"Women Are Our Coffee":
Historical Factors and Current Variables
in Smallholder Coffee Production
in Papua New Guinea

Holly Wardlow
Emory University

Coffee is second only to crude oil as the most important internationally traded commodity in monetary value (Araim 1991, Marshall 1983, Stolcke 1988, Stewart 1992). In Papua New Guinea, coffee is the leading agricultural export, and, in fact, is the second earner of foreign exchange after copper. While coffee production in Papua New Guinea began on expatriate plantations and was intended to supply the Australian market, Highland villages soon began growing coffee in their own gardens, and indigenous smallholder production rapidly overtook plantation production despite government policies, both before and after independence, designed to encourage plantation style production. 70% of the Papua New Guinea's coffee is now produced by Highlands smallholders in plots averaging 400 trees (Stewart 1992), and Grossman asserts that the Highlands commercial economy is "precariously dependent upon a single export cash crop, coffee, which can account for up to 90% of rural commodity income" (1984:28). Smallholders' decisions to plant coffee have transformed the Highlands economy and ecology, and continue to influence gender roles and elations, land tenure customs, and traditional exchange rituals. Coffee production has had such a pronounced effect on social relations that in one area of the Highlands men and women agree that "women are our coffee." An examination of the factors contributing to, and some consequences of, smallholder production of coffee is therefore in order. This paper will draw on a agricultural decision making model to direct attention to some of the most important historical factors and current determinants affecting smallholder coffee production. I will demonstrate that the history and structure of the international market have forced Papua New Guinea smallholders into one of the most labor intensive methods of producing coffee. Papua New Guinea's government policy, lack of infrastructure, and mountainous topography have constrained producers' options and compounded the labor necessary for success. Since the social organization of labor in Papua New Guinea makes women the primary agricultural producers, access to female labor is the limiting factor in smallholder coffee production. Smallholders themselves often cite low prices on the international market and lack of access to coffee buyers as barriers to success in coffee production; however, these two factors are more likely to lessen the likelihood of harvesting coffee than to deter planting. A Brief History Coffee production grew up around Goroka, the capital of Eastern Highlands Province. Goroka originated as a lone airstrip built during World War II to deter a threatened Japanese advance. After the war, the Australian Administration extended its control into the Highlands, and several patrol officers and gold prospectors settled around Goroka with the intention of exploiting the agricultural potential of Goroka's montane valleys. With no road to the coast, they required an agricultural commodity with a high value per weight ratio to make airfreight economically feasible. Coffee was ideal. During the late 1940's and early 1950's, local clans were eager to sell some of their land to the prospective plantation owners in order to gain immediate cash, possible future wage labor, and information about cash-cropping. However, by 1957 there was criticism from both Highlanders and Australians about excessive alienation of land, and a ban on further land sales was instituted (Finney 1987). The Papua New Guinea economy was meant to compliment Australia's and in order to meet the Australian demand for coffee, agricultural extension services were begun in 1953 in the Goroka area to teach Highlanders proper methods of coffee growing. These services were initially aimed at group leaders who could marshall land and labor resources for commodity production. However, by the early 1960's individual smallholder coffee production had overtaken both indigenous and expatriate plantation production, and was expanding so fast that extension services were halted. In 1975, harvests in Brazil, the world's biggest producer of coffee, were destroyed by frost, and prices for coffee on the international market skyrocketed. 1976 and 1977 marked the period when New Guinea smallholders first experienced profits from coffee on a grand scale (see Diagram 1: Stewart 1992). In 1992 coffee prices on the international market were the lowest they have been since the early 1970's, and during the past five years its production has been an ongoing disappointment to smallholders. The International Coffee Agreement, which serves to stabilize prices, fell apart in 1988 and a new agreement has not yet been reached (Araim 1991). It is difficult to predict how Highlanders will respond to the current poor market since the history of coffee production in New Guinea is relatively brief: while Brazil, for example, had dominated the world market since 1840, indigenous production of coffee in New Guinea only began in the 1950's. Anthropology and Coffee Production in the New Guinea Highlands Anthropological interest in the New Guinea Highlands also began in the 1950's, but very little mention is made of coffee production, even in ethnographic work from the 1960's when 60% of all coffee acreage was under indigenous control, and when the output of indigenous growers increased by 28% per year (Good 1986). By 1970 there were more than 45,000 indigenous coffee growers in Eastern Highlands Province alone (Stewart 1992), but it seems that many anthropologists have been reluctant to perceive Highlanders as anything other than horticulturalists engaged in labyrinthine and ostentatious ceremonial exchange and/or chronic warfare (but see Finney 1968). Those few anthropologists who have paid attention to coffee production conducted their field work in 1976 and 1977 during the great coffee boom (Grossman 1984, Sexton 1986); their data is representative of a rather unique period in PNG's history of coffee production. Unfortunately, then, there is little ethnographic information concerning the explosive spread of smallholder coffee production in the 1960's and 1970's. Precise data concerning the social organization of labor or proportion of land allocated to coffee is not available. Nor do we know how smallholders decide to allocate their resources and manage potentially competing desires - whether to produce coffee, raise pigs, acquire wives, look for work, send children to school, or demand remittances from urban dwelling kin. It is clear, however, that coffee production has been quite important to Highlanders themselves, and continues to have a great impact on their lives, perhaps particularly now when coffee prices are low and recourse to other sources of cash are, for many, nonexistent. The Agricultural Decision Making Model The allocation of household resources to alternative ends is felt in many arenas of individuals' lives, and decisions to alter allocation strategies can result in overarching patterns of change for a region or nation. Papua New Guineans' decisions to produce coffee have influenced land tenure customs, gender roles and relations, traditional exchange systems, and the Highlands economy as a whole. Understanding how producers perceive their situations, how they want to respond to them, and the constraints on their decisions can help us understand how micro-level strategies for household success emerge as macro-level patterns of change. Unfortunately, understanding decision making processes is far from simple. As Bartlett points out, "In any attempt to say what causes a certain land use pattern, the only accurate answer is everything" (1978:1). However, as she also points out, there are some factors which consistently play a part in the decision making process, namely, household resources and needs as they are shaped and constrained by the natural and social environments in which they are embedded. How all these factors are evaluated by an individual producer in a particular cultural context may also play a key role in decision making. Historically there have been two basic approaches to understanding decision making. The statistical approach examines people's behavior, and, using a path analysis, determines which variables influence individuals to make decisions one way or another (Chibnik 1980, Bartlett 1980b). Cognitive approaches attempt through in-depth interviews to create models outlining precisely how individuals arrive at their decisions (Gladwin 1980). However, there often seems to be a lacuna between what people say and what they do: while statistical models may be predictive, they do not always represent factors which individual decision-makers themselves have