a half prior to adjustment Tanzania had followed a path of 'socialism and self-reliance', which involved nationalisation of the 'commanding heights' of the economy, 'confinement' of most productive acting and trade to parastatal or state-run cooperatives and holding the exchange rate at a very high level in order to subsidise the import of capital goods. Simultaneously the rural population was forcibly resettled in villages, resulting in a massive disruption of agricultural production. Marketed agricultural output also declined as parastatalmargins steadily rose, resulting by the end of the 1970s in a sharp decline in staple exports and a severe foreign exchange crisis. From 1979 onwards, while the Tanzanian government and the international financial institutions were deadlocked over the terms on which the latter might extend balance of payments support, there was a widespread contraction in official economic activity and a growth in the so-called 'second economy'. The latter term subsumed both the mass diversification of popular livelihood sources away from export crop agriculture and official marketing channels generally, and the more specialised trends of smuggling, diversion of official imports, hoarding, and illegal dealings in confined goods and forex. After efforts to suppress these trends ended in fiasco and the death of their main architect, a gradual official opening began in 1984. The Tanzanian shilling was allowed to rapidly depreciate and traders holding hard currency were allowed to import goods into the country on a 'no questions asked' basis. This latter measure regularised a lot of already existing 'second economy' activity. As deconfinement of exports and an easing of regulations governing the right to retain earned foreign exchange followed, growth in exports as well as imports was stimulated. The prawn boom was lead for a very long period by relatively small-scale large numbers of locally-based traders making large profits from the difference between low Tanzanian shilling-denominated local costs and high hard-currency denominated international prices. These profits were often supplemented by large profits made on exploiting pent-up demand for luxury imports purchased with export earnings, or by selling undeclared hard currency earnings locally to realise their still considerable parallel forex market premium, or both (see Gibbon, 1997a). The Nile Perch boom occurred only towards the end of the period in which such 'supplementary' benefits were important. From a policy viewpoint, the decisive factor bringing Kenyan buyers to Tanzania was the real lowering of local costs brought about by devaluation. Their buying agents of course benefitted from the other money-making opportunities which the trade opened up for them, but these were generally of secondary importance. From 1992-93 the Kenyans were forced to set up plants in Tanzania itself by the banning of whole (and later 'semi-processed') fish exports. Many would have probably shifted to Tanzania anyway to take advantage of low local costs and factories were also built by citizens. The cost of their investments was cheapened by the various investment incentives which were made available between 1990 and 1995 (tax holidays, remission of import duties and sales taxes on capital equipment, 100 per cent forex rention, automatic access to leases on land for intended investment sites), while their recurrent costs were cheapened by the de facto deregulation of Tanzanian labourlaws brought about by growing state incapacity. On the other hand, these advantages were partly offset by the very high costs entailed by the appalling state of the local infrastructure. By 1996 the international financial institutions were proclaiming Tanzanian market liberalisation a decisive success. However, they were also expressing strong concern about the failure of state revenue collection to keep pace with improved levels of economic growth. This proved to be one context for a substantial increase in royalties, levied on the factory exporters from April 1996 (there were other contexts too, see below). Combined with a steady but very strong appreciation in the price the factories were having to pay for fresh Nile Perch, and with a stabilisation of the international value of the Tanzanian shilling, this was to lead to some factory owners publicly wondering about the continued viability of their investments. However, the factory owners had still not accepted the relevance of the ostensibly more fundamental question, namely the length of time in which the Nile Perch might be expected to be still available for economic extraction at current rates.
The discussion here will consider Nile Perch exports in various forms, as well as the internal trade in Nile Perch, from a quantitative standpoint. It will also make some very preliminary observations concerning the numbers of persons involved in the Nile Perch industry and about some elements of the local income generated by it. The preciseness of the figures appearing in the tables needs to be taken with a pinch of salt, as will become clear. Official published figures on Tanzanian fish fillet export volumes and values could be traced only for 1993 and 1994 and are found in Table 1, along with official figures for 1992 given in van der Hoeven and Budeba (1993) and an 'officially-based' figure for ex-Mwanza export volumes and values from Mwanza derived from examining individual record of export declarations made at Mwanza Region Fisheries Office (RFO). To this should be added an estimate of exports from Musoma. The export earning figures declared by exporters in each year are systematically understated and Table 1 includes a rectification of them for 1996. It may well be that several companies also systematically underdeclared their real export volumes. Two company managers both told me 'we are the only ones who don't do it'. On the other hand, because their export earnings underdeclarations were being taken at face value until 1996, there was no particularly strong incentive to underreport volumes also, and in the case of exports by rail the volume records of the Tanzania Railways Corporation more or less correspond to declarations made at the RFO. Even if currently reported, volume figures for 1995 are not a good guide even for 1996, since twofactories only started up in the course of the year and the largest underwent a major expansion of capacity. Table 1 Nile Perch fillet exports 1992-96, Tanzania
Sources: 1992 a:Actual Tanzanian production 1992 b:Estimated export to Kenya (calculated as fillet equivalent on basis of fillet: wholefish ratio of 38 per cent), van der Hoeven and Budeba (1993) 1993-94:Wizara ya Utalii, Maliasili na Mazingira (annual reports) 1995:Booked exports for royalty payment purposes, Mwanza RFO 1996:Based on personal estimates of individual factory outputs, April 1996 and catch data (see below) applying a fillet: wholefish ratio of 38 per cent 1996 values:Based on estimated average export price of USD 3.3/kg Key: *Includes fillets other than Nile Perch +Excludes production in Musoma ( )Indicates systematic underdeclaration on price When factory managers or owners were interviewed in March-April 1996 they were asked (inter alia) what their current production was, whether it exceeded levels of 1995 and what they thought the total current factory throughout was for the Tanzanian part of the Lake. Most were rather evasive about their own production level, a fact no doubt influenced by the dispute with the Tanzanian government over underdeclaration on price which had just broken out. On the other hand, all stated that their products exceeded the levels of 1995 and there was a surprising degree of agreement on estimates about total current throughput. On this basis, an estimate for fillet exports for 1996 has been generated. Nile Perch is exported in several other main forms besides fillet, namely as kayabo or sangara chumvi (dried salted steaks), 'chips' (dried salted belly flaps? and fillet trimmings), and mabondo (maws or gas bladders). The mabondo trade is sufficiently different to be considered separately (see below). There is also an export trade in fins, tails and skeletons but this is not particularly significant. Table 2 Exports of artisanally-processed Nile Perch, 1995, Tanzania
Sources: Kayabo volumes: a: Mwanza RFO; b: Kirumba mwalo market office, based on local estimates of unit weights of tenga at 700 kg and pakacha at 200 kg, and estimating sales for 1-15 January (missing in receipts books) as the same as for 11-31 January; c: Mwanza South railway freight office, Owners' Risk Goods Consignment Note counterfoils. Kayabo values: based on average wholesale purchasing price in Mwanza of Tsh 550 (USD 1)/kg Chips volumes: a: Mwanza RFO; b: Mwanza South railway freight office, as above. Chips values: based on average purchasing price (processed) in Mwanza of Tsh 200 (0.36 USD)/kg No official figures of any description have been published on any of these trades, although some quantitative information is available by sifting through a number of unpublished sources. These include individual records of export declarations by foreign traders made at Mwanza RFO, individual records of sales recorded as exports by market officials at Kirumba mwalo, and counterfoils of Owners' Risk Goods Consignment Notes prepared at Mwanza South Railway freight office for traders using the Mwanza-Kigoma goods service (during 1995 the main export route to Zaire). None of the total export figures derived from these sources can be taken particularly seriously, both because of systematic underrecording to avoid market levy and royalty payments, and possibly also to reduce rail freight charges. Secondly, artisanally processed fish is never sold by weight, but rather by the piece, so weight figures recorded by market offices tend to be based on estimates of the volume of the containers (normally very large parcels (pakacha) or baskets (tenga)) which are used to shift fish from one place to another. A further complication is that such pakacha and tenga of different standard capacities are used in different parts of the country, meaning that where volumes are recorded in terms of numbers of containers rather than weights, consistent aggregation is difficult. Figures for export values of artisanally-processed kayabo and chips given here are based on prices paid by exporters in Mwanza itself and do not represent true f.o.b. values. Transport costs to the border with Zaire (the destination for almost all of these exports) will be considered in a later section. Table 3 Internal Tanzanian trade in Nile Perch, ex-Mwanza, 1995 (tons)
Sources: 1. Fresh frozen:a. ATC, Mwanza. Excludes export cargoes carried by ATC and all cargoes by other carriers; b.Eestimated on basis of passenger parcels counterfoils, January-April, Mwanza Town station 2. Dry salteda. Kirumba mwalo market office; see note 6 on Kayabo/Table 2; b. Own estimate 3. Smokeda. As per 2a above; b. own estimate 4. PunkBased on traders' estimates of 6 trips/week by 3.5 ton trucks during dry seasons (22 weeks) from all artisanal sites in Mwanza. No official figures exist for the internal trade in Nile Perch either, although quantitative information of various kinds is again available from diverse sources. For the internal trade in frozen and so-called 'fresh' Nile Perch, which is conducted exclusively by air and by passenger train, information may be derived from counterfoils of individual weight bills for air freight held at the Air Tanzania office in Mwanza and counterfoils of Owners' Risk Goods Consignments Notes for passenger parcels loaded at Mwanza town railway station. As for the trade in dry salted and smoked Nile Perch is concerned, more or less the same reservations apply to the sources employed here as were elaborated a moment ago in relation to artisanally-produced exports. The basis for my own estimates concerning these trades are stated later in the text. After installation of a new director at the Mwanza RFO in 1995, a serious fisheries census was undertaken for the first time in living memory. By the time of my own stay in Mwanza, returns had been received from all the districts in Mwanza Region which bordered the lake except Geita. No similar census had been undertaken in Kagera or Musoma Regions although it is almost certain that the industry in Mwanza dwarfs that in neighbouring regions. Numbers of Mwanza-based fishermen specifically involved in the Nile Perch fishery has been estimated on the basis of other information in the census concerning the distribution of all enumerated gear between different fisheries. Estimates of numbers of other types of workers in the trade have been generated according to principles explained in the notes to Table 4 and justified elsewhere in the text. Table 4 Estimated population directly involved in Nile Perch fishing industry in Mwanza region, 1995
