---_----1 \ . -;-.-,,~" Jr;.lT ... ' r{1 ,ti. \U ~IAN DEMAND S lES \ CO.~~uOM HIS10RlCA --- '¡ CC~~TT BIBLIOTECA A Discussion Document for the Annual Review 1986/87 Centro Internacional de Agricultura Tropical \ l J...I ~12~ : Asian Demand Study: Status Report The economice study of cassava in Asia focuses on detailing the current status of the crop in the principal producing countries. The study relies almost exclusively on aecondary data aources. The only primary data collection involved a cost survey of chipping and pelleting factories in Thailand. Depending on existing data sources has often left araas where further detail would have been valuable. This is particularly true' for production issues. Nevertheless, Asian countries have well developed data systems, and the economic iasues facing cassava could be explored at a sufficient level of detail to give a reasonable outline of a feasible strategy for the crop in the region. The study ie essentially complete in terms of its major findings. A few sections of the study still remains to be completed. !he 1ntroductory chapter ie still being written. It ia an historical chapter and has depended on some major bibliographic research which is st111 underway, especially on early Spanish or Portuguese dissemination of the crop in Asia. The rest of the study 1e essentially written except the animal feed sections 1n the Malaysia and China chaptera, and the conclusions section to the Ph1l1pp1nes, Malaysia, China, and world trade chaptere. F1nally, time has not permitted the development and typing of all the tables and figures. In terms of the scope of the study the only significant gaps in a comp~ete regional study are analyses of the cassava sector in Vietnam and Sri Lanka. Sri Lanka is the smallest producer in the region with distinct policies that negat1vely affect cassava and Vietnam, while a larger producer than Malaysia or the Phil1ppines was not accessible. Leaving out these two countries does not greatly diminish the conclusions of this regional analysis. r;' 2'.).. '1.. ..(. The Cassava Economy of Asia: Adapting to Economic Change 1 Taking Root: A History of Cassava in Asia 1. Introduction and early distribution 2. The integration of cassava into Asian food economies 3. The advent of steamship trade and the rise oi tapioca pearl 4. Shifting comparative advantage: Malaysia, Indonesia and­ Thailand 5. Cassava: a crop in permanent transition II India: Kerala and Tamil Nadu 1 • Produc t ion a. Production trends and distribut10n b. Cassava product1on systems l. Kerala 2. Tamil Nadu c. Yields d. Costs of production and labor utilizaiton e. Technology 2. Markets and Demand a. A synthesis of production and utilization l. Kerala 2 . Tamil Nadu 3. Other Sta tes b. Cassava for direct human consumption c. The dried chip market d. The starch market e. Pricing and market efficiency 3. Conclusions 111 Trends and Distribution of Chinase Cassava Production and Use (by Bruce Stone, IFPRI) 1. Production a. Production trends and distribution b. Cassava production systems c. Yields d. Costs of production and labor utilization e. Technology development 2. Markets and demand a. A synthesis of production and utilization b. Cassava for direct human consumption c. The starch market d. The domestic and export animal feed market 3. Conclusions IV Indonesia· lo Production a. Production trends and distribution b. Cassava production systems c. Yields d. Costs of production and labor utilization e. Technology development 2. Markets and Demand a. A synthesis of production and utilization b. Cassava for direct human consumption c. The starch market d. Gaplek in feed markets e. Pricing and market efficiency 3. Conclusions V Malaysia 1. Production a. Production trends b. Cassavs production systems c. Yields d. Costs oí production and labor utilization e. Technology development 2. Markets snd Demand a. A synthesis of production and utilization b. The domestic and export market for starch c. The domestic animal feed market d. Pricing and market efficiency 3. Conclusions VI Philippines 1. Production a. Production trends and utilization b. Cassava production systems c. Yields . d. Costs of production and labor utilization e. Technology development 2. Markets and Demand a. A synthesis of production and utilization b. Cassava for direct human consumption c. The starch market d. The dried chip market e. Pricing and market efficiency 3. Conclusions VII Thailand l. Production a. Production trends b. Cassava production systems c. Yields d. Costs of production and labor utilization e. Supply response f. Technology development 2. Markets and Demand a. A synthesis of production aud utilization b. The cassava pellet export market l. Price formation 2. Profitability of the cassava pellet industry c. The cassava starch market d. The animal feed market 3. Conclusions VIII World and Regional Markets for Cassava Products l. Protectionism and Súhstitution: Decline in