articulated, and while cognitive models reflect individuals' perceptions and evaluations of what they think is important, they are sometimes not predictive of actual behavior (Bartlett 1980b). Producers may be quite able to articulate the factors they deem most prohibitive or encouraging; however, it may be those factors that they take for granted and do not articulate that are most important in their decisions. Bartlett has arrived at a model which accounts for many of the factors influencing and constraining producers' behavior. The structure of this paper will follow that laid out for decision making in Diagram 2 (Bartlett 1980b). I will discuss the human and ecological environments in which Papua New Guinea coffee production is carried out, and then I will examine household level factors in terms of land, labor, and capital. Social organization, population density, degree of infrastructure, and history of contact with the Australian administration vary from place to place within the Highlands, and these factors have had an important impact on smallholder coffee production, making generalizations difficult. However, since the whole of the Highlands commercial economy has been built primarily on coffee production, and since developments in one region have influenced other regions, I will generalize as much as possible and use specific comparisons where useful. I would only add to the decision making model in two ways. First, in the case of commodities produced specifically for export, the role of the international market must be highlighted. As Roseberry urges, We need to concentrate on relationships that transcend spatial boundaries, that take the apparently external and make it internal to our model of a social structure. 'Anthropologists don't study villages', wrote Clifford Geertz (1973:22), 'they study in villages.'...We need to be creative in our conceptualization and study of relationships, institutions and networks that are apparently foreign to the community (1989). While the links between macro-level international forces and micro-level decisions may be difficult to tease out, in the end the international arena for any export commodity affects the small-holders producing it. Second, as Stewart points out, "in order to explain social and political relations in an industry, it is necessary to know how a commodity is produced" (1992:36). I begin my discussion, then, with an examination of coffee production in general, and then look at the international market in which it is embedded. Coffee and its Production Coffee is indigenous to Ethiopia, Uganda and Zaire, and Africans were the first to use the stimulatory effects of coffee by chewing the cherries and eating the ground beans mixed with animal fat (de Graff 1986). Coffee's consumption as a beverage spread from the Middle East to Europe where it immediately became fashionable. As early as the 17th century Indonesia was producing coffee for the Netherlands to support the growing number of coffee houses in Europe. A perhaps apocryphal story is that the Dutch made a gift of coffee to Louis XIV who was so impressed that he ordered in his will that it be spread throughout his tropical empire (de Graff 1986). From French Guyana, Haiti and Martinique, then, coffee spread to Brazil and the Central American nations that eventually became the primary producers for the world. Although there are a number of different coffee species, the two most important types in terms of commodity production are arabicas and robustas. Arabica requires temperatures of 18 - 25 C, and the high altitude areas of countries located near the equator seem to be most suited to its growth. Being evergreen, it requires constant water except for 2 - 3 month dry season to slow growth and initiate flower buds. In Papua New Guinea most of the Highlands area is suitable for its production. Arabicas begin producing small cherries 3 or 4 years after planting, and during the 4 months of the year that cherries ripen, each tree needs to be harvested continuously or the fruit falls to the ground and rots. Robustas grow at lower altitudes, bear more and bigger cherries than arabicas, are more resistant to disease, have relatively little pulp that needs to be disposed of in processing, and do not drop ripe cherries, making them easier to pick and giving the producer some flexibility as to when and how often he will harvest. From the producer's perspective, robustas are preferable to arabicas. However, the robusta beans contain more acid and caffeine, and their stronger taste seems to make them less desirable on the world market. Both kinds of tree produce their best yields between 5 and 15 years after planting and stop producing well after 20 years, but can be fruitful as long as 30 if they are pruned, weeded, fertilized, and mulched. Arabicas account for 76% of the world market and robustas 23% (de Graff 1986). Papua New Guinea primarily grows arabicas, although some robusta is grown outside of the Highlands in warmer areas closer to the coast. The international coffee market deals with three coffee types: robustas, mild or "washed" arabicas, and Brazilian or "unwashed" arabicas. Arabicas are divided into washed or unwashed types depending on how the cherry skin and pulp are removed from the beans. In Brazil, arabica cherries are directly spread out in the sun to dry for 3 or more weeks. The skin and the pulp shrivel up, leaving the beans ready for hulling. Other arabica producing countries, including PNG, use the wet or washed method. This method requires that coffee cherries be pulped within 24 hours of being picked. Cherries can be pulped with the fingers, or in hand turned or diesel pulpers. The viscous layer remaining on the beans after pulping disintegrates during 3 or 4 days of fermentation in wooden tubs, and the beans are then washed in fresh water. Finally the beans are spread out to dry for a week or more leaving the beans enclosed in their sturdy hulls and surrounded by a silvery parchment. Regardless of the method used for initial processing, the hulls must be cracked. The beans are then dried once again, and only at this point quality as "green beans" which can be traded on the world market. Hulling and final drying are almost always carried out in the producing country in factories with oven driers. Blending, roasting, decaffination, and conversion to instant coffee generally take place within the importing nation. The washed method is quite labor intensive, but produces better tasting coffee. Brazilian unwashed beans are of much lower quality than washed, and are sold at lower prices. They are valued highly in the United States, by far the largest coffee importer, where lower quality but cheaper coffee is desired by the public. The unwashed Brazilian beans, with their neutral flavor, are used as a base and blended with other higher quality beans to produce the standard American diner cup of coffee. The International Market Coffee is primarily an export commodity, so to understand smallholder production, one must understand the international market in which it is embedded. Papua New Guinean smallholders are particularly influenced by the international market. Since Brazil has long dominated the market for unwashed arabicas, Papua New Guinea, like other smaller producing countries, must invest in the more labor intensive washed method of processing. And, unlike Brazil, for example, where domestic consumption accounts for almost 50% of its total coffee production, Papua New Guinea has not developed processing/roasting facilities nor cultivated a domestic market, and thus consumes only 1% of the coffee it produces and exports 99% (see Table 1: de Graaf 1986). Moreover, Papua New Guinea derives 15 - 25% of its export earnings from coffee (see Table 2: Stewart 1992). Since coffee is completely an export commodity in Papua New Guinea, the structure and state of the international market not only directly affect prices for the small-holder, but also shape government policies which can encourage or deter small-holder production. Coffee's unique characteristics as a commodity have made it the object of the longest lasting international commodity agreement (Araim 1991). Like oil, bauxite and copper - other valuable commodities governed by international agreement - vast amounts of money change hands in coffee's exchange; in 1985 there was a turnover of nearly 4 million tons of coffee and US$10.5 billion dollars. But unlike these mineral commodities, coffee is vulnerable to natural phenomena such a frost or disease; one bad year of frost can drastically alter the tenor of the market for the next 2 or 3 years. And, unlike minerals, coffee must be harvested and processed in a timely manner. The harvesting of coffee is necessarily labor intensive: human labor cannot be replaced by capital intensive measures since no one has yet invented