Sources: 1a. Calculated from Mwanza Region fisheries census and assuming the Nile Perch fishery represents two months of all fishery employment. 1b.Own estimate. 2a.Assuming 1 processor for 10 fishermen. 2b.Own estimate; majority at Nyegezi; large number working in own houses. 3 a,b.Assuming 1 cook and 1 net mender/fundi for every 20 fishermen. 4a.Based on 3 crew members for every factory collector boat (75 registered between February 1992 and March 1996) plus pick-up, truck and land-based agents. 4b.Based on 3 crew members for every non-factory collector boat (59 registered between February 1992 and March 1996; many other unregistered or registered under other users, possibly 75 in all); remainder bicycle traders in Ukerewe and Mwanza Gulf and a few pick-up traders. N.B:, a large majority of 'independent' collector boats owners are factory sponsored. 5a.Own observation. 5b.Comprises c.15 airfreight traders; 30 'punk' traders; 65 others. 5c.Own estimate. Most either serve the Geita minefields via Nkome or inland areas of Magu from Nyamikoma. 6.Own estimate. 7.Own estimate, based on 15 functioning trawlers prior to trawler ban. 8.Mainly trolli crews and pick-up truck drivers on Kirumba-Mwanza south and Nyegezi-Shinyanga and Nyegezi-Magu routes. Table 5 Main local incomes from 'outsiders' in the Nile Perch industry, Mwanza, 1995 (M. Tsh)
Sources: a.based on wholefish equivalent of declared production of 9904 t, assuming fillet: wholefish ratio of 38 percent and assuming 1995 average buying price of Tsh 300/kg. (includes commission to collector). b.local currency equivalent of values in Table 2. c.based on price to Karickoo wholesalers of Tsh 0.63 m/ton excluding transport, and volumes given in Table 3. d.based on wholesale price given in note on kayabo values, Table 3 and average of two volume figures given under 'dry salted a' in Table 3. e.based on wholesale price of c. Tsh 500/kg and average of two volume figures given under 'smoked' in Table 3. f.assuming average price of Tsh 315/kg at markets in Shinyanya and Magu and volumes stated in Table 3. g.Assuring average wage of Tsh 17,500/month and employee numbers stated in Table 4. N.B.Mabondo had been excluded because of difficulty apportioning overall returns. Estimates of total income generated for the population of Mwanza region by the trade have been made by estimating total expenditure by 'outsiders' (mainly factory owners) on different kinds of fish. An effort has been made to avoid double counting but it may well be that some has still occurred. There is a probable overstatement of the item on factory expenditure on whole fish and understatement of the item on employment, for in 1995-96 several factories were turning from purchasing from more or less independent artisanal fishermen to developing their own fleets of fishing vessels. Even on the basis of the very rough figures indicated here, it is clear that the macro-economic significance of the trade is very considerable, and grossly underestimated in official statistics covering exports and GDP. On the other hand it is probable that official statistics seriously understate levels of exports and GDP across the board, meaning that trying to calculate the Nile Perch fishery's share of either is a futile exercise. What can be said with some certainty is that huge amounts of money have been made in the trade and that the factories have been its mainbeneficiaries, not least at the expense of state revenues. Equally certain, the trade also supports a broad swathe of the population of Mwanza region and very considerable sums have been recycled in the local economy - in 1995, probably around half of all export earnings.
Unlike in the case of prawn, there is today only one major marketing chain for Tanzania Nile Perch - although this chain is a tangled one, resembling a river system with different tributaries, collection points, outlets and short-cuts between them. This section concentrates on the structure of the chain in 1996, while the next will discuss the nature of the actors and addresses found at its various links, the social relations between them and the strategies which they have employed in order to establish, maintain and improve their position. The simplest way to describe the chain's structure is to take the process of fishing as a starting point and to work downstream. 5.1Production Functions Until 31 December 1995, when trawlers were banned from the Tanzanian (and apparently the Ugandan and Kenyan) waters of Lake Victoria there was both semi-industrial and 'artisanal' fishing for Nile Perch. At this time there were about 20 trawlers licensed to operate in the Tanzanian part of the Lake, of which 17 were based in Mwanza. All were extremely small, between 9 and perhaps 18 metres and with engines of only 65-175 BHP. None had on-board processing or freezing/chilling equipment and only a few had fish-finding equipment. The largest had a crew of 4 and 2 officers . They were mostly using four inch trawl nets. The vessels had a mixed history. Trawling on the Tanzanian part of the Lake dated from the early 1970s when the Dutch built a plant to mill the then still plentiful furu for the local animal feed industry and three vessels (in 1996 still the largest) were imported to supply the plant and became the property of the parastatal TAFICO. Three others originated from overseas assistance programmers to government-owned local fishing institutions dating back long before the Nile Perch boom. Seven were vessels originally registered in Kenya, which followed the factory owners south along the lake shore in the early 1990s, while the remaining were locally-built and privately owned in one case by a factory. Some of the latter were designed for furu trawling and a few since the late 1980s for Nile Perch. Unlike with the marine trawler fleet, there was no special central registration and licensing system for the vessels, except that their existence had to be notified to the Mwanza RFO as well as to a District Fisheries Office (DFO). Prior to the ban the only regulations applied were that trawling was restricted to waters of 20 m or more and to daylight hours. The ban had been announced in an amendment to the Fisheries Regulations made as early as October 1994. This amendment appears to have been a trade-off for the 'adoption' of fisheries institutions around the lake by the United Nations Environmental Programme, but the exact mechanisms of this process are not known. Nor is the precise reasoning behind the ban, although in a general sense it obviously stemmed from concerns about overfishing. Actually, the nature of the trawling possible before the ban was restricted not merely by the regulations but also by the nature of the vessels involved and the local infrastructure, and this made its actual contribution to any generalised overfishing doubtful. Without space for processing, cold rooms or crew accommodation, the range of the trawlers was limited to a tour of a day; the state of the roads further meant that their anchorage could not be much more than 100 km from the factories. As a result, the main trawler grounds were in the Speke Gulf in the lake's south eastern corner, about half an hour's sailing time from the anchorage of Nyamikoma, Magu district (about two hours by road from Mwanza and a little more to Musoma). When operating, the trawlers all delivered direct to factory jetties. Functioning factories directly owned four of them. The operators of the others (which included a factory under construction) mostly had semi-formal contracts with specific factories; only one 'sold to anybody'. The trawler ban was not strongly opposed by the factories as a whole, although it was clearly an inconvenience to them. The lack of opposition was probably related to the relatively low share of trawler catches in their total throughput, and possibly to the expense of running them. Average daily catches were mostly in the range 1 - 1.3 t/day/vessel in 1995, or equivalent to 7300-9500 tons/year for the whole trawler fleet. Government owned 'research' vessels, of which there were three operating out of Mwanza, were exempted from the ban and as they had done before it was introduced - continued to supply the factories. At the end of February 1996 the vessels owned by the parastatal TAFICO were also exempted and one private operator told me that he also anticipated an exemption shortly. This issue will be returned to later. There was one more or less unique by-product of the trawler fishery. The four inch trawler nets which were being used tended to clog after about 500 kg or so had been caught and the subsequent catch was indiscriminate, with a high by-catch content. The commonest saleable by-catch was juvenile Nile Perch, of around 8 cm, locally called jowa. Around the trawlers' main anchorage a localised trade in jowa developed prior to the trawler ban. The presence of larger trawler catches of jowa also provided cover for it to be traded by the owners of illegal small-mesh beach seines operating in the same area. Jowa was 'barbecued' on the beaches by the trawler crews and net operators by being laid between layers of grass which were then set on fire. Bicycle traders buying 40-80 kg were the main customers; for the most part they were carrying the jowa to internal markets in Magu where it was resold in small quantities. The remainder of the adult Nile Perch catch was accounted for by the 'artisanal' fishing fleet. The size of this fleet is unclear. According to the Mwanza Region fisheries census in 1995 there were a total of around 4000 registered and unregistered fishing vessels active in the region, excluding Geita district, where there were possibly another 500-1000. Assuming vessels to be distributed in roughly the same proportions as those assumed for fishermen, this implied perhaps 3000 craft involved in the Nile Perch fishery in Mwanza region and perhaps 3500 on the Tanzanian part of the lake generally. A supplementary source of information about the Nile Perch artisanal fishing fleet in Mwanza is provided by the boat registers of the Mwanza DFO, which could be inspected as far back as the beginning of 1992 commonly said to be the starting point of the local boom. Perhaps because Mwanza is the town with a DFO most frequently visited by vessel owners, and/or perhaps because many owners living in other district of Mwanza region are not particularly anxious for the extent of their ownership to become local public knowledge, a disproportionate number of vessels in Mwanza Region are registered here. Between the beginning of 1992 and the end of 1995 there were 1450 vessels registered at Mwanza DFO whose primary purpose was clearly fishing or collection, or more than half of all the vessels which in the census are recorded as registered ones. 790 of these 1450 were vessels equipped with gillnets, and a further 103 were vessels equipped with other types of gear specific to the Nile Perch fishery. A further 129 were collection vessels. Assuming that there was one unregistered vessel entering the fishery for every two registered ones (the proportions pertaining to all fishing vessels in Mwanza region, according to the census), at least 1500 vessels had entered the fishery in the previous four years. The peak year of recorded entry was 1995, when no fewer that 285 vessels carrying gill nets, 25 others with Nile Perch-specific gear and 61 collector boats were registered. 