the World Stsrch Trade 2. Protectionism snd Substitution: The Rise in Trade in Cassava • Feedstuffs a. Demand for cassava in the EEC b. The Asian regional market for cassava feedstuffs 3. Conclusions IX A Comparative Analysis of Cassava Production and Utilization in Tropical Asia l. A Comparative Analysis of Production 2. A Comparative Analysis of Consumption 3. Marketing aud Price Formation 4. Cassava's Future Role in Asia INDIA = KERALA AND TAMIL NADU lt i5 almost an aphorism that India i8 a vast, diverse "sub-continent", where over three-quarters of the 684 million people (1981 Census) live in the rural sector subject to the vagaries of the annual monsoons. As might be expected a major concern of agricultural policy has been the capacity of India to feed itself and this has resulted in a commitmentto attaining self-sufficiency in food grain production. This goal was "achieved in the mid-1970's, essentially by focus1ng on development of the more productive agricultural regions (Sarma, 1982). Self-sufficiency, while implying a termination in imports, is nevertheless a relativa concapt because it implies that demand i8 defined by production availability rather than by consumption needs. The central government has attempted to overcome this problem by intervening in grain marketing to manage demando The government operates a public food distribution system at subsidize prices to ensure that a certain m1nimum level of universal distribution for food grains is achieved independent of income levels. As Sama has noted, "This (self-sufficiency) strategy, which was confined to certain crops and are as with assured irrigation, also resulted in the widening of interpersonal and interregional disparities.... The social justiee objective. in terms of reducing unemployment or underemployment and alleviating poverty in rural areas, remained largely unfuIfilled" (p. 24). The cassava-growing areas in the south of India have been such a region which has remained largely outs1de the area of impact" oi the Ifgreen revolution" technology. Although cassava ia very much a regional crop in India, thia ia true of moet other crops except rice. Analyzing cassava in southern India thus provides some insight into rectifying the dieparities between regiona in India. z PRODUCTION Production Trends and Distribution: Cassava is very mueh a regional crop in India, although given the size and diversity of the country, this could be said of most any crop. Cassava production i9 concentrated in the 90uth of India in the state of Kerala and the western part of Tamil Nadu. These two states make up 97% of caSSava production in India (Table 1). On a country wide basis cassava makes only a amal1 contribution to total calorie supplies, with production being more or less equivalent to some of the minor coarse grains, su eh as barly or che small millets. However, in the south of the eountry cassava ranks second to rice as the major calorie producing erop. Given the range of temperature and rainfall conditions in India, this type of regional specialization in crop production would be expected for non-irrigated crops. According to the official data series, area planted to cassava in India increased slowly from the mid-sixties to the mid-seventiea, reaching a peak area of 392 thousand hectares in 1975-76 (Table 1). Since then cassava area has declined quite markedly, reaching a level of 310 thousand hectares in 1981-82. The trends in area are due principally to changes in cassava plantings in Kerala. Cassava has been widely planted in Kerala since at least the turn of the century. In the 55-year period from 1920 to 1975 cassava area expanded at a relatively slow and uneven rate of 1.3% per annum (Table 2). Since 1975 cassava area has declined rapidly to the same level as the early sixties. On the other hand, area planted to Cassava in Tamil Nadu has remained relatively constant at around 50 thousand hectares since the late 1960's. Production trends are more difficult to evaluate sinee the basia on which yie1d has heell estimated has heen changed twice. In 1963 yield levels in Kerala were revised sharply upward from a trend of 7 tlha to a rising yield trend starting at 12 t/ha. In 1979 a erop cutting survey was in8tituted in Kerala and Tamil liIadu and whst had been a rising trend in yields in Kerala was revised downward. In Tamil Nadu, on the other hand, yield estimates were dramatically increased. Given these revisions in yield estimates, production trends, which follow from the area snd yield estimates, are somewhat meaningless. What can be said with some degree of confidenee 18 that production in Kerala has declinad markedly since 1975 at "an annual rate oC about 5% per annurn. Cassava production in Tamil Nadu in the same period has shown a slight increase. The dorninant question that arises is the reason behind the declining area and production of cassava Kerala. Cassava production systems: Kerala: Kerala i5 one of the mast populous rural areas in the trop1cs. Population densities in sorne districts exceed 1000 people per square kilometer. About 81% of the population reside in the rural area according to the 1981 eensus, while a little less than half of the work force are d1rectly involved in agrieulture. However, a more accurate reflection of the population pressure is that while average farm size is on1y 0.49 of a hectare, only one third of the work force in the agricultural sector have access to land. Moreover, over 70% of the population who do own land have leas than half a hectare (Table 3). 