a machine which can distinguish ripe from non-ripe cherries and harvest a tree without doing extensive and lasting damage. Many of the producing countries rely on coffee for most of their export derived income; for example, in 1980, 98% of Uganda's export earnings came from the sale of coffee (Marshall 1983, Stewart 1992; see Table 2). Producers, then, have a large stake in maintaining good prices for coffee. Demand, however, is subject to consumers' changing tastes. Thus, robusta, for example, is currently becoming more desirable since it can be used more efficiently than arabicas to make the now fashionable instant and decaffeinated coffees. Since producing countries tend to specialize in one kind of coffee - e.g. Indonesia and the African nations produce robusta coffee, while the Latin American countries produce arabicas - such shifts in taste cause shifts in the market. And, unlike other commodities, coffee consumption does not usually go up when prices go down - world demand tends to remain fairly constant unless new markets are cultivated in countries which do not traditionally drink coffee, as has occurred in Japan over the last 20 years. Because the coffee market is vulnerable to all these contingencies, coffee prices have tended to fluctuate dramatically. When Brazil experiences frosts, for example, the world supply of coffee drops precipitously, prices soar, and smaller coffee-producing countries experience a boom for a few years. This situation motivates coffee producers to increase plantings, and motivates the average American family (the world's leading consumers) to indignantly reduce a consumption and demand the cheaper coffee to which they had become accustomed. Predictably there is a glut on a dampened market a few years later when those newly planted trees began to produce cherries and Brazil's production has recovered to normal levels. The glut is, of course, accompanied by low prices distressing to producing countries. It is because of this "tendency toward persistent disequilibrium between production and consumption...and pronounced fluctuations in prices which can be harmful both to producers and consumers" (International Coffee Agreement 1962, in Araim 1991) that in 1962 the International Coffee Organization (ICO) was created, and the first International Coffee Agreement (ICA) was reached to regulate the exchange of coffee. Since direct regulation of prices was deemed undesirable, the organization instead estimated the tonnage of coffee needed in the world and allocated quotas to each coffee producing country, limiting the amount they could export each year. Quotas are allocated in "bags" of 60kg. each, or in tonnes of 1000 kg. each. Presumably the idea is that exporting governments will somehow regulate their smallholders or plantations so that they do not produce more coffee than can be exported or stocked. This does not always happen and many countries end up storing vast amounts of coffee in government subsidized warehouses hoping that Brazil will once again experience bad weather so that quotas will be lifted and they can release their stock onto the market. The establishment of the ICO has not been completely beneficial to Papua New Guinea. To begin with, PNG joined the ICO as a colony of Australia, and was allowed to export only as much as Australia could consume. By the mid 1960's, Papua New Guinea was already producing more than this allotment. Secondly, the 1962 ICA quotas were based on the previous few years' production instead of on longer historical trends or projected capabilities. Brazil was therefore granted 39% of the total shares even though historically speaking it had been producing a smaller and smaller percent each decade: before World War II, one state in Brazil, Sao Paulo, produced 90% of world coffee needs, but this proportion declined rapidly so that in 1960 Brazil produced only 30% (Singh 1977). The International Coffee Agreement, then, while stabilizing prices, essentially served to freeze the market as it was before 1960, a situation particularly beneficial to the U.S. and to Brazil. The agreement also served to preserve bilateral trade agreements - such as those between the United States and Brazil, or France and its former colonies - which provided importing countries with discounted coffee and exporting countries with assured markets (Stewart 1992). The ICO was structured so that nations were allotted votes based on their shares in the market. The biggest coffee producers and consumers were granted the most votes, and thus Brazil and the USA together had and still have a strong influence over allocating new quotas and changing the terms of the agreement. Papua New Guinea has experienced difficulties in getting its quota increased and, in fact, in 1982 had to take a cut in its quota from 700,000 bags or 42,000 tonnes, to 610,000 bags or 36,000 tonnes (Good 1986). Good found that in the early 1980's when Brazil experienced a frost and export quotas were lifted, PNG exported up to 50,000 tonnes (see Table 3: de Graaf 1986), and Stewart estimates that at that time PNG was capable of exporting 58,000 tonnes (1992:227). Obviously, PNG's production capacity far outreaches the export quota it has been assigned. Like many countries whose coffee production was relatively small when the ICO was formed but thereafter expanded rapidly, PNG has suffered under a quota that does not correspond to its production capacities, and being a weak member of the ICO, has been unable to avoid these constraints. A small quota is felt directly by small-holders: when export companies cannot export or stock any more coffee, they stop buying, which means that coffee buyers no longer come to villages to buy smallholder parchment. Likewise low prices on the international market are absorbed primarily by the smallholder. However, it is important to note that while these low prices make coffee a variable and unreliable source of income, they probably do not deter smallholders from investing in coffee. Grossman found that when prices were low, the Kapanara people simply neglected to harvest their coffee; they did not decide to dig up their coffee and plant something else (1984). The Natural Environment Having examined coffee as an agricultural product, and situated it in its international content as a commodity, we can now turn to those features of the Papua New Guinea natural and human environment that influence smallholder production. The Papua New Guinea Highlands region, with its tropical climate tempered by high altitude, heavy rains, and brief but reliable dry season is ideal for growing arabica coffee. Soil quality and weather, then, are not limiting factors for coffee production. But the topography of the land, in combination with the international market, has almost certainly influenced production. Since the United States is the biggest importer of coffee, and obtains cheap unwashed beans from Brazil, PNG can only compete by producing the high quality and labor intensive mild or washed arabicas. The washed method requires initial processing of cherries within 24 hours of harvesting. Many Latin American and African nations have established community or government run processing plants in central locations so that smallholders can sell their cherries directly and not be concerned with processing (de Graaf 1986). In the PNG Highlands, with its rugged mountains which make the construction and maintenance of roads quite difficult, transportation is irregular. Smallholders have therefore resorted to another strategy not widely practiced in producing countries, which is to do the initial processing of coffee themselves. Presumably, smallholder processing of coffee to the parchment stage was government policy since it was initially extension officers from the Department of Agriculture, Stock and Fisheries who trained people in this technique. Only those producers living near provincial capitals on the Highlands Highway, like Goroka, Mount Hagen, or Kainantu, can take their cherries directly to processing factories. Anyone further than a couple of hours from these centers must pulp, ferment, wash and dry their cherries. Thus, the natural environment is a constraining factor in that it forces smallholders to process coffee, a time consuming activity. Processing beans to the parchment stage entails more labor than would be necessary otherwise, but it also means that producers can store their beans and thus have some choice as to when and to whom they will sell. The Human Environment The environments which constrain and influence producers' actions are not shaped only by ecology. Humans create the environment around them, enabling some activities while deterring others. It is therefore important to identify those factors dependent on human action which shape individuals' choices. In the case of smallholder coffee production