8.6 per ent of vessels carrying gillnets which registered after 1992 were motor powered; in 1992 the proportion of all lake fishing vessels with outboard motors in Wilson's survey had been only two per ent (Wilson, 1993). Almost all the collector boats were powered. From the outset gillnetting has been by far the commonest fishing method for Nile Perch. In 1987, when Reynolds and Greboval (ibid.) visited the Tanzanian part of the Lake, most fishing operations consisted of a single paddle-powered planked canoe carrying between 10 and 30 (occasionally up to 50) six inch nets of about 90 m unhung (c 40 m hung), operated by a crew of two or three. Vessels carrying smaller numbers of nets and crew tended to be dugouts, those carrying larger numbers were all constructed from plank. Since manufactured nets were still largely unavailable at this time, gear tended to be made from unravelled polythene sacking or unravelled tyre thread. Nets were set in a chain in late evening, then returned to and hauled early the next morning. Catches were so high that no particular strategy was followed to increase them. At this time, fishing effort was mainly confined to the 15-21 darker nights during the lunar month. When Ward et al.(1994a) conducted fieldwork in Ukerewe in 1994 the typical net size had fallen to five inches and most vessels observed at the landings were now plank boats carrying 40-70 factory-produced nets. A significant element of the fishery was no longer based in the closest onshore areas and sail power was now frequently employed. Crews had begun to stay on the lake overnight with their nets. A concentration in ownership was observed but Ward et al.did not comment on how operations were coordinated between groups of vessels owned by the same tajiri. By 1996 mesh sizes of the same magnitude were still being used but further changes could be observed in most other aspects of gillnetting. It was clear that a strong differentiation had taken place in the fishery and that there had been both an extensification and an acceleration of the intensification of operations as they were observed by Ward et al.(1994a). There were still some hundreds small-scale operations in the fishery, with single dugout canoes (mitumbwi) and perhaps 10-20 nets. It is not clear however if these were operated by the same people referred to by Reynolds and Greboval. More likely, a large proportion of them were probably capital-poor new entrants 'encouraged' by the example of the growth of larger-scale operations. By now a clear majority of operations consisted of one or more planked canoes. These operations were differentiated according to how many plank boats were operated and how many nets each carried. A middle sized operation consisted of three or four vessels, a large one anything more than this number. Most seemed to be now carrying upwards of 60 nets, invariably hung double in order to entrap Nile Perch swimming at lower depths. A significant number of vessels, particularly those in the larger fleets, were carrying more than 80 nets and many were carrying more than 140. Double hanging was said to have been introduced in 1994. By 1996 some vessels had adopted triple or even quadruple hanging. Increases in numbers of nets used meant that bigger crews and larger vessels were necessary. Most of the newer planked canoes which had entered the fishery since 1994 were upwards of 8 m long and had crews of four men. The largest were up to 10 m long. Where multiple vessel ownership existed, sail power had been increasingly supplemented by the use of outboard engines. Boats equipped with the latter were used less for fishing itself and more for towing canoes to the fishing grounds and for patrolling to prevent theft of nets and fish. Net theft is reported by Bituro (1974) to have already been rife in the Tilapia fishery in the early 1970s, but had reached very serious proportions by 1996. One reason for this was clearly increased opportunities, for nets were being left overnight every night of the year, as well as in some cases for considerable periods of daylight. Declining unit catches and larger fleets of vessels had stimulated the use of more systematic and less intuitive fishing strategies. Fleets entering a new area would typically start a fishing cycle by stationing sets of nets at a considerable distance from each other. According to whereconcentrations of catches were detected at particular points in this 'picket', the fleet of nets would be redeployed according to calculations about patterns of fish movement. Depending on the net fleet size, net chains of a dozen kilometres or more had become common. Next to gillnetting, the most common fishing method followed in 1996 was beach seining. According to the 1995 Mwanza Region's Fisheries Census there were 324 Nile Perch seines operating in the region, not including Gieta the district which probably had the highest concentration. Seines were banned at the same time as trawlers, but unlike in the latter case, attempts to enforce this ban were comprehensively and publicly abandoned almost immediately (see below); by the time of my fieldwork beach seine owners made no effort to hide them or stay silent about them even during visits by Fisheries Department staff. Unlike gillnetting, beach seeing for Nile Perch was mainly carried out between 2 pm and 11 pm. Most seines I observed were 300-400 in length with meshes of about an inch and a half at the cod end. They were shifted from beach to beach on a daily or 3-4 times weekly basis and set in the water using one or two canoes. All were setting and hauling more than one a day and some smaller ones were doing so four or five times. The smallest hauling crew I saw was eight men, the largest upwards of 30. It seems that, locally, the third most common method of fishing for Nile Perch, long lining, was becoming more popular at the expense of gillnetting. According to its practitioners this was mainly as a response to the prevalence of gillnet thefts, but more obvious reasons were that long lines had far lower capital costs and were highly productive. Long lines whose operations I observed were typically around 4-6 km in length, with up to 2000 hooks baited with live furu and set up at 2-3 m intervals along the line. The lines were typically set in a stationary position and left overnight, although it was quite common for a second setting of them to have been introduced, particularly in the middle of the long rains, 'when fish tend to die in the water'. Setting and hauling was by paddle or sail-powered canoes with crews of three. I also heard of long lines being trolled from moving vessels, but never observed it. The practitioners of every single method complained of declining unit catches, despite the substantial intensification of production methods and the greater mobility of much of the fleet , whose effective range where camps were being used had now increased to perhaps 12-15 km from the mainland shore. Hard evidence about catch levels in earlier periods is hard to come by; only Ward et al. (1994a) seem to have collected any systematic daily catch data and this was limited to mwalo in Ukerewe. The catches thus reported were 240-300 kg/vessel/day during the rainy season and 57-90 kg/vessel/day during the dry season, from vessels with between 40 and 70 nets. I personally interviewed around 25 vessel owners at the end of the inter-rain dry season and the first week of the main raining season in 1996, concerning catches from vessels using between 40 nets at one end of the scale and 200 at the other. These were getting daily catches varying between 10 kg and 250 kg/vessel. Most interviewees' catch levels fell into one of four categories of roughly equal size, corresponding to around 10-20 kg/vessel, 20-40 kg/vessel, 50-60 kg/vessel and 80-100 kg/vessel respectively. Perhaps an overall average was 40 kg/vessel/day. Those catching over 50 kg/vessel/day all had 80 nets or more. The interviews were divided evenly between Ukerewe and Geita/Sengerema. Although the sample is extremely limited, the data does seem to point to a clear downward trend. Three beach seines owners were interviewed in Geita/Sengerema at the beginning of the main rainy season. Their seines were fairly constantly returning catches of around 200 kg/haul. Long line boat owners I interviewed were getting catches in a range of 80-200 kg/vessel/day, at the end of the inter-rains dry season and using one setting a day. No comparable data is available from earlier periods for the last two of these fishing methods. [ Making the conservative assumption that of the c 3500 vessels engaged in the Tanzanian Nile Perch gillnet fishery in 1996, 2500 (including a total factory fleet of perhaps 300) were plank boats with catches of 40 kg/day over a year of 330 fishery days and 1000 were dugouts with catches of 5 kg/day over a year of 250 fishing days, the total gillnet catch would have been about 34,250 tons/year. Assuming the operation of 300 beach seines with average catches of 100 kg operating 350 days/year and 200 planked long line boats with catches averaging 60 kg/day operating 330 days/year, the total ‘artisanal ’ non - gillnet catch would have been around another 14,500 tons. Assuming the continued operation of the publicly - owned 6 trawlers with an average catch of 1.15 ton/day, operating 350 days/year, the total trawler catch would have been 2600 ton/year. This gives a total outtake of about 51,350 tons/year, slightly less than the estimates currently used by the Fisheries Department (but whose basis I never discovered). The factories accounted for a minimum of 80 per cent of this outtake (the majority of the remainder representing fish the factories declined to purchase).] 5.2Trading Functions Most people I spoke to in the fishery thought that a majority of boats were tied to the factories in terms of supply of gear and counter-supply of catch. My own feeling was that this perception probably rested on a slight confusion. The great majority of fishermen predominantly supplied the factories, one way or another, but the factories tended to offer significant sponsorship only to fishermen who were 'established', i.e., who already had at least one vessel and who could use the sponsorship for acquiring the gear for a second. By the same token, almost every tajiri I encountered who owned more than one gillnetting vessel had received some nets from a factory. Sponsorship normally took the form of providing partial sets of nets, and occasionally (to those already owning a few vessels) engines. Nets seemed to be most commonly (though not only) distributed in units of about 10-50/vessel, the latter amount equivalent in value to Tsh 1 m (USD 1800). Systems and strategies for distributing them will be examined later. 15 HP engines cost around the same amount, 25 HP ones half as much again. These were distributed to an elite ofmatajiri with already proven records of fish supply, either to enable them to expand their direct fishing capacity, or for towing fishing vessels (for which the larger size engines were required), or for collecting fish from other fishermen, or for all three. In the cases of both nets and engines matajiri were expected to repay in delivered fish; the usual system was for 50-60 per cent of the latter's purchase price to be withheld by the factory as repayment of the principal, although there were also instances of contracts with fixed-sum repayments (also denominated in fish). Formerly at least no interest was levied, despite an annual inflation rate of 30-40 per cent. Table 6 Distribution of ownership of vessels in the Nile Perch fishery (excluding trawlers and collection vessels) registered January 1992 - March 1996