3 As a consequence óf this population pressure, land use 15 very 1ntensive. Exclud1ng forest reserves and non-agricultural uses, 87% of available land i5 cultivated. The cropping intel\sity index in Kerala in 1977 /78 was 132 percant, well aboye the average for India as a whole. However, this figure i5 more remarkable when ie 1s considered that two-thirds of cult1vated area 1s under permanent tree crops. Thus, for area under annual crops the eropplng intensity index i5 192 percent; that is, a substantial portian of the land under annual erops is double or triple eropped. Cassava 18 the most important annual erop 1n Kerala after rice, making up 38% of the net are a sown to annual crops. Two faetors explain why eassava has achieved sueh importanee in sueh an intensive agricultural system. First, the non-irrigated upland areas are eharaeterized by lateritie soils which are Iow in inherent 5011 fertility, espeeially phosphorus, and are quite acidic. Cassava in eomparison to most other annual crops, is well adapted ta sueh 50ils, even with relatively minimal amounts of fertilizer. Second, ca8sava g1ves very high carbohydrate y1elds under these conditions. W1th average y1elds around 15 t/ha oo1y triple cropp1ng of r1ce under irrigation gives h1gher dry weight yields in the sta te. While rice is grown on the irrigated bottomland, cassava 15 grown on the sloping upland areas. On these upland s011s cassava competes primarily with tree crops for land and it is the general concensus that cassava i9 being displaced by higher value tree crops. However, for the principal tree crops increased plantings of rubber and cashew nut are more then offset by dec1ining area of coconut and black pepper (Table 4). The crop or crops that are disp1acing cassava remain as unc1ear from the aggregate data but the strongest hypothesis stil1 remains some combination of tree crops. Cassava production systems in Kerala are relatively simple, compared to countries such as Indonesia. This 15 partly due to the constraints on potentia1 intercrops lmposed by so11 conditions. Annual rainfall in the state average s about 3000 mm, and varies from about 2000 mm in the south to 3800 mm in the north. There 1s a long dry period from December to Mareh when little rain at all 18 received. The rains etart in Apr11-May when 60-65% of the cassava crop is sown (Hone, 1973). The monsoons arrive in full force in June-July. From 35-40% of the crop i8 planted in September-October when the rains have fallen off but before the etart of the dry season in December. 1.and preparation is done completely by hand and any green vegetation in the plot i8 concentrated 1n the soi1 below where the cassava stems are to be sown. The stakes are sown vertically at populations of 10 to 12 thousand per hectare. In such intensive systems weed control is fair1y meticulous and when farmyard manure or woad ash is available it is incorporated in the same form as the green manure. Some chemieal fertiIizer is certain1y used on cassava in Kerala, although there is conflicting data to suggest just how extensive this use 18. Certainly potassium fertilizer consumption is a much higher percentage of total fertilizer consumption in Kerala than in India as a whole (33.~3% of eonsumption as compared to 11.4% in the whole country). Cassava has a higher potassium requirement than grain erops. A National Couneil of Applied Economic Researcb 9urvey in 1975/76 found tbat 83% of casssva area in Kerala was fertilized but that only 19 kg/ha of nutrients were applied to the area fertilized. Desai (1982) has found this survey to substantially overestimate aggregate fertilizer consumption in Kerals. 'He provides estimates for India as a whole, suggesting that in 1976/77, 38.2% of e8SS8va area was fertilized at arate of 33 kg/ha. Tbe limited data available thus suggests thst thare i9 some fertilization of eassava but at very low rates of application. Tbe cassaVa is harvested at about 10 months, with the bulk of the crop heing harvested in the dry period from December to February.Tbe percentage of the crop that 1s sold off the farro is open to some question. A relatively dated report (Tapioca Market Expansion Board, 1972) estimates that about 40% of produetion enters market ehannels (Table 5). Tbis would appear a b1t low considering that easSava i9 sueh a pervasive consumption item in Kerala, that about two-thirds of households in Kerala do not grow cassava, snd thst household consumption surveys show higher consumption levels for purchased cassava than own produetion (Table 6). Tbe pervers1ty of the lattar ls due tú the positiva relation between income and land ownership in Kerala and the shitt from cassava to rice at higher incornas. 