these are Papua New Guinea's government policies, marketing mechanisms, and roads. Government Policy In Papua New Guinea, the direct consequences of the establishment of the ICO in 1962 was that the administration, concerned about over-expansion and the minimal export quota allotted to PNG, actively discouraged smallholder production of coffee by withdrawing all agricultural extension services, closing the coffee nurseries that supplied smallholders with seedlings, and promoting village production of alternative commodities such as cattle. These actions had little impact on smallholders. Maclean reports that the Kwima Maring traded bird plumes for coffee seedlings from their neighbors in the Jimi Valley who had, in turn, obtained them from the Simbu (Maclean 1981). The Kwima were thus able to establish coffee gardens in 1964 with no initial cash investment and no government assistance. In this manner, smallholder production throughout the Highlands increased at the fastest rate ever, competing with plantation produced coffee for PNG's small export quota. Between 1958 and 1969, PNG's production increased from 1,000 to 20,000 tonnes, most of which was due to expansion of the smallholder sector (see Table 4: Stewart 1992). The administration's actions seemed to deter coffee production only in parts of the Southern Highlands, the Highlands province last "contacted" and least developed. The Highlands Highway was not extended to the Southern Highlands until the 1980's, and the people there did not experience the explosive development of expatriate plantations all around them as Eastern and Western Highlanders did. In Koroba, coffee plantings were begun in 1965, but once assistance was curtailed in 1967, production did not expand (Harris 1972). Instead, in 1969 cattle projects were actively promoted with government subsidized fencing and training through the Dept. of Primary Industry (DPI - formerly the Dept. of Agriculture, Stock and Fisheries). It would seem, then, that government policy had little impact on smallholders' actions, particularly for those who had witnessed the potential benefits to be had from coffee production, and therefore may not be an important factor in constraining decisions. Certainly, unlike many African nations whose government's play a large role in regulating coffee production by setting prices, running processing factories, and controlling export facilities (de Graaf 1986, Ruthenberg 1968), PNG's government has maintained a hands-off policy, intervening only to divvy up the national export quota among the various export companies (Stewart 1992). However, policies enacted in the years following independence in 1975 had a great impact on the distribution of land and access to credit, and thus indirectly on small holders. At independence most of the expatriate coffee planters left Papua New Guinea, and it was the new government's desire to maintain the productivity of the plantations they left behind. Therefore land was not redistributed amongst the clans who initially owned it, but was bought up by the government and sold to individual Goroka business leaders or groups through the Plantation Redistribution Act. In 1978 a "consolidation" program was initiated so that individual businessmen could obtain a loan through the national Development Bank to buy plantation size tracts of land. All that was required was a guarantee of a labor force, usually clan members, and the employment of a plantation manager, usually an expatriate. These initial policies were specifically designed to promote large scale plantations over smallholder development, and utilized the existing social organization of Big Men and clan solidarity to do so. The government specifically hoped to create a class of indigenous "Big Men" plantation owners who could provide wage labor for non-owners. In 1962, Palas Wingti, now Prime Minister, stated: It might not be wise to artificially redistribute wealth when our own communal system is already based on the principle of sharing. It would be wiser to aim at building the wealth in certain areas, secure in the knowledge that it will be shared out...by the "Big Men" into whose hands the money first accumulates (in Good 1986:74). This approach to rural development has led to the emergence of a type of business organization unique to PNG, the "development corporation". Development Corporations are huge business groups usually started by one entrepreneurial clan member and based on clan affiliations. Encouraged by government policy and the concommittant coffee boom, many of these groups, such as the Gouna Development Co., began by buying up coffee plantations (see Table 5), and have grown to encompass trucking companies, shares in export firms, coffee processing factories, rental properties, and trade stores. The Gouna Development Corporation is now worth about $10,000,000 (Finney 1987). Government policy and the resulting rise in development corporations has had a great impact on smallholder choices and therefore on their decisions. First, government emphasis on large plantations had meant a deemphasis on developing rural infrastructure, such as roads or centralized marketing structures for smallholders. Secondly, the existence of these plantations and businesses has created the need for seasonal wage labor. While development corporation clans supply some of the necessary labor, much of it is supplied from less developed provinces. Good asserts that this was the intention of the new government; Simbu Province, for example, was meant to remain underdeveloped and provide wage labor for the expanding number of plantations in the Eastern Highlands. Strathern notes that the Southern Highlands and part of Enga province have become a labor pool for Western Highlands plantations (1982a). Out-migration in the Southern Highlands is particularly high, and the development of income generating projects - stores, coffee, or cattle - has been slow (Harris 1972, Strathern 1982a), reflecting decisions of some people not to invest in coffee but rather to seek wage labor. Marketing Mechanisms and Roads The marketing structures for coffee in Papua New Guinea are quite decentralized. Smallholders can take their parchment coffee to town and sell it right at the factory door, they can transport their coffee only as far as the Highlands Highway and see it to roadside buyers, they can sell it to plantations which sometimes need additional beans to fulfill their obligations to importing firms, or they can wait in their village for buyers to come to them. Brookfield notes in trying to assess the amount of production in Simbu that "the multiplicity of marketing outlets would make virtually impossible to obtain adequate data even under continuous field observation" (1973:154). While some roadside and itinerant buyers are commissioned by factories and can only offer a certain price, many are independent. During coffee flushes, producers can take advantage of competition between itinerant buyers and decide to whom they will sell for how much (Stewart 1992, Strathern 1982b, Grossman (1984). Grossman notes that while producers can usually obtain higher prices at the factory door, they prefer to sell to itinerant buyers. Often smallholders do not have access to transport, or must pay transport owners a higher fee in order to get their coffee in town (personal communications from many Highlanders in 1988 and 1992). Moreover, there are often holdups on the roads, and smallholders prefer that coffee buyers, who have to carry large sums of money, take this chance rather than risking themselves the loss of a season's worth of labor (Grossman 1984, Strathern 1982b). However, the decision to sell to itinerant buyers means that when national coffee production is high and prices are low, producers may have to wait weeks for a buyer to come to the village and then sell at whatever price is offered. And, in fact, buyers may not come at all. In Dundi, Gende smallholders stated that access to buyers was a major problem blocking success in coffee production. They complained that the usual Simbu buyers no longer ventured into their area, and they debated whether harvesting was worth the effort that season. Some families had their children diligently pulping coffee cherries into old tin cans; others left their trees full of ripe red berries for all to see. Individuals questioned stated that while some buyers were lazy or had forgotten their Bundi cousins, inadequate roads contributed heavily to buyers' neglect of their community (personal observation, 1992). When plantations were first established around Goroka after World War II, there was no road to the coast and all coffee was flown out by DC3 aircraft (Stewart 1992). The Highlands Highway, the only permanent stretch of road in the whole country, was built in the 1960's specifically to transport coffee to