Source: Data base derived from Mwanza DFO boat register Table 7 Nile Perch vessel owners registering vessels involved in other activities on Lake Victoria (excludes factories), January 1992 - March 1996
Source: Data base derived from Mwanza DFO boat register. In all cases loans were made in kind rather than cash, although presumably a tajiri could have sold his gear or engine in cash if he had been so inclined. I never encountered a case of capital being advanced for the purchase of a boat, on the other hand. New planked fishing boats varied in cost between Tsh 0.15 and 0.3 m (USD 270-550) depending on size and the wood used; I was unable to establish a reliable price for a dugont. I also encountered no cases of factories sponsoring either beach seine operators (whose capital investment for a 400 m seine was in the range of Tsh 1 m (USD 1800) or long liners (cost of a 2000 hook line about Tsh 50,000 (USD 90). It was unclear to me why this type of investment was not undertaken, either by the factories or by independent collectors. As Tables 1-3 above indicate, there were only two significant buyers of fresh fish: the factories and their agents, and individuals sending frozen or chilled fish by air or train to Dar es Salaam. Supply shares of fish destined for freezing were 95 per cent or more in favour of the factories. Besides being supplied by trawlers, the factories in 1996 were obtaining fish from their own 'artisanal' fishing fleets, secondly from tied and untied vessel-owners either direct to their jetty or via a system of tied collectors and collectors' agents (some of whom were also tied fishermen), and thirdly by independent collectors. Each of the different factories pursued a different strategy with respect to their specific combination of these methods. In particular there was considerable variation in investment in 'own' fishing operations and in the nature and status of collectors whom the factories engaged. After the trawler ban, while no factory procured anywhere near a majority of its catch from its in-house 'artisanal' fishing effort, four were supporting in-house fleets of various sizes and compositions. The largest comprised 113 registered vessels, almost all constructed in the factory's own boatyard, while another had 53 registered vessels distributed between three camps in Ukewere, Geita and Muleba. The bulk of all the factories' output came from their collection systems. In most cases these were radically decentralised, although each of the factory's had a chief agent (karani, pl. makarani) based at the factory jetty who was responsible for lake-based intelligence, strategic coordination (including making recommendations about price), distribution of working capital and buying from small independent collectors operating close to the factory itself. Decentralisation was usually to a number (5-15) of field agents, distributed between the different fishing grounds and operating from stations equipped with large collector boats and ice. These were responsible for the selection and location of sub-agents, allocation to them of working capital and transport (including smaller collector boats), making decisions about whom of the local matajiri should receive loans (and if so, to what amounts), buying from small local independent collectors and bulking and dispatching loads to the factory jetty or gate. The sub-agents were responsible for most 'front line' collection, both from own tied vessel owners and from anyone else they could buy from. They might deal regularly with 20 vessel-owners, implying loads of perhaps 0.5-0.6 ton/day and receipt of daily advances of Tsh 0.2 m (USD 363). Field agents would normally be in charge of 3-5 sub-agents. Such operators were sustained by infrastructures of more or less extensive collector boat fleets, ranging between a handful of vessels and almost 30, as well as smaller fleets of pick-ups and trucks. The only factories with collection fleets of less than five vessels were collecting primarily from Speke Gulf area by road (the only fishing ground served by a tarmac road from Mwanza). Collector boats generally varied in size between 10 and 15 m, with fishholds of 1.5 to 10 tons, and had engines up to 48 BHP. In most cases they remained the property of the factory. All agents and sub-agents worked off capital advanced by the factory and received instructions about maximum price. Systems for renumerating them were many and varied and ranged from implied commissions (the sub-agent being paid by simply keeping the difference between the instructed price and whatever he had to pay to get the fish) to actual commissions, to mixtures of wages, commissions and bonuses. Actual commissions seemed to be the commonest system. Private individuals with large fleets of their own, most of them had been sponsored by at least one factory in the acquisition of nets and/or engines, tended to be dealt with separately from smaller sponsored and untied matajiri. They might have longer-term and more favourable price contracts and/or double as semi-independent collectors responsible for a specific area around their kambi. The largest factory in Mwanza made a point of sponsoring larger matajiri to become independent collectors, by allowing them to buy back not only the gear and engines advanced them, but collector boats themselves. Sponsoring collection by matajiri rather than through specialized agents was indeed the main collection strategy of this factory. Larger matajiri would normally deal directly with field agents, or in the case of the largest, with factory makarani. In Mwanza region there was only one large-scale independent collector who was not a tajiri. This was a former trawler-owner and air freighter. He was renting two large collection boats and buying 6-7 tons/week from his own network of matajiri and small collectors around the Lake. He had delivery contracts, one written and one verbal, with two different factories. This was also the case with some of the large matajiri-collectors, although it was probably more common for them to engage in serial than multiple contracting. The much smaller independent collectors referred to were mostly bicycle and karua (small planked passenger/cargo vessel) traders. It was widely agreed that their number was dwindling rapidly and that they were found in particular niches rather than in competition with more organised elements. The most important such niche was Mwanza Gulf itself, close enough to factories to allow easy and cheap delivery but no longer with enough fish to justify concentrated investment in production. Here several motorized small karua traders, a number of bicycle traders (including one combined operation of four traders with three cycles) and a couple of independent pick up traders operated out of mwalo like Nyegezi and Kikongo. Another niche was mwalo where the dominant fishing methods were non-sponsored ones like beach seineing and long lining. Such mwalo included Mchangane, near Nkome (Geita) and bays and inlets along the northern shore of Speke Gulf. Independent collectors around Mwanza and Speke Gulf tended to deliver directly to the factories and to factory field agents based in Nansio town respectively; those operating in Mchangane took the fresh fish direct to the mining areas nearest to Geita. Finally there was still a handful of traders buying fresh fish at Kirumba and reselling it to the factories. The amounts in question were not large; I was unable to establish from whom it was being obtained. With the exception of a couple of the air freighters (see below), nobody except the factories themselves were in a position to undertake significant sponsorship of the fishing fleet. Independent collectors and the lowest rang of factory collectors nevertheless often did offer sponsorship of a kind to matajiri not contracted to the factories via the supply of nets and engines. This took the form of providing small cash advances or advances in kind of consumption goods. Such advances in fact seemed to be extremely widespread. As indicated, the trade with Dar es Salaam in frozen Nile Perch was longer established than the export trade. Reynolds and Greboval (1989: 59) report that in 1987 'air freighters' were buying fish gutted and cleaned at Mwanza Soko la Kati (central market) and having it frozen (or, more probably, chilled [ Recommended deep freezing temperature for fish is - 30 ° C or below. If properly packaged and handled, such fish will not visibly deteriorate for a few days after leaving the freezer. Chilling involves reducing the fish ’s temperature to only 0 ° C and involves less expensive plant (e.g. home ‘freezers ’ (in reality, home cold stores)) and the consumption of less energy. If a fish is to be transported by air and arrives at its destination within a few hours, chilling should be sufficient to prevent its visible deterioration. Mwanza air and rail freighters also used the category ‘soft frozen ’ to designate something between chilling and ‘hard freezing ’.] ) at the National Cold Chain Organisation (NCCO) plant before sending it tothe capital. Actually, probably the largest single 'air freighter' of this time (and for a considerable period after) was a partnership of three local businessmen who were obtaining fish from three wooden trawlers which they owned. Somewhere along the line this group seems to have acquired a lease on the NCCO's freezing plant, which they sublet to other air freighters on a contract basis when not using it themselves. The partners had their own outlets in Dar es Salaam, at the old NCCO plant there and in the fresh fish section of Kariakoo market. By the time of a study by Ward et al. (1994b), the partners seem to have dropped out of the trade except as letters of freezing and chilling space. The bulk of the air freighted trade now comprised fish bought at the Igombe mwalo half an hour by vehicle from the airport. The trade was now monopolised by individuals connected to Air Tanzania Corporation (ATC)'s Mwanza operation, supplying a small number of regular customers in Dar es Salaam. According to Ward et al. (1994b) the main outlet was more formal butcher-cum-fish shops around the city; they put the trade's size at somewhere between 400 and 600 tons/year. In 1995 officially recorded freight cargoes of frozen/chilled wholefish to Dar es Salaam amounted to 520 tons, averaging 1.6 t/consignment. A further 155 tons, in slightly larger average size consignments, was sent from Mwanza to Kilimajaro International Airport (KIA), midway between Arusha and Moshi towns. Some fish was probably still bought at Igombe mwalo. Air freighters were important customers for many matajiri here and one I interviewed had been tied to an air freighter by the supply of an engine. Others had been promised engines and it was said that sometimes 15 tons/day was being delivered to this individual. Other air freighters were obtaining catches at the islands simply by offering significantly above the market price. It was not completely clear where the fish sent by this route was now being frozen/chilled/stored. The old NCCO plant, which was taken over by the fisheries parastatal TAFICO in 1991, was still the main source of rented blast freezer space in Mwanza and in 1995 worked at full capacity (2 tons/day). [ TAFICO was charging Tsh 40/kg for freezing.] However, it was clear that most of its customers were not air but 'rail freighters', since