40% is then probably a minimum estimate oi marketed surplus oi csssavs in Kerala. Tbe most common practice i8 for farmera to sell the standing cassava crop to purcha8e agents for a lump suro payment. Tbe agent", do not necessarily harvest straight away but must harvest before the start of the raina. Farmers, as well, gradually harvest the crop themselves, selling in small lots by the roadside or in local roarkets. When marketing of the fresh root is, problematic, particularly in the north of Kerala, the roots are peeled, sliead and dried as chips during the principal harvest period in the dry season. Wholesale merchants and weekly markets serve as assembly points for roots and chips. Taroil Nadu: Tbe other major cassava producing zone is in the western part of Taroil Nadu where production is principally concentrated in Salem District. Production systems for cassava are considerably different from those in Kerala and this arises frem a change in the limiting production constraint from soil factora in Kerala to moisture avai1abi1ity in Tamil Nadu. Rainfall in the major production area of Salem District averages 820 mm per year. This average, however, masks a very high variation, with annual rainfall in the 1as1: tenyears ranging from 550 mm to 1250 mm. !bere is a five-month dry season from January to May when rainfall averages no more than 14 mm in tbe whole periodo Tbis limited rainfall is in many cases supplemented by irrigation. Farro s1ze forcassava farmers in Tamil Nadu are somewhat larger than that in Kerala. A sample of 70 cassava farroers in Salero District found an average farro size of 2.6 hectares, with an average area sown to eassava of .75 ha (Uthamalingam, 1980). Tbe larger farro size reflects in par1: the much drier conditions in Taroi1 Nadu snd the relative scarcity of irrigation water. Cassava is g~o:..m almost- st-rict{y as a cash é~op in these eropping systems and competes for lsnd principally with cotton, and to a leseer extent, rice and sugar caneo Cassava's role in these cropping systems ia defined by ita aCcess to a ready market (the industrial starch market) and cassava' s efficieney in water use. Over 85% of the irrigation water ia provided by wella and the farmer must plan his cropping pattern around expected rainfall and available water stored in the wella. When irrigation water ia in ahort aupply, farmera turn from rice and sugarcane to casaava or cotton, depending on output priees. According to the sample of 70 farma in Salem District, 90% of the farma grew caasava under irrigation. The crop cutting survey in all of Iamil Nadu found that 72% of the plots were grown under irrigation. The irrigated crop ia planted at the end of the rains in January. Up to four or five irrigations are needed for establishment. Frequency of irrigation afterwards depends on water availability in the wells and the arrival of premonsoon showers in June. On average 20 irrigations are given at an interval of 15 to 20 days. The rainfed crop is sown at the atart of the southwest monsoon in August. The crop is aasured of no more than five montha of rainfall before the atart of the dry season in January, which is followed by the pre-monsoon showers in June-July. A rained crop is often grown on as litUe aa 500 mm of rainfall. Ihe irrigated crop is usual1y harvested after 8 to 10 months while the rainfed crop requires 12 manths before it Can be harvested. Land preparation relies on bullocks and for the irrigated crop the lsnd is ploughed four or five times before forming either beds and channels or ridges snd furrows. Plant population 19 approximately 10,OOO/ha. Stakes are sown vertically and normally six or seven weedings are done during the course oí the crop year. Fertllization or manuring i8 a common practice for cassava in Tamil liadu, especially for the irrigated crop. The crop-cutting survey found that 74% of the cassava plots were either fertilized or manured, using either animal manure or a vegetable compost. The farmer survey in Salem found an average application of 18.5 t/ha of farmyard manure or 15.1 t/ha of compost. Manuring ls often combined with application oí compound fertilizer. Moreover, eaasava is uaually planted in rotation with other crops and viII often take advantage of residual fertility from fertilizar application on prior erops. However, whare cassava i6 grown in succes6ive years in the same plot, there is a marked tendency for yield to drop. A typical trend 16 35 t/ha in the first year, 24 t/ha in the second sud 17 t/ha in the third (Tapioca Experiment Station, Salem District, private