the coast (Good 1986). While it is now easy to transport coffee from the major towns along the highway to the coast, getting coffee from villages to these major towns is still quite difficult due to lack of roads. Maclean reports of the Kwima Maring that having access to a road was so important that the community took it into their own hands to build one in the 1970's, and that "enemies or strangers often worked side by side" (1981:39). The Gende also boasted roads built by the community to facilitate business. However, heavy rains often made roads impassable, so that, for example, landslides in Kwima meant moving the location of sale to another area; "coffee and trade store goods had to be carried for 4 1/2 to 6 hours instead of 2 1/2 to 3 hours as in the past. This has had an adverse affect on trade" (41). Maclean states that "the local population views further economic development, especially expansion of coffee plantings, as dependent upon the construction of roads which will facilitate markets" (39). I have heard similar assertions from people living in Morobe, Madang, and Simbu Provinces with people citing lack of a road, lack of road maintenance, or lack of transportation as the most important hindrances to successful coffee production. It is certain that being able to get coffee out of the village and into the hands of a buyer is a key factor influencing coffee production. Whether access to a road actually affects investment in coffee is not clear. It may be more likely that people store their parchment coffee or do not bother to harvest their trees when the likelihood of being able to see coffee is lessened. Maclean found that the Kwima, faced by continual landslides and lack of government support, either worked to maintain the road themselves, or were willing to carry their coffee further to sell it. In contrast, Bambok village, a long days' walk from the nearest road, was surrounded by coffee trees grown wild. The whole village had decided that the benefits of coffee did not outweigh the costs, and had abandoned their trees (personal observation in Morobe Province, 1988). Maclean points out in a telling anecdote that roads cannot be the only limiting variable to coffee production: The Kwima think that the Tabibuga have many opportunities to make money because they have the road...At a bride price presentation, the Kwima contrasted themselves to the Tabibuga, saying that they had no road...All they had was the odd bag of coffee which they had to carry great distances to sell. The reply form the Tabibuga people was enlightening. "You say we have lots of money and "bisnis' because of the road...Well I tell you I spend a lot of time working on that road. When I want to clear a garden they tell me to work on the road. When I want to build a new house they tell me to work on the road. And I see lots of cargo and wealth going back and forth along this road but I don't know where it goes. Certainly none of it comes into my hands." (1981:47) What other variables, then, constrain coffee production and determine whether "cargo and wealth" come into people's hands? While there are probably a number of factors, the household micro-level variables of land, labor and capital will be the ones that concern us now. Land Territory in the Highlands is owned and defended by a tribe or clan as a corporate group. However, individual gardens within the clan's territory are the property and responsibility of individual men, and upon marriage a man is granted his own land for gardening. In many areas of the Highlands there is no exclusive and permanent individual possession of land. Men readily transfer tracts of land to fellow kinsmen; land is informally redistributed when men die or younger men set up a new household; and relatives from other groups, affines, and friends lend or make gifts of land (Brown 1978). As Strathern points out, "Rules of group membership and access to land were flexible...small-scale readjustments and transfers of claims could occur" (1982a"141). Thus, the decision to plant coffee is not a once in a life time decision, and over the developmental cycle of his family, a man may have many opportunities to assist his parents with their coffee gardens, start his own, expand his holdings or, theoretically, loan them out to someone else. There are conflicting reports as to whether there is a shortage of land in the Highlands. Grossman states that land is still widely available among the Kapanara, and that people were still expanding their gardens in the late 1970's (1984). Good, in contrast, states that land is short in many areas of the Highlands due to rapidly rising population levels (1986). Strathern concludes that there has been a "universal change in the perception of land as a resource which may be exchanged against money and/or may be used to produce goods convertible into money. Such a perception is itself likely to increase 'land shortage'" (1982b:142). Boserup proposed that population pressure is the causal mechanism that forces farmers to use their land and labor more intensively. Thus, "when populations are less dense, farmers will make decisions based on the returns to labor. Only when pressure on the land reaches a certain point are returns to each land unit the prime criterion of land use decisions" (Barlett 1978:19). Highlanders do not seem to cultivate their coffee gardens intensively. Most authors note that smallholders provide only a low level of care which tends to constrain yields; they rarely weed or prune trees, and never mulch. Thus, if intensity of land use is any indication of availability of land, then one could conclude that in most areas of the Highlands, land is not short. However, Grossman states that every married family in Kapanara owns some coffee, and Sexton found that parents want their sons to inherit their coffee gardens. Coffee, is should be recalled, is usually a 20 to 30 year investment of land. It would seem, then, that with increasing population, land will eventually not be abundant, land ownership will no longer be flexible, and coffee may not so readily be an option for all. While at this point land may not be short, the use of land for coffee production does seem to entail changes in the allocation of land for other agricultural products. Most Highland families plant two types of gardens: one for sweet potatoes, the staple crop for both the family and the pigs they raise, and a mixed vegetable garden of beans, greens, tomatoes, sugar cane, bananas and yams. Mixed vegetable gardens tend to be planted near the home, while sweet potato gardens are located further away. Coffee gardens have consistently displaced these mixed vegetable gardens in all parts of the Highlands (Strathern 1982, Brookfield 1973, Grossman 1984). Brookfield found in a study of Simbu land use that from 1958 until 1967 Total area under cultivation, and under cultivation per head of population, increased substantially. Both open field (sweet potato) and coffee area increased, while mixed gardens declined greatly...Almost all coffee is in the central block, while the net increase in food-crop area is almost wholly in the outlying tracts (1973:139). The decision to invest in coffee seems to entail reducing mixed vegetable plots and planting these plots further from the house as coffee is planted closer. The proximity issue is perhaps easier to understand: people want their coffee gardens close to home because theft of coffee is a ubiquitous problem (Grossman 1984, Strathern 1982, Good 1986, Sexton 1986). Grossman states that old people guard coffee beans drying at home while other family members are away working, while Sexton notes that people who must plant their coffee far from the village often move their houses away from the nucleated settlement and closer to the coffee garden to prevent theft. Why it is mixed vegetable gardens which are displaced or reduced is a more difficult issue. Brookfield asserts that the nutrients these gardens supply is replaced by foods that can be bought, but he does not specify what these foods are, where they are bought, or whether in fact people actually buy them. In any case, Brookfield's findings concerning land allocation suggest that when investing in coffee production, smallholders attempt to maintain sweet potato production at normal levels and are willing to sacrifice vegetables. Significantly, in Simbu and many areas of the Highlands, up to 50% of the sweet potatoes produced are used to raise pigs (Grossman 1984). Pigs are prestigious wealth and use objects which are important for bridewealth and other ceremonial transactions. Brookfield's findings concerning Simbu investment in land for coffee and sweet potato production suggest that the Simbu are trying not to make a decision between pigs and coffee in their agricultural production strategies. Investing in coffee production and