sending fish by rail to Dar es Salaam involved a journey of at least two and a half days and therefore required something resembling 'hard' freezing. [ Actually the TAFICO blast freezer could not get down to more than - 20 ° C, with obvious implications for the quality of the fish frozen there.] Both freezing and chilling space was said to be also available for rent at the Mwanza Refrigeration Company, which must have had a capacity of something like 800-900 tons/year for all the small-scale freezing and chilling trade to have been covered, and which must have been used by the air freighters. At the ATC office in Mwanza I was allowed to examine all the counterfoils for air cargo between Mwanza and Dar es Salaam issued in 1995 (total figures and average consignment sizes were derived from this source). The names of the senders and receivers of all frozen wholefish consignments between 1st February and 30 June were recorded and analysed. The 121 consignments between Mwanza and Dar es Salaam involved only 14 senders and 13 receivers; the top five senders jointly accounted for 81 of them and the top five receivers for 93. The 37 consignments between Mwanza and Kilimanjaro involved only three senders and two receivers. ATC staff confirmed that all the senders still worked for the airline or the airport authority. The characteristic second-hand cardboard cartons, held together with string and spilling sawdust, used to transport Nile Perch from Mwanza by both air and rail, [ These parcels, each around 150 kg, are known as bayi .] could also be observed being loaded onto flights of carriers other than ATC (e.g. Air Afrique) bound for Dar es Salaam. It was not clear whether these were processed at the ATC office or not. The estimate of a rail traffic of 800-900 tons of frozen Nile Perch from Mwanza is based on an analysis of counterfoils of Owner's Risk Consignment Notes issued for parcel luggage carried on passenger trains from Mwanza town railway station from January to April 1995, when a total of 213 tons of frozen fish was carried. These were mostly in the form of slightly smaller consignments than the air freighted Nile Perch. 80 per cent by weight was destined for Dar es Salaam, the remainder for various points along the Central Line up to and including Dodoma - but mostly for stations close to the gold fields around Kahama (Shinyanga). Common sense suggests that a large number of the parcels were sent by staff of the Tanzania Railways Corporation (TRC). Because of the admitted or suspected involvement of ATC and TRC staff in these trades, it is probable that the volumes carried have in both cases been understated. Ward et al.'s (1994b) study reported the control of the Dar es Salaam market for airfreighted Nile Perch by five leading 'receivers', just as my own analysis of weigh bills showed. Ward et al. went onto trace the activities in Dar es Salaam of these receivers. The largest owned three butchers' shops as well as some home-based cold storage equipment and a pick-up. He rented cold storage at the airport and distributed the fish to his own outlets and to the places where the other leading receivers stored their fish. Some of these also owned butchers' shops. The receivers traded most of the fish they obtained on a wholesale basis however. Ward el al. (1994b) state that a total of 15 butchers' ships around the capital were supplied in this way, as well as some large-scale frying operations in Kariakoo itself, who bought fish the wholesalers considered too 'stale' to sell on. My own fieldwork turned up a different, though also very small, group of receivers controlling most of the Dar es Salaam market for rail-freighted 'frozen' Nile Perch. Five of them were based in the so-called fresh fish section on the ground floor of Kariakoo market hall. One whom Iinterviewed had a mixture of owned and rented domestic freezers with a total capacity of around 1.3 tons and employed a supervisor (a qualified butcher) and two assistants. She was turning over around 3.5 t/week which was ordered by phone from a regular supplier and collected at Dar es Salaam railway station with a hired pick up or, for small consignments, a taxi. She was selling most of the fish as 'fresh', after allowing it to thaw. Until 1994 her biggest customers were locally-based Zaireans who took the fish downstairs to the Kariakoo blast freezer where they paid for it to be re-frozen (!) before being dispatched to an unknown destination. These days she was selling mainly to 'suburban fishmongers' in weights of 30-50 kg. Private individuals were also buying in Kariakoo, in quantities of 0.5 to 5 kg. 5.3Factory Processing and Export Functions The factories operating in Mwanza in April 1996 were with one exception located along the lake shore, either close to Mwanza South docks and railway yard or off the road to the airport north of the town, at Ilemera. Only two of the seven operating were purpose built; the remainder occupied old godowns or textile plants, although most either had or were building new extensions. Plant mostly comprised plate and blast freezers, cold storage, ice-making equipment and generators. One plant, which dealt only in chilled fillet, had only cold rooms. Total rated freezing capacity was perhaps 120-150 tons per 24-hour period, giving an overall capacity utilisation rate of less than one-third. However there was considerable variation both in capacity and in utilisation. One plant had a capacity of over 40 t/day and was working at over 60 per cent capacity; in the second rank came three plants with capacities in the range of 20-25 t/day, which were each working at less than half capacity. One of the smaller plants was also working at more than 60 per cent. The lower capacity utilisation ratio appeared to be mainly an effect of choices about length of the working day rather than shortage of inputs (most of the factories had taken deliberate decisions to work only one shift). All the factories produced ice, but no calculation was made of total production. Ice seemed to be used only by the larger factory collection operations and was not really distributed through the chain. The plants had virtually identical labour processes, beginning with unloading and proceeding to selection, washing with chemicals, skinning, gutting, filleting and removal of mabondo, cleaning the fillet with a chemical wash, trimming, washing for a third time, packing and then freezing or chilling prior to storage and export. Employment inside the plants varied between 80 and perhaps 400. After completing export documentation and customs clearance (organised through Mwanza RFO) the factories were exporting to northern hemisphere countries and Australia, by four different routes. Each of the four Kenyan-owned factories was ferrying its fish in refrigerated containersby road to Mombasa, where it was loaded onto container vessels. The two Tanzanian-owned plants were ferrying their fish in refrigerated containers on flatbed rail wagons from Mwanza South to Dar es Salaam port. One of the Kenyan-owned plants was also sending chilled fish by air (ATC) to DSM for airfreight export to Europe. [ This trade was not counted in the figures given earlier for the domestic airborne trade between Mwanza and Dar es Salaam.] The remaining plant, owned by a European expatriate, had been sending chilled fish by the same method but using a different carrier (a private Zairean-owned airline). When the latter company ceased operating it was unable to get space on ATC and began sending by road to Dar es Salaam airport, but using the tarmac road via Musoma, Kisumu, Nairobi and Arusha rather than the direct unpaved route. By the time the factories opened in Tanzania, an international market for Nile Perch had already been established. Operators of the Tanzanian plants told me that from 1992, 'the market made itself. Enquiries came to us rather than the other way round'. Particularly decisive was a temporary shortage of Atlantic Cod in 1995, which pushed the international average price of frozen Nile Perch fillet almost to USD 4/kg, f.o.b. Subsequently it fell to around USD 3.3/kg by April 1996 (chilled Nile Perch averaged nearer USD 4.2/kg). On the other hand these average prices concealed sharp variations between markets in different parts of the world and for different types of customer. This point will be returned to. Table 8 Distribution of recorded Nile Perch fillet exports by destination, Kenya 1987 and Tanzania 1995
Sources: 1987, Kenya: Reynolds and Greboval (1989:18); 1995, Tanzania: booked exports for royalty-payment purposes, Mwanza RFO (* excluding Musoma). Table 8 shows Nile Perch's apparent conquest of increasingly mainstream markets in the years between 1987 and 1995. One contributing factor to the demise of Israel as the main destination was that Israeli vessels ceased calling at Dar es Salaam. A considerable proportion of the Nile Perch arriving at the three currently dominant destinations - most obviously Gibraltar - probably ended up being reexported, so no very firm conclusions can be drawn. The exports bound for Gibraltar were accounted for by a single factory, which only began producing in bulk in the last five months of 1995. Exact information on total investment by the factories was hard to come by, except in the cases of three purpose-built plants currently under construction. Excluding fish collection and transport investment operations, start-up investment in these three plants ranged between USD 0.6 and 3.5 m. The largest of existing plants, which had just been considerably expanded, was the only one to remotely approach the upper end of the scale. This plant was said to have had new machinery installed from the outset. In the remaining, smaller plants, there tended to be a mixture of reconditioned as well as new machinery. Some of the Kenyan-owned plants had developed their Tanzanian operations while reequipping at home, transferring their old machinery to the Tanzanian branches in the process. Other plants had started with reconditioned machinery, but as their profitability had been demonstrated, had invested in new machinery. Total investments in the existing plants varied between perhaps USD 0.2 m and USD 2.5 m with the majority falling in the upper part of the range. Besides plant, the factories had to invest in fish procurement and transport, local and long-distance. Investment in fish procurement included investments both in fishing vessels and gear and in collection; if investment in self-built sail-powered fishing vessels with gear was around USD 3000/unit and if investment in self-built 25 BHP outboard-motor powered collection boats was around USD 4000/unit, then the factories had each invested up to around USD 0.15 m in their current in-house collection fleets and USD 0.45 m in their current in-house fishing ones. On the other hand, engines had only a limited life on the lake, particularly in the hands of collectors. One factory, which had a collection fleet of around ten vessels and an in-house fishing fleet which included about 20 powered boats, and which had sponsored about 40-50 artisanal fishermen (not all with engines, but some with more than one) since opening in 1992, had been obliged to import around 150 engines in all since this time at a total cost of around USD 0.45 m. Cumulative investment in these fish procurement methods may therefore have been around 50 per cent higher than the figures given above. Investment in gear and engines for resale to artisanal fishermen on hire purchase cannot really be considered as direct investment by the factories. Rather, where the loans were repaid, they were direct investments by their ultimate purchasers and a long-term element of working capital as faras the factories are concerned. Default on loans made by specific factories ranged between zero and 30 per cent (these figures could be