communication). In·contrast to Kerala most of the cassava is harvested and marketed by farmers; only a small percentage is sold standing in the 10t. In the Salero farm sample 87% of the eassava was marketed directly by farmera. Ihe reason for this la the very decentralized nature of the cassava stareh processing industry. The industry consists of upwardsof 500 relatively small-scale plants distributed throughout the distriet. Coordination of harvesting by the farmer and processing of the fresh roota at the factory are easl1y managed wlthout the naad of middlaman or larga axpandlturea on transporto Yields: By world standards cassava yields in India are high. Yields ln the 1980-81 crap year averaged 16.8 t/ha in Kerala and 28.9 t/ha in Tamil Nadu. With tpe generally intensive level of cultural practices used in Kerala and TamU Nadu, this high yield ia not surprising. The difference in yields between Kerala and Tem1l Nadu ia due essentially to the poorer 90ils in Kerala and the use of irrigat1an and associated higher input levels in TamU Nadu. The author i9 unaware of any farm-level data on distribution of cassava yields in Kerala and therefore of any estimates of yield variance across farms in the state. The district-level data suggest a slight tendency for yields to be higher in the southern and central parts of the state, and lower in the north. Thus, the 1980-81 crop estimates suggest average yields of 15 tlha in the four aouthern districts and of 11 t/ha in Kozhikode and 12 t/ha in Malappuram in the north. This limited data suggest little variation in yields acrosS the state but has litt1e implication for across farm variation. In Tamil Nadu a crop cutting survey in 7 districts ln the state found a significant variation in farm-leve! yields (Table 7). The yield distribution vas skewed tovard the lower side of the mean and as ve11 exhibited a very extended upper tail; that ls, a more or less typical distribution for farm-Ievel cassava yields, apart from-the very high mean. Over 15% of the plots had yields of over 37 t/ha with a maximum yield of 84.2 t!ha. Tamil Nadu provides a perfect example of the yield potential of cassava when grown under very favorable production conditiona. Part of the reason "hy national cassava yields in other parts of Asia never approach such levels i5 that cassava i8 usually grown under more marginal agro-climatic conditions. Yet even vithin a highly productive region such as Tamil Nadu, over a quarter of the farmers are getting less than 15 t/ha. Such typica1 yie1d distributions lie at the heart of production research: vhat factors explain the difference in yields at the lov and high end of the distribution and to what extent are these factora a function of farmer management or a function of more or less uncontrollable biological and edaphic factors facing the farmer? The question ia critical sinee it begs the issue of the substantial yield gap for cassava between experiment station and farm-level yields and how closely experimental yields translate into farm-level yields. Costs of production and labor utilization: In such densely popu1ated rural areas and in such intensive production systems as exist in southern India, the expectation is that relative to "other cassava production areas wage rates wi11 be low, labor input per hectare wi11 be high, inputs that substitute for land wil1 be applied at high levels, and labor costs viII be a lower portion of total costs. The available data suggest per hectare labor inputs of 265 days "for irrigated systems in Tamil Nadu, 139 days for rainfed systems in Tamil Nadu I (Uthanalingam, 1980) and 200 to 220 days for production systems in Kerala (Kerala State Planning Board, private communication). !be breakdown of labor activities for Tamil Nadu shows that weeding is the principal labor requirement, and makes up 60% of total labor demand, with inputs in rainfed systems requiring about half that in irrigated systems (Table 8). Labor for harvesting forros the next major component in botb systems followed by land preparation. Although tbere is no breakdown for Kerala, labor input per activity probably lies somewhere between tbe irrigated and rainfed systems of Tamil Nadu, with the exception that labor tor land preparation in Rerala is much higher. Labor input in cassava systems in India i8 lower than that in Indonesia but significantly higher than labor input in !bailando Malaysia and tbe Philippines. !bis result is expected given the relative differences in the land-labor ratios in the cassava growing regions of the different countries. Moreover, labor costs are a lower proportion of total production costs in India as compared to the latter three countries. In Tamil l'Iadu labor makes up only 35% of variable production costs and less than 20% of total costs. !bis i8 due to the large expenditures on fertilizer and land rental. A comparison of production costs between Kerala and Tamil Nadu (Table 