other forms of cash procurement while continuing to raise pigs has been interpreted as men's attempt to maintain two competing strategies for prestige. While this interpretation may be correct, it is also true that pigs are no longer simply prestige objects to be competitively exchanged by Big Men or clans. Pigs are also sold as a source of income: when asked how they paid for children's school fees, many Gende women said that since coffee buyers no longer came to their villages, they now raised pigs to be sold (personal survey of 35 women, 1991). While most smallholders may have enough land to invest in both pigs and coffee, it is doubtful that they all have the labor resources to do so. Labor Given the rather amazing financial accomplishments of clan based corporations like the Gouna Development Corporation and individual entrepreneurial Big Men (Finney 1986), it is not surprising that much of the anthropological work that has been done on coffee production in Papua New Guinea has neglected household production to focus on the Big Man/entrepreneur and/or clan solidarity (Strathern 1979, 1982a, 1982b; Finney 1968, 1973, 1987). Finney argues that Gorokan society was preadapted for capitalism precisely because of the labor and capital a Big Man could marshall to his cause. Strathern also argues that since warfare was subdued in the colonial period, clans, facilitated by government policy, have invested in coffee, and now compete in the form of business oriented corporate groups. He notes that when warfare does occur, clans will often make a point of uprooting their enemies' coffee trees to debilitate them and assure that they will not be a strong force again for a number of years (1982:142). However, Stewart, a political economist unburdened by anthropological biases, notes that 70% of Papua New Guinea's coffee is produced by nuclear family based smallholders, and has been since 1977. What then are we to make of the ethnographic emphasis on the Big Man and the clan? There are a number of reasons why the clan might have been foregrounded as the unit of analysis. First, anthropologists initially interested in the Highlands were trained to use social structure as a key to understanding society, and social structure in the Highlands is oriented around agnatic kin groups. While the actual groups do not, in fact, rigidly conform to descent group membership rules (Barnes 1962), the male groups formed, through agnation or otherwise, tend to think of themselves in patrilineal terms and employ a strong agnatic rhetoric to accomplish things as a group (Strathern 1982). Secondly, until recently there has been, with some exceptions (Strathern 1972), a strong emphasis on men's activities in Highland ethnography, whether it be exchange rituals, warfare, or, apparently, coffee production. And finally, since the development corporations grew conspicuously and rapidly right after the colonial period and were encouraged by government policy, they might have been the most obvious and interesting features of coffee production at that time: Finney asserted in a 1987 follow-up to his research in the 1960's that coffee production seemed to be more oriented towards nuclear families than when he was first studying the emergence of the capitalist Big Man. While Stewart's data suggest that smallholder production was just as important then, perhaps nuclear family enterprises were not as conspicuous. Those few studies that have examined smallholder production have tended to focus on the male head of the household. Hayano found of the Awa that it was younger men who had spent time outside the village in wage labor who were successful coffee producers (1973). Harris, in contrast, found of Southern Highlanders that it was older men who invested in coffee in order to compete with the large numbers of out-migrating younger men who returned home with cash (1972). Not one of these studies, however, examined the actual processes of coffee production and measured precisely how much time is put in by whom to maintain a coffee garden of a certain size. As described above, arabica coffee production and processing to the parchment stage is labor intensive. It involves going to the small trees weekly to harvest ripe cherries, carrying them to where they are to be pulped, pulping them by hand or by means of a hand-turned pulper, carrying water for the fermentation and washing of the beans, turning and guarding the beans as they dry, and carrying the dried beans to the nearest road if a buyer cannot come directly to the village. One agricultural economist estimated that for a household to maintain a plot of 400 trees, 31 days of the 4 month ripening period must be devoted to coffee production (Anderson 1977, in Stewart 1992:59-61). On these days 6 people must spend 8 hours harvesting, and another 3 hours must be spent by two people pulping and washing the cherries. Stewart claims that the one available study of smallholder coffee production is inadequate; thus, his assertions concerning the time required for harvesting are not actually measurements for smallholders. Rather, they are estimates based on the speed at which plantation workers can harvest, calculating that smallholders could attain one quarter the productivity (Haarer 1962, in Stewart 1992:60). It is my opinion that these estimates underestimate how quickly smallholders can harvest and ignore the amount of picking children often do; therefore, his labor calculations for harvesting may be high. However, his calculation that pulping takes only 3 hours per harvesting day is based on the assumption that all smallholders have mechanical pulpers. Grossman found that 31% of Kapanara households did not have access to pulpers and removed cherry skins by hand. Gende children are sometimes used to hand pulp coffee cherries (personal observation), and Gainj women remove cherry skins with their teeth (Patricia Johnson, personal communication). There are no data to my knowledge on the percentage of Highlands households with access to mechanical pulpers. Therefore, Stewart's estimates concerning pulping may not be generalizable. Despite these qualifications, it is clear that coffee production requires a huge investment of labor resources. Household labor resources are not infinite and choices must be made. Grossman quotes the Kapanara as saying "kofi kalabusim mipela", coffee imprisons us. He found that during coffee season the amount of time spent gardening decreased and the quality of pig husbandry declined (see Diagram 3: Grossman 1984); pigs were fed fewer sweet potatoes less frequently, and consequently there was significantly more damage to people's gardens as pigs foraged for themselves. It is important to note, however, that Grossman's study was done during the coffee boom when prices for coffee skyrocketed, and that Kapanara is on a reliable road not far from Kainantu, a major town on the Highlands Highway. Intensive investment in coffee would have been an easy and profitable decision for the Kapanara during this period. Perhaps the most important labor issue in smallholder coffee production is that it is women who do most of the work. This issue has been somewhat obscured by the emphasis on entrepreneurial clans and Big Men, and by the failure to carry out detailed analyses of coffee production. While development corporation plantations may depend on cooperative clan labor, it is primarily the individual household that carries out all the labor involved in smallholder coffee production (Sexton 1986, Johnson, 1988, Grossman 1984, Strathern 1982). In the arena of household food production, women are usually responsible for planting and harvesting sweet potatoes and other vegetable crops, while men plant "male" crops which include most trees. Women, in most areas of the Highlands, may harvest the fruit of trees planted by men. Coffee trees have assumed the status of most other tree crops in the Highlands. While men clear the ground, plant the coffee trees and perhaps spray insecticides, the existing literature suggests that women do most of the harvesting, much of the pulping, and all of the carrying. Johnson found of the Gainj, a fringe highland group, that household success in coffee production correlated most highly with the number of unmarried women in the household between the ages of 20 and 60 (1988). While 32% of these women were widows, 68% were young women either belonging to the household and not yet married or brought in from other households. In Johnson's study, this variable of unmarried women was more important than number of wives. She concludes that while wives must be involved in child care, and I would add pig care, unmarried women can be devoted mostly to coffee and food production. Strathern also notes that taking on new wives is not an easy route to recruiting labor: wives are expensive and do not often work