generalised to working capital generally, for default rates by agents buying on working capital advanced by the factories was broadly similar). Assuming on the basis of Table 6 that almost a quarter of the artisanal plant boat fleet targeting Nile Perch were part of private fleets with two or more vessels, and that the total plank boat fleet was around 2500, then over 600 vessels were part of medium or large-scale fleets. Assuming all of these to have been sponsored (on the principle that they were all probably added to the fleet by fisherman 'established' prior to 1992), that they were sponsored to an average of 30 nets/vessel and that 10 per cent (65) were also sponsored with engines, then the long-term working capital injected by the factories into the lake was probably in the region of a further USD 0.9 m. The lowest long-term working capital investment of this kind by a factory was worth around USD 40,000; the highest perhaps ten times this amount (i.e. being well over 200 vessels). These sums need to be discounted somewhat in order to take account of various remissions on import duties and taxes enjoyed by the factories, however. Some factories had invested in road-based collection, although most of these tended not to be the same ones as had made heavy investment in fishing and water-based collection. Road collection 'fleets' averaged about four vehicles, generally pick-ups or three ton trucks with total values of perhaps USD 60-70,000. Because of the appalling state of roads in the area, the life of these vehicles was not long and most factories had to replace about a quarter of their fleet every 18 months. The factories using road transport to the ports also had their own long-distance transport fleets, comprising articulated flatbed units for carrying refrigerated containers. In some instances the factories owned the refrigerated containers too, in others they were rented from the shipping companies. An articulated flatbed unit capable of taking a 20 ton container cost perhaps USD 50, 000 new; some factories had five or six. The absolute minimum short-term working capital a factory required was perhaps the equivalent of 100 tons whole fish (at early 1996 prices of USD 81 c/kg), USD 80 000) plus the costs of collecting this amount and of processing it. On the basis of the above calculations, total investments (excluding buildings) by the generation of factories operating in 1996 probably ranged between around USD 0.5 m and around USD 2.5 m; while combined short- and long-term working capital requirements ranged roughly between USD 0.12 and USD 0.5 m. 5.4'Artisanal' Processing Functions and Trade in 'Artisanally' Processed Fish As already indicated, the trade in artisanally-processed Nile Perch has two quite distinct sources: processing on the islands and on lakeshore mwalo away from Mwanza, of fish not purchased by the factories; and processing in and around Mwanza town of fish or parts of fish that factories had purchased and then discarded. The first was easily the larger of these trades. There were two basic artisanal processing methods for Nile Perch on the islands. For most of the 1980s, and possibly in the early 1990s, too, the most important was smoking, to produce sangara moshi. The labour process involved removing guts, mabondo and scales, then splitting into three pieces and washing before smoking for 8-16 hours under a papyrus mat in multi-rack kilns built of earth and usually set on a mound. Kilns were of various lengths and might be roofed or open to the elements; they were all more or less temporary structures. Earlier, fish of all sizes were smoked, but by 1996 smoking was mainly of smaller-sized fish (2 kg or less), which were said to be more suitable. The product is brittle and easily damaged during transport. It has a shelf-life of around a month. By the time of my fieldwork the most popular artisanal processing method had become drysalting, to produce kayabo (known in Dar es Salaam as sangara chumvi). This form of processing is not mentioned in Reynolds and Greboval (1989). The labour process involved the same initial preparation as sangara moto except that washing was with a wire brush rather than by dipping. After this, incisions were made for salting and salt applied, on average on a 20 per cent by weight basis. The poorer the condition of fish, the more salt was used. More salt was used also in the rainy season, when risks of deterioration increased. The fish was left overnight to 'absorb the salt' and then dried for about three days on makeshift cypress wood table frames with tops set at an angle and covered with papyrus mats. Kayabo travels well and has a shelf life of up to three months. The fish 'artisanally' processed on the islands was fish which the factories were unable or unwilling to buy. Inability to buy generally implied breakdown of collection vessels or running out of ice, or cutting short of journeys because of bad weather, etc. Unwillingness to buy referred to fish size and/or quality. The factories were not interested in fish of below 2 - 2.5 kg, since these were unsuitable for filleting. Secondly, agents on the lake rejected any fish suspected of having started to deteriorate. Agents and sub-agents were simply not paid for fish which were rejected at the next step up the chain, although they would have themselves paid for them from their advances from the factory. Hence there were strong incentives to reject fish of doubtful quality. Fish not bought by the factories were generally sold to processors operating in the camps themselves or close to landings, although a few were retained for feeding camp members themselves or given to various categories of camp 'helpers' as mboga (literally 'side dish', but in this context a token or supplementary payment). Processors fell into three main categories. Firstly there were those - on various scales of operation - who were based most of the time at a specific landing or permanent camp. Secondly there were those, this time mainly small-scale, who were based more or less permanently at inland, market centres, who came to specific landings on a more or less regular basis to buy and process in situ before returning with the processed fish to 'their' market centre. Thirdly there were those, usually on a larger scale and found overwhelm ingly in the kayabo trade, who shifted together with the larger more mobile and more productive camps from one location to another. All processors bought by the piece rather than by the kilo; in March 1996 an average sized piece cost Tsh 500 (US 90c); some large pieces were fetching up to Tsh 1000. 'Artisanally'-processed fish was also always sold by the piece. In respect of kayabo there were clear rules allocating processed fish to one of two quality grades and three (Mwanza) or four (Dar es Salaam) size ones, but no such clear guidelines existed for sangara moshi, partly because there was greater size standardisation for the type of fresh fish considered appropriate for smoking. Rejection rates by fresh fish collectors visiting the camps were said at most places to be roughly 10 per cent. Processors in the fishing grounds perhaps 'captured' a further 5-10 per cent of total catches, from purchases from untied fishermen in areas where the factories' presence was weak and from fish 'accidentally' uncollected. At any rate, there were a surprisingly high number of processors everywhere, particularly at the larger camps where there was usually one processing operation for every two vessels. Since a large proportion of processors were also involved in mobile trading of one kind or another, they were usually obliged to hire assistants. In all there were perhaps one processor or processor's assistant in the fishing grounds for every four or five fishermen. Of these at least three quarters were working with kayabo processing. The only area of the Tanzanian part of the lake where smoking still seemed common was at the mainland landings (not the islands) in the extreme south west corner of the lake, although most larger camps elsewhere also usually had a smoker or two. Evidence of the decline of smoking was graphically visible in many places. The lakeshore was dotted with crumbling kilns, and even in those areas where smoking was prevalent there were so many kilns vacant that those wishing to use them did not have to pay rent. Two main reasons were generally given for this decline (and the unresponding rise of kayabo). Firstly, as already indicated, smoked fish did not travel well and were unpopular with traders because of losses in transit. Secondly, demand for smoked fish appeared to be much more price sensitive than for kayabo. The effect of increased competition for fish in the lake had led to a steep increase in the price of smoked fish, which in turn appeared to have almost eliminated it from the diet of the wide swathe of previous purchasers in the lake's hinterland who had adopted it over the previous decade. On the other hand the market for kayabo, which was mainly either in neighbouring countries or in traditional fish consuming areas of Tanzania which were well away from the lake, proved to be comparatively price inelastic. The decline may have also been influenced by therising price of firewood, although outside of Mwanza relative to salt the former was still extremely cheap. [ In some places, e.g. western Ukerewe, where wood was still plentiful, a m 2 of firewood could still be obtained for Tsh 1000. This would have been enough to smoke 500 fresh whole pieces (roughly one ton of fish). In Nkome (Geita), where wood was harder to come by, the same amount cost Tsh 3000. To produce kayabo , one ton of fresh fish would need 200 kg of salt, at a cost of Tsh 32,000. In Mwanza, where it was in extremely short supply, a m 2 of firewood was c Tsh 8000, but if bought in smaller quantities could be Tsh 16 000.] Of course, there was still a market for sangara moshi, although it is difficult to make reliable statements about its size. According to lists of officially recorded transactions at Kirumba, somewhere between 740 tons of sangara moshi [ Equivalent to around 1500 tons of fresh fish assuming a 50 per cent freight loss during processing.] were traded there in 1995. Officially recorded transactions are normally around 50 per cent of trade that actually takes place, [ The system in most markets is that most transactions are recorded at half their actual volume for levy purposes. Half of the remaining unpaid levy is then retained by the trader; the other half is paid to the market official as a bribe.] but even so there was no evidence at all of sangara moshi being bulked at Kirumba during my own fieldwork in the first third of 1996. By this time the trade seemed to have three main directions, two of them originating mainly from landings in Geita district (Mwanza) and the neighbouring Biharamulo district (Kagera region). The first was direct to the mining areas south of the Lake in Geita district and around Kahama (Shinyanga region). The second was direct to Mwanza town rather than via Kirumba. The markets served in Mwanza town were the old central market (Soko la kati) and several new magenge markets around the town. [ Magenges are stall holders in semi - formalised open air markets, usually situated in high density areas.] The latter were also supplied by processors smoking at the larger landings near to Mwanza town, like Igombe. At Soko la kati, where there were about fifteen stalls selling sangara moshi, the main customers were consumers from Mwanza town and wholesalers from the main towns in neighbouring regions, notably Shinyanga and Tabora, with transactions of up to 500 pieces. [ c 0.25 ton of processed fish, assuming an average wet fish weight of 1 kg/piece.] Traders at Soko la kati also reported regular sales to traders from Dar es Salaam, but it is difficult to believe that