9) ebows that per ton costs are higher in Kerala than Tamll Nadu. The difference i8 due in large part to differences in yield levels, particularly when it ia considered that rainfed eystems in Tamil Nadu are of only marginal importance. Moreover, when average yields reported for the state are used in place of the study' s 8ample yields, the difference becomes even more marked. l'Ieverthelesa, the flow of cassava i8 from Kerala to Tamil Nadu and not vice versa. This is due to the very seasonal nature of cassava supply in Tamil Nadu and the fact that opportunity cost of irrigated land when there is sufficient water is much higher than is reflected in average rental rates. Technology Development: Not on1y ia there very limited potential for expanding are a in cassava in southern India, but competition from other crops has actua11y resulted in declining area planted to cassava in Kerala. There is an obvious demand for technology that would lead to increases in cassavs yields. !be question arises, since the production systems are so intensive and cultural practices are of such a high level, whether there i8 a significant yield gap to exploit? !bis iasue ia at che heart ef the work of the Central Tuber Crops Research Institute (CTCRI) in Kerala. Undar the Indian Council of Agricultural Research the institute as sumes principal responsibility for research on cassava in India. Most of their work i8 focused en conditiona in Kerala .mere research has been carried out since 1963. Independent research on casssva ia carried out in Tamil Nadu at the Tamil Nadu Agricultura! University in Corimbatore and the Tapioca Experiment Station, established in 1971 in Salem District as part af Horticultural Department of Tamil Nadu. !bis division in activities allows research to faeus on the very different preduction systems of Kerala and Tamil "1'Iadu. Moreover. India has had the longest period of continuous research on cassava in Asia. v The search tor yield increasing technology in Kerala has focused ..o n essentially four principal faetors: (a) improved, high-yielding varieties, (b) soi1 terti1ity management, (e) control of African cassava mosaic virus, and (d) intercropping systems. The two principal constraints on increased productivity are perceived to be soil factors and the virus disease. Given the high level of cultural practices in the state, overcoming these two constraints would probably not lead in themselves to much higher yield levels. Major inereases in per hectare productivity would have to combine as well improved varieties and intercropping, wíth the problsm in rhe later being ths identificatíon of an adapted legume crop. During the early years of CTCRI when a germplasm bank was being assembled, one selection from Malaysia, M-4, was released and found wide acceptability with farmers. This variety has since set the standard and developing hybrids to replaee M-4 has been a difficult task. Only five hybrids have been released since the inception of the institute: B-165, H-97, and H-226 in 1970 and H-2304 and H-1687 in 1977. A fertility trial earried out at the experimental atation arguably gives Some indication of potential yield gain with these varieties (Table 10). Average yields of M-4 at intermediate fertilizer levels are at about the state average of 15 t/ha indicating little gain to be aehieved by agronomic practicas. Tha hybrid H-2304 yielded 24 t/ha at intermediate fertilizer levels and 32 t/ha at relatively high fertilizer levels. Because most cassava grown in Kerala is consumed as a boiled root, quality characteristics are very important. This has probably been one of the principal faetora limiting tbe wider adoption of the hybrids. These quality eharacteristics include HCN eontent, short cooking time (due to limited fuel resourees of households), softness with cooking (apparently related to the ratio of anylose to amylopectin), good consistency (high stareh content), and to a more minor extent, whiteness of the flesh (H-1687, for example, is yellowish due to a high carotene content). K-4 is recognized to have good culinary quality and for these properties to be stable across locations and through the growing season. The result is usually a price discount for roots from the hybrids; for example. farm priees of 0.90 rupees/kg for M-4 versus 0.75 rupees/kg for H-1687 (field notes, 1982). Thus, a 25% yield advantage is almost canceled by a 20% price diseount.. Besides higher yielding ability and root quality characteristics, the other major breeding objective i5 field tolerance to cassava moaaie virus. M-4, though brought from Malayaia where the diaease does not exist, has relatively high field toleranee as do almost all the released hybrids. Toleranee does nat imply immunity with this disease and tolerant varieties must be combined with adequate selection of clean planting material. sinee this i5 the principal meana of spreading the disease. Unlike in West Africa