together. In any case, it would seem that the amount of female labor a man has in his household is perhaps the most important variable in determining whether he can invest in coffee production or not. This is perhaps not surprising since women's labor is the limiting factor on food and pig production also. The importance of female labor in coffee production is having a marked affect on gender relations. Johnson notes that widows, traditionally suspect members of society, are now more easily accepted into new households. She says that they are quite aware that they are working harder, but appreciate the fact that they are no longer so isolated and are recruited into new households usually on the basis of effective ties with female household members (personal communication). Money derived from coffee is an object of contention also influencing gender roles and relations. Finney noted of Gorokan entrepreneurs that they tended to be "'conspicuous investors', gaining prestige and status by investing in visible commercial assets" (1987:ix) such as tradestores or PMVs (public motor vehicles). Brookfield found, in contrast, that smallholders do not accumulate their earnings to reinvest, but rather engage in short term consumption of tradestore goods (1973). Grossman, Good, and Sexton also found that coffee profits were used to buy consumer goods, particularly beer. Men's use of coffee money to buy beer has been contested in various ways by women in many areas of the Highlands. In Simbu Province and Eastern Highlands Province women have joined to form "wok meri" (women's work) groups. These groups have developed extensive saving and loan networks amongst each other, and created elaborate rituals based on traditional bridegiving ceremonies. Upon becoming a member of a wok meri group, a woman will donate money set aside from selling coffee or produce. After thousands of dollars have been saved, a group may invest in a community tradestore or PMV business, or may make a loan to a related group Sexton 1986). In Kapanara, Grossman found that women sometimes own their own coffee gardens. While men and women usually cooperate in harvesting coffee, women will sell and keep the proceeds of the coffee they pick. These strategies pursued by women as groups or individuals are explicit responses to men's expenditure of coffee money on beer, and are effective ways of keeping some cash out of this consumption pattern. However, both Grossman and Sexton did their work when coffee prices were extremely high and men could probably afford to give women coffee money. It is not clear whether these strategies are still being pursued now that coffee prices are low. Strathern found that coffee has an ambiguous position among Hageners. As the product of trees planted by men, coffee is rightfully owned by men. Moreover, coffee is intercropped with casuarina, another tree traditionally used by men for building fences and houses. However, the products of female labor, such as sweet potatoes usually belong to women to distribute as they see fit. Coffee is, for the most part, a product of female labor, Sometimes women are permitted to sell coffee and keep the proceeds, but men often remind women that the land on which coffee is planted belongs to the clan and that coffee is ultimately men's property to be used for male dominated projects such as clan prestations. Since vegetable gardens are usually sacrificed to plant coffee and women often sell extra produce as a source of cash, women are the double losers if they fail to make a convincing claim on coffee proceeds. It is not clear from any of these ethnographies whether men would articulate access to female labor as the variable determining their ability to invest in coffee. Sexton notes that in the Daulo region women are clearly associated with coffee. Women decorate themselves with springs of coffee when they marry, and people say that coffee is like a woman in that it bears fruit that will bring a family wealth, just as women bear children who will obligate affines to make wealth payments to a man's clan (1986). And in Bundi men and women agree that "meri-ol kofi bilong mipela", women are our coffee; in other words, I was told since coffee prices are low and coffee buyers rarely come to the village, Bundi people try to obtain cash by marrying their daughters to the wealthier Simbu or Coastal people for a high brideprice. However, nowhere in the literature does it say, " I wanted to invest in coffee but I couldn't because I don't have enough women to work for me." Whether the number of women in his household enters a man's decision making processes is not known. Capital Smallholder coffee production does not require much capital. Buying a hand-cranked mechanical pulper is about the only way to replace labor with capital. Bush knives are also necessary for weeding and pruning, but most households use these items in their food gardens, and they would not require an extra expense. If buyers do not come to the village, capital may be necessary to pay for transport of coffee beans to town. Capital for living expenses is needed in large amounts these days. School fees and brideprice both require cash. Tradestore goods also require cash, but there has been little research done on how much a smallholder family tends to spend on such items as rice or tinned mackerel or corned beef. Nor has there been any study of a possible relationship between coffee production and the amount or kinds of tradestore goods bought. Most households do not have many means of getting cash. Women often sell extra produce at markets but the income from this is not large. Some families receive remittances from kin earning wages, but there have been few studies of the percentage of families benefitting from remittances, how much they receive, and how reliable such a source is. Conclusion It has been shown that the history and structure of the international market have forced Papua New Guinea into producing labor intensive washed arabica coffee. Topography and lack of infrastructure have meant that smallholders must process their coffee to the parchment stage, making coffee production yet more intensive. Since the labor required for coffee production must be supplied primarily by each individual household, and since the burden of agricultural labor falls mainly to women, a household's access to women's labor is probably the most significant variable in smallholder's ability to invest in coffee production. Although no time allocation studies or large scale statistical analyses have been done test this hypothesis, Patricia Johnson's study among the Gainj suggest that it is correct. However, it is probable that interviewing smallholders would reveal other concerns, such as maintenance of roads, access to buyers, or prices for coffee. And these are in fact important variables in coffee production. But it may be that the state of roads or vagaries of the international market only influence whether smallholders bother to harvest their coffee or make an effort to sell it during a particular season. At this point in the history of smallholder production, a drop in the price of coffee or a landslide may simply mean that one stores parchment coffee until better times or allows coffee cherries to rot on the ground for one season. It probably does not mean that one digs up coffee plantings to allocate land and labor to some other alternative. The foregoing discussion of smallholder coffee production demonstrates the need for future research. First, the effect of economic change on the lives of women should be investigated. One would like to know how much work women are putting into coffee in relation to other tasks, what rights women have over coffee money, and to what degree women contribute to decisions concerning household investment in coffee production. Some studies on the impact of development on women have suggested that with the expansion of cash cropping, women's labor increases while their power decreases (Tinker 1990, Bossen 1984). It is not clear to what extent this may be so in the Papua New Guinea context. While their work may be increasing, if women can make convincing claims to coffee money, they may be benefitting more in terms of economic and social power from their labor than they have in the past from raising pigs for male dominated ceremonial exchange. If not, cash cropping may mean decreased power and prestige for women. Finally, it must be acknowledged that coffee and the production of other commodities have long been important to Papua New Gunineans themselves, whether anthropologists choose to study this facet of their lives or not. The ways in which people act to benefit from commodity production, wage labor, and other aspects of a market economy are making for interesting changes in Papua New Guinea that anthropologists should pay attention to.