this trade was particularly significant. A third direction of the sangara moshi trade was from the islands, passing through Kirumba only to pick up transport (and pay levy), direct to down-country destinations. Kirumba market receipt books for the domestic trade were examined for February, June and October 1995. While because of under-recording the figures in them cannot be used to read off total volumes of the trade, they can perhaps be used as a guide to its direction (assuming all transactions have been under recorded in the same way, then the shares of volume recorded by destination would be the sameas those occurring in reality). Dar es Salaam accounted for almost 70 per cent of recorded transactions by weight; no other centre of consumption accounted for more than 7 per cent. The Lakeside sangara moshi processors who were based most of the time at specific landings or permanent camps operated on a variety of scales. The smallest, producing under 200 pieces/month, tended to simply receive visiting traders, almost all of them magenge from the mining areas or (at Igombe) Mwanza town. Traders from areas away from the lake who came to buy fresh fish for smoking tended to operate on a similar scale. Those based at the landings and producing between 200 and 500 pieces/month tended to leave the landings to sell inland themselves, either at local markets like Nkome, or in the mining areas, or at Mwanza Soko la kati. Purchases by traders from the mining areas visiting local markets like Nkome were on the same scale. [ The day I visited Nkome there were 28 sangara moshi transactions, all involving one or two tengas (200 - 400 pieces), i.e. a total of around 4 tons. I was told this was typical for a weekly market.] Those producing over 500 pieces/month were the ones likely to be shipping and selling downcountry, more often than not directly to regular customers at Kariakoo market in Dar es Salaam. Just as at Kirumba, it is probable that receipt record books under-recorded the real volumes passing through Karakoo market by half. In 1994 2252 pakachas of sangara moshi (known in Dar es Salaam as sangara moto) were recorded as having been traded; in 1995 the number was 2050. A Dar es Salaam pakacha is around 500 pieces (0.25 ton). Unlike the case of kayabo/sangara chumvi (see below), the customers in Dar es Salaam for sangara moto were said by Kariakoo traders to be heavily local: the main purchasers were local magenge. To sum up, given the confusing information about the sangara moshi trade at Kirumba, it is hard to estimate its overall size, but in 1995 it was possibly around 400 t/year around the lake hinterland, 1000 tons for the Dar es Salaam market and 100 tons for the rest of the country (equivalent to about 3000 tons wet fish in all). The trade in kayabo/sangara chumvi was longer distance, larger and more centralised. A large proportion passed through Kirumba mwaloni and the largest single market for this was Bukavu in eastern Zaire. The central players were a group of 50-60 active (of about 170 registered) kayabo traders at Kirumba. Many of these typically worked in partnership with a roving lake-based collector and also bought from incoming traders or processor-traders from the islands. The large roving collectors typically bought 5-10,000 pieces/month catch in two collections; incoming traders or processor-traders at Kirumba were generally bringing in single collections of 1000-1500 pieces/month each; there were a lot more of the latter than the former and they probably accounted for the bulk of total kayabo purchasers on the lake. Both would be buying from small and medium-scale processing operations of perhaps 100-600 pieces/month. Thoseprocessing more than 600 pieces/month themselves would almost certainly be taking them direct to Kirumba, or if they were producing more than 2000/month probably direct to Dar es Salaam. Traders from Dar es Salaam and other down-country centres by-passing Kirumba and going direct to the islands would be buying from the same small and medium scale processing operations already described, probably also in total loads of 1000-1500 pieces. There was no local trade in kayabo, which was not announced anywhere around the lake itself. Processing by the kayabo method was introduced to the Mwanza area by visiting Zairean traders in 1989. Earlier efforts by Tanzanians to produce a regionally-marketable salt-dried Nile Perch had failed. From 1990 until 1992 processor-traders based around the lake and traders at Kirumba itself were taking kayabo by road to the borders of Rwanda and Burundi, and sometimes - where they could find 'hosts' (see below) - just beyond it. As both visa and export licensing procedures were relaxed between 1991 and 1993, incoming Zairean traders rapidly took over the main part of the cross-border trade. Over the same period, traders at Kirumba managed to ensure that they achieved a near-monopoly in selling to them. By 1996 there were probably 50 Zairean kayabo traders coming to Kirumba on a regular (on average six times a year basis). Until the Rwanda war of 1994 they were coming by road through Kigali, bringing gold which they bought in Zaire for resale at banks in Mwanza. In 1994-95 the gold price in Tanzania fell and the RPF stopped them crossing Rwanda, claiming that they were supplying the Hutu militias in the camps around Goma. By 1996 they came with hard currency, using the longer route across Lake Tanganyika from Uvira to Kigoma and then by train (changing in Tabora) to Mwanza. Since they were all from the Kivu province in eastern Zaire, these traders could speak Kiswahili and theoretically could therefore conduct business on their own account. In practice, they all used local agents working on commission (known at Kirumba, for reasons which will be explained, as magofi) to do their buying, to deal with Tanzanian officials of various descriptions, and to organise the homeward carriage of their purchases. The peak period of Zairean purchases was in the wet season of January-May, when demand in Tanzania itself tended to be at its lowest. It was during this time that supplies from Lakes Albert and Tanganyika, where their traders normally obtained their fish, tended to fall. Normally, visiting Zairean traders never purchased less than 1000 pieces (roughly equivalent to 1.75 ton). An average in 1996 was probably 1500 - 3000 pieces (c 2.5 - 5 tons) and a maximum 10,000 (c 17.5 ton). Through their agents, the Zaireans would place their orders, then wait around for a week or two until their load was assembled. On payment and completion of formalities consignments were packed (in tengas), labelled individually and a truck was hired jointly with other Zaireans to take them the five kilometres or so across town to Mwanza South rail freightyard. Here the group of traders would jointly hire a 13 or 40 tons freight car to transport the fish to the ferry terminal at Kigoma, where it was taken by a hired transport-boat across to Uvira or Kalumba and then on by road to Bukavu. The consignments were resold in Bukavu in relatively small amounts to traders from within 100 km of the town. According to market receipt books at Kirumba only around 300 tons of kayabo were sold for export at the market in 1995. This figure would appear to have little relationship to reality. In the first place, counterfoils for Owner's Risk Goods Consignment Notes equivalent to almost 1800 tons were issued at Mwanza South rail freight office for kayabo consignments between Mwanza and Kigoma in 1995. In the second place, there was general agreement amongst all those interviewed at Kirumba that the external trade comprised close to half of the total for this commodity, whereas in market records it was under a quarter. As will be described in a moment, there was some kayabo bound for Zaire which the Kirumba traders did not control, but it is unlikely to have exceeded more than 100 tons/year. The Zaire kayabo export trade through Kirumba was therefore at least around 1800 tons/year and possibly over 2000. Most of the domestic trade in kayabo was to the Indian Ocean coast; very little was consumed in the interior. As with the sangara moshi trade, Kirumba market receipt books for the domestic trade were examined for three months in 1995. These showed 56 per cent of the domestic trade by volume being bound for Mtwara and Lindi, 21 per cent for Tanga and Muheza and 13.7 per cent for Dar es Salaam. Mtwara and Lindi, nearly 2000 km from Mwanza in the south east of Tanzania, are areas with virtually no livestock and almost certainly have the highest per capita fish consumption in the country. Certain Mtwara-based traders interviewed at Kirumba personally owned chains of 'fish shops' in each of the main towns of the area (Mtwara and Lindi themselves, Nachingwea, Newala, Masasi and Tunduru) and were shipping very large cargoes for distribution between them. One trader I interviewed had flown to Mwanza via Dar es Salaam and was supervising the assembly of a 20 tons load. On completion, he was planning to organise another 30 tons load before returning home. According to him there were probably 30 others who, like him, came to Mwanza perhaps six times a year and bought perhaps 10 tons kayabo/a visit, (equivalent to 1800 tons/year). This trade was all by road. Unlike the Zairean traders, those from Mtwara , Tanga, Dar es Salaam and other domestic centres did not purchase from Kirumba alone, or even mainly from Kirumba. They were at liberty to buy direct from processors on the islands and had strong incentives to do so if they discovered, via local intelligence, that fish could be obtained significantly cheaper on the lake itself. However, they were generally still obliged to physically pass through Kirumba and pay market levy there as a rule, for there was no alternative landing - at least in Mwanza region - where a 10-20 tonstruck for transport down-country could easily be hired without costly and complicated pre-arrangement. The largest incoming traders seemed to mainly buy from Kirumba itself, while the smaller ones tended to buy directly on the islands. This decision was clearly related to the respective costs of assembling large and small loads from the islands themselves. Unlike some larger processor-traders based on the islands themselves, the Kirumba traders were only extremely rarely taking consignments to Dar es Salaam in 1996, although many observed that this had been common a few years earlier, and in 1990 had been the main form of the trade. The reasons for this change will be examined in the next section. I never discovered much about the kayabo trade between Mwanza on the one hand and Tanga and Muheza on the other, except that according to rumour, informal export on to Mombasa was an important component of it. The trade with Dar es Salaam was also predominantly an intermediary one, for kayabo/sangara chumvi is not consumed widely in the capital. Here the onward trade was principally once more to Mtwara and Lindi. Whereas in the sangara mushi trade, traders from the islands were mainly shipping themselves to Dar es Salaam, in the kayabo trade, the Dar es Salaam trade was dominated by traders coming from Kariakoo and elsewhere to the Lake. These traders came to Mwanza individually, obtained market intelligence concerning where on the lake kayabo was currently most plentiful and cheap, proceeded to purchase it and then returned to Kirumba to arrange transport. Consignments were smaller than those going direct to Mtwara and Lindi, implying that transport had to be shared either with other Dar es Salaam traders and/or with those shipping to other important destinations near the coast like Morogoro and Chalinze. Those with very small consignments (one or two pakacha), who were generally based not in Kariakoo but in other markets in Dar es Salaam were probably mainly using the passenger train luggage service - as were smaller traders from centres along the Central Line generally. The Mtwara and Lindi buyers who came to Kariakoo to buy kayabo/sangara chumvi were clearly in the second rank of these regions' traders, although this had not always been the case. So long as most of the trade out of Mwanza was accounted for by outward journeys by Kirumba traders, Dar es Salaam traders remained in Kariakoo and 'received' traders of all sizes from Mtwara and Lindi. When Kirumba traders ceased making outward journeys, larger traders from Mtwara and Lindi started travelling direct to Kirumba itself while smaller ones remained 'faithful' to Kariakoo. This corresponded to a sharp decline in the kayabo/sangara chumvi trade through Kariakoo itself. Receipt books at Kariakoo market office record only 910 pakacha of kayabo passing through the market in 1994 and 825 in 1995 (i.e., c. 228 and 206 tons respectively). My own estimate is that the internal (including Tanga/Muheza) trade in kayabo was somewhere around 3000 tons/year, and that the total internal and external kayabo trade may have been 5000tons/year, equivalent to around 10,000 tons of wet fish; the total trade in kayabo and sangara moshi was perhaps 6500 tons/year or equivalent to around 13,000 tons/year of wet fish. It is important to underline that while almost all of the larger of the two trades under consider ation (kayabo/sangara chumvi) passed through Kirumba, part of the Kirumba trade emanated not direct from artisanal processors on the islands, but from processors at Kirumba itself. On the shore at the end of the market furthest away from Mwanza town there was a regular procession of 3 ton trucks from the factories unloading reject fish weeded out in the factory's own selection process. It is unclear what proportion of the factory intake was rejected in this way; given the stringency of selection on the lake itself it was probably not more than 10 per cent. These fish were bought on the spot for Tsh 400-700/piece depending on size, either by Kirumba or outside traders, and immediately prepared for drying by teams of highly organised casual labourers who assembled as if by magic the moment a truck pulled in. Kayabo processed in this way was dried on racks at a site a couple of hundred metres further down the road, alongside other Nile Perch products which will be described in a moment. Kayabo made from fish dumped at Kirumba by the factories made up part of the poorer grade II quality - most of which was said to be destined for Zaire. The remainder of grade II quality kayabo was sold by the factories themselves. The most substantial sales were by the factory with the largest in-house artisanal fishing fleet (133 fishing vessels registered at Mwanza RFO). Although Zairean traders were apparently the sole exporters of this second category of fish, a group of local magofi had managed to insert themselves as commission agents and took care of all direct dealings with the factory (the ex-factory price was similar to that at the dumping area in Kirumba itself). The final element of the trade in sangara consisted in the sale or disposal, further processing and subsequent resale of waste fish products from the factories. Waste products were of two basic kinds: waste generated in the production of fillet and waste generated in the finishing of fillet. The Tanzanian trade in waste generated in the production of fillet dates from 1993, according to legend, when local fish friers (and smokers) found that they could no longer obtain fresh fish at the landings around Mwanza town. The story passed around that the factories at Mwanza South were giving away the remains for free and a number of friers and smokers set up a processing camp about 300 m south of the gate of the largest factory, at a place called Mwalo ya Pamba (Cotton Landing). This mwalo, just below the old railway engine shed which gave the district its name (Sheddi) had for some time been used as a dumping ground for cotton husks, which were used by the friers and smokers as free fuel. Of the fish remains which the factories discarded in this way, the only part certain to contain meat was the head. This was therefore sliced from the rest of the frame, opened up or split, and then usually smoked over an open fire on a wire meshsupported by stones. Around the same time the square shaved ('flat-top') hairstyle popularised in the US by Carl Lewis and in England by Ellery Hanley took off amongst Mwanza's male youth. As with many other imported items, cultural or material, it took on a Kiswahili version of the northern hemisphere name for a similar object rather than the actual one: 'punk'. Since the heads of the sangara were also 'shaved', they became christened 'punki'. From the onset, the processor-traders at Sheddi began getting complaints about the smoke and the smell from people living around the site. After a series of manoeuvres they eventually shifted to the site at Nyegezi in April 1994. By this time the factories had started to sell the waste rather than giving it away free, and a market had begun to be established for the product, mainly amongst previous consumers of fried and smoked sangara proper. At Nyegezi in early 1996 processing had become highly organised. Traders went to the gates of the factories at fixed times when frames were being sold and bought job lots. They then organised their transport by trolli up to Nyegezi where around 200 casual workers, themselves organised into different specialisms and renumerated at different rates, prepared punki for smoking or frying. The principal division of labour was between splitters and washers/driers, each hired by the piece. Subsequently most of punki was smoked in the same way as normal sangara on one of the site's 38 kilns, by the trader and/or a minder. Cotton husks were still used as fuel. Most of the kilns belonged to the bigger traders using the site and those not owning them had to pay a fee. Pieces in poorer condition and deemed to be unsuitable for smoking were fried in pans over open fires, using oil scraped from the fish intestines with a knife or fingernails into a basin and subsequently 'refined' by boiling and then left to cool and finally bottled. Intestines were bought or begged for this purpose by a distinct group of 'scrapers' also based at the site. A similar group could be observed at Kirumba (the factories themselves also refined and sold fish oil). Finally the smoked or fried fish were packed into tenga, a vehicle was hired by a group of traders and the punki was taken to one of a number of weekly markets in the Mwanza and Shinyanga region hinterlands (Madaha, Malanpaka, Hulyumalwa, Salawe, Nyasamba and Isaka). Most of the traders using Nyegezi site were local, but traders from other areas could and did use it. I encountered punki traders as far away as Tabora and Singida town markets who came to Mwanza, bought frames at the factory gates and then organised the processing of punki for wholesaling to other traders in their 'home' markets. In the case of all the local markets mentioned the main customers were described as 'local traders from the villages'. A second, much smaller and relatively unorganised punki producing site was located on the airport side of town at Ilemera (Mwanza North) where there were about 10 processors working in April 1996. There was also a considerable home-based smoking and frying trade in several parts of Mwanza town, especially Pasiansi. The punki trade was concentrated in the dry season, when alternative foodstuffs were in short supply around the lake. Most of the approximately 50-70 traders involved nevertheless brought rice, maize or cassava back from the hinterland. There was considerable differentiation between the traders, but more in terms of their ownership of capital equipment (i.e. kilns) at the site than in levels of trading activity (a trolli load of 1000 pieces appeared to be a standard consignment size for both Mwanza and externally-based traders, costing either Tsh 15-30 or Tsh 70-90/piece at the factory gate, depending on size). At the same time a 'good size' processed piece cost around Tsh 130 on site at Nyegezi. Altogether the total punki trade was approaching 500 ton/year. The term 'punki' may have been a Tanzanian invention, but a trade in waste fish products from the Kenyan factories was already noted by Reynolds and Greboval in 1987 (Reynolds and Greboval, 1989). The Kenyan factory owners in Mwanza said that the equivalent trade in Kenya was more extensive and expressed surprise and regret that the Tanzanian one was not larger. One told me that his Kenyan plant could pay all its local overheads out of its income from selling frames: in Tanzania he was having to pay for most of his factory's waste to be disposed out of town. The main waste generated in the finishing of factory-produced fillet was trimmings from cut fish steaks made by filleters to leave fillets of standardised sizes for packing. These trimmings, ranging in size from 10-200 gm, were known locally as chips. A trade in them started only in first half of 1995. Zairean traders were the main players, but again they acted only through intermediaries. The trade in this case was under the control not of the Kirumba traders but mainly of small groups of 'master processors' who were operating at several sites around town. One group had reopened the old punki site at Sheddi and built about 20 drying tables there under the watchful eye of a large colony of storks. The Zairean traders would approach the processors and pay them an all-in fee, either wholly or 50 per cent in advance. The factories charged Tsh 100-150/kg for the trimmings; the traders were giving the processors money to buy the chips and paying them an equivalent of about Tsh 250/kg for purchasing, transport, salting (as per kayabo), drying and packing into gunny sacks. The drying process took around three days, so the processors could dry 0.5 - 1 ton/week each. There was a shifting pool of casual labour, of whom only the salters were paid. Chips processors also shared the much larger drying site close to Kirumba used for ex-factory kayabo processing. The Zairean chips traders mostly also traded kayabo. The same magofi organising their relations with Tanzanian officialdom and their onward transport covered both kayabo and chips. At Sheddi, some of the Zaireans were buying enough to keep all five processors occupied (i.e. up to 2 tons). According to counterfoils of Owner's Risk Goods Consignment Notes held at MwanzaSouth rail freight office, 370 tons of chips were shipped in freight wagons from Mwanza to Kigoma between April 1995 when the trade started, and the end of the year. Chips was a popular commodity amongst traders. It was easy to handle and instantly 'took off' with customers in Zaire, not least because of its availability in smaller and more divisible units than kayabo. According to traders, it also cheapened the cost of giving bribes at Zairean army roadblocks: 'they always ask us for 100 of whatever we're carrying; 100 pieces of chips costs us a lot less to give away than 100 pieces of kayabo'. The sangara chain had experienced several transformations up to the time of my fieldwork. A few months later it was to experience another profound one when fighting which had been simmering in eastern Zaire for some time exploded into all-out war. Bukavu was seized by opponents of the Zairean government, with the probable support of the Rwandan army, and hundreds of thousands of Rwandan Hutus who had been in UNHCR camps around the town either dispersed toward Kisingani or were obliged to return to Rwanda. How much of the processed sangara exported to Zaire was destined for the camps is unclear, but even if none had been, the trade would almost certainly have still been severely disrupted. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