where the disease i9 easiIy spread by the white fly vector, effective infection in India is only 2 to 5%. The final two breeding objectives are short maturity and plant type compatible with intercropping systems. The latter 15 complementary to the research on intereropping systems. Most of the cassava in Kerala 1s grown in monoculture, due in large part to the lack of adaptatíon of potential commercial intercrops to the lateritic soils. The institute i9 having some 9 suceess in promoting peanuts as a suitable intererop with cassava. Moreover, sinee eassava is planted continuously for many years in the same plot, maintaining so11 organie matter i8 diffieult. Long terro fertllity trials have shown that applying farro yard manure with fertilizer gives a significantly higher yield than fertilizer alone and that manure appears to be necessary in maintaining yield levels over time (CTCRI, 1980 snd 1982). Incressing cassava production in southern India is dependent on increasing yields. These yield incraaaes in turn, depend on the development of high-yielding varieties that do not !lacr1fi~ quality for yield and that are tolarant to cassava mosaie virus. The improved varieties in turn imply heavier demands on soil fertility and thus higher rates of fertilizer application. Although the researeh objeetives are quite straight forward, after twenty years of consistent breeding effort, CTCRI has found the progress to be slow, in part beeause substantial effort at the beginning had to be devoted to more basie studiea, since little basie researeh had been done on cassava upto that point in time, in part because their varietal evaluation system requires approximately ten years froro cross to potential release of a new variety and, possibly, in part because the reeombination of all desired characters has a low probability. The efforts upto this point in time suggest that a goal of average farm-level yields of 25 t/ha i8 a feasible objective. If the goal is worth pursuing depends in turn on the prospective outlook for utilization of the eassava erop. MARKETS AND DEMAND A synthesis of production and utilization: !he modicum of uncertainty surrounding the cassava production eatimates and the paucity of data on caasava consumption in ita various end uses makes the deve10pment of a consistent supply and distribution series a speculative enterprise. The exerclse wll1 be attempted by f1rst sepsrst1ng Kerala and Taroll Nadu, then rev1ewing the avallable consumptlon data for each state, and flna11y 1ntegrating these estimates with the production estimates. The result wil1 then provide the basis for an evaluation of cassava markets and demand in aouthern India. Kerala: An sna1yais of cassava utilizat10n must beg1n with an est1mate of human consumption of fresh roots. Several estimates ex1st but as can be seen in Table 11, there is a substantial range in these estimates. Given that Kumar' s sample introduces a substantial upward bias in the caasava consumption estimate - consumption is higher in the southern districts, in rural areas, and 1n the lower income strata -, the striking feature is the difference between the estimatea from food balance sheeta and those from sample surveys. The George and Kumar samples have upward biases in their estimates of per eapita consumption. The National Sample Survey is probably the best structured sample and thereby estimate of consumption levels. Since fresh human consumption 1s considered the largest single market for cassava, the difficulty arises of how to account for the difference between the consumer sample estimate and that derived from production estimates. Dried cassava chips are also produced in Kerala, principal1y in the northern districts and primarily in the period October to April. These chips go into various end uses. Dried cassava can be prepared in the home and eaten, especially when fresh cassava is not available. Cassava flour is also produced by grinding the chips. At least one factory operates in Malappuram exactly for this purpose. The flour is in turn usad to produce fine noodles. Often the flour is produced in the home. Also, larga starch factorias also buy chips for processing, particularly for glucose production. Finally, from 1955 to 1966 cassava chips were exported. Afeer that exports ceased until just recently and since 1977 India has again been exporting modest amounts of cassava chips. , Statistics on production and utilization of cassava chips are practically non-existent. The Tapioca Market Expansion Board provides the single estimate of household consumption of processed cassava products and estimates an annual consumption of 9.5 kg per eapita of drled cassava. It can only be assumed that cassava flour is included in this figure. Cassava chip exports were initiated again in 1977 after a lull of about 10 years. Exporta remain small and irregular. Imports into the European Community from India were 7,949 t in 1977. 