References Cited

Amarchi, A. K. Good and R. Mortimer 1979 Development and Dependency: The Political Economy of Papua New Guinea. Melbourne: Oxford University Press.

Araim, A. S. 1991 Invergovernmental Commodity Organizations and the New International Economic Order. New York: Praeger Publishers.

Barlett, Peggy 1978 What Shall We Grow?: A Critical Survey of the Literature on Farmers' Decision Making. Prepared for The Civic Participation Division: Bureau for Program and Policy Coordination, Agency for International Development.

1980a Introduction: Development Issues and Economic Anthropology. In P. Barlett (ed.), Agricultural Decision Making: Anthropological Contributions to Rural Development. Orlando: Academic Press.

1980b Cost-Benefit Analysis: A Test of Alternative Methodologies. In P. Barlett (ed.), Agricultural Decision Making. 1980c The Anthropological Approach to Fertility Decision-Making. In T. Burch (ed.), Demographic Behavior. Boulder: Westview Press.

Barnes, J.A. 1962 African Models in the New Guinea Highlands. Man 62:5-9.

Bossen, Laurel 1984 The Redivision of Labor: Women and Economic Choice in Four Guatemalam Communities. Albany: State University of New York Press.

1989 Women and Economic Institutions. In S. Plattner (ed.), Economic Anthropology. Stanford: Stanford University Press.

Brookfield, Harrold 1973 The Pacific in Transition: Geographical Perspectives on Adaptation and Change. New York: St. Martin's Press.

Brown, Paula 1964 Highland Peoples of New Guinea. New York: Cambridge University Press.

Finney, Ben R. 1968 Bigfellow Man Belong Business in New Guinea. Ethnology 7:394-410.

1973 Big-Men and Business: Entrepreneurship and Economic Growth in the New Guinea Highlands. Honolulu: University Press of Hawaii.

1987 Business Development in the Highlands of Papua New Guinea. Honolulu: Pacific Islands Development Program.

Good, Kenneth 1986 Papua New Guinea: A False Economy. London: Anti-Slavery Society.

Graaff, J. de 1986 The Economics of Coffee: Economics of Crops in Developing Countries No. 1. The Netherlands: Pudoc Waganingen.

Grossman, Lawrence S. 1984 Peasants, Subsistence Ecology, and Development in the Highlands of Papua New Guinea. Princeton: Princeton University Press.

Harris, G.T. 1972 Labor Supply and Economic Development in the Southern Highlands. Oceania 43(2):123-139.

Hayano, D. 1973 Individual Correlates of Coffee Adoption in New Guinea Highlands. Human Organization 32:305-314.

Johnson, Patricia 1988 Women and Development: A Highlands New Guinea Experiment. Human Ecology 16(2):105-122.

Lucier, L.L. 1988 The International Political Economy of Coffee: From Juan Valdez to Yank's Diner. New York: Praeger Publishers.

Maclean, N.L. 1981 Politics of Development in an Underdeveloped Area: A Case From the Jimi Valley. In The Plight of Peripheral People in Papua New Guinea: Volume I. The Inland Situation. Cambridge: Woodhead-Faulkner Ltd.

Ortiz, Sutti 1980 Forecasts, Decisions, and the Farmer's Response to Uncertain Environments. In P. Bartlett (ed.), Agricultural Decision Making.

Roseberry, William 1989 Peasants and the World. In S. Plattner (ed.), Economic Anthropology. Stanford: Stanford University Press.

Sexton, Lorraine 1986 Mothers of Money, Daughters of Coffee: The Wok Meri Movement. Studies in Cultural Anthropology, No. 10. Ann Arbor: UMI Research Press.

Singh, S., J. de Vries, J. Hulley and P. Yeung 1977 Coffee, Tea and Cocoa: Market Prospects and Development Lending: World Bank Staff Occasional Papers #22. Baltimore: John Hopkins University Press.

Stewart, R. G. 1992 Coffee: The Political Economy of an Export Industry in Papua New Guinea. Boulder: Westview Press.

Stolcke, Verena 1988 Coffee Planters, Workers and Wives: Class Conflict and Gender Relations on Sao Paolo Plantations, 1850-1980. Oxford: The Macmillan Press.

Strathern, Andrew 1979 Gender, Ideology and Money in Mount Hagen. Man 14:530-548. 1982 The Division of Labor and Social Change. American Ethnologist 9:307-329.

1982b Tribesmen into Peasants. In A. Strathern (ed.), Inequality in New Guinea Highlands Societies. Cambridge: Cambridge University Press.

1984 A Line of Power. London: Tavistock Publications.

Tinker, Irene 1990 A Context for the Field and Book. In I. Tinker (ed.), Persistent Inequalities: Women and World Development. New York: Oxford University Press.