37,182 t in 1978. 26,799 t in 1979 and 11.915 t in 1980. Chips purchased by the starch factories are assumed to be included in starch production figures. This leaves only potential exports of dried cassava to other states. Data on transport through selected checkposts for the period May 1975 to May 1976 give the following: 11 Quantity (M.T.) Value (100,000 rupees) Tapioca chips N.A. 78.80 Dry Tapioca 90,150 44.34 At the Kozhikode ",holesale price for cassava chips in this period of 62 rupees/lOO kg, the volUllle of tapioca chips implied is 12,710 t. On the other hand, the per ton price for dried cassava imp1ied by the above value and volume figure is 49 rupees/t, a figure undervalued by at 1east a factor of ten. A selection of either the volume or value figure ie arbitrary. Processing the chips into starch is possible but 90 thousand ton s 1s a bit excessive in relation to starch production capacity in Tamil Nadu. Moreover, assemb1y of this volume is a M.t large compared to more recent international export volumes. lt is therefore assumed that 90 thousand quintals were exported to Tamil Nadu, implying a total export volume of 21,725 t. Starch 1a the other major consumption form of cassava in Kera1a. The 1ndustry is reckoned to run at undercapacity and to be a much more minor producer than Tamil Nadu. A 11sting of reported starch plants --although not necessarily a complete listing-- and the1r estimated annual production gives a starch product10n f1gure oi approximately 57 thousand tons. An alternative unpublished est1mate for 1977/78 i9 110,608 t of starch (State Planning lloard, private communication). Tbe latter figure ",ouId imply a much larger industry than 16 commonly reckoned. Tbe f1nal entry in the accounting of cassava utilization in Kerala i9 root export to Tamil Nadu. Most reports on the starch industry in Tamil Nadu cite imports of cassava roots from Kerala. Tbe roota principally cO,me from Trichur district in the north. Estimates of these exports are fe",. lIone (1974) presents an estimate of 400-600 thousand tons and cites a figure that licenced exports of up to 400 thousand tons are permitted. Tbis 18 a remarkable volume considering tbat road transport i8 relatively acaree and expensive--transport costs add as mueh as 40% to root purchase priee in Kerala. A transport priee of 150 rupeest ",as cited (fieId notes, 1962), compared to a wholesale root price in Triehur of 519 rupees in 1981. Tbe higher eost of root production in Kerala together with the transport eost is bound to make cassava roots from Kerala competitive only outside the principal harvest season in Tamil Nadu. Moreover, easssva produetion in Trichur distríet ia one of tbe lowest in Kerala, producing 114 thousand tons in 1960/61. A more reasonable estimate ia probably in. the range of 50 to 75 thousand tons. A synthesis of these various consumpt10n estimates 18 presented in Tabla 12 for· the year 1977. Compar1ng tbe consuroption aggregate to the 1977 /78 production figure, tbat is after the production series has been radically revised downward, reveals that about a million tons sti1l rema in unaccounted for. Wastage in an economy such as Kerala with the small distances to roarket and the well developlng marketing services 1s probably small but may be assumed to be in the neigbborbood of 10 to 12%. At tbis point there i8 no more justification for revislng the consumption figure upward as for rev1sing the pro~uction figure downward. Assuming that the human consumption figure is underestimated and putting the remainder in tbat category would imply a per capita consumption level of_l03 ~ k~LYea:r_. Compared to the other sample estimates this is not unreasonable but eertainly suggests that earlier estimates of per eapita consumption froro food balance sheets were substantially overestimated, generally by more than 100%. Tamil Nadu: The market for cassava in Tamil Nadu as compared to Kerala, i8 dominated by demand for industrial uses as opposed to food uses. The starch and tapioca pearl industry centered in Salero District ia considered to be the major end user of cassava in Tamil Nadu. There are 611 starch factories in Tami! Nadu, 497 of which are located in Salero District and the other 114 of which are located in Dharampuri, South Arcot and Coimbatore districts (Salem Starch and Sago Manufacturers's Cooperative, private communication and Uthamalingam, 1980). Utilization of cassava roots would then fo1low from the operational characteristics of these plants. A sample of 30 starch and pearl factories were selected in Salem town and in out1ying rural areas. The operational structure is given in Table 13. There are 228 pearl factories aud 269 starch factories in Salem and assuming a distribution of 75% small-scale and 25% large-scale, leads to an average annual output per factory of 499 t. This annual average starch output thereby implies an annual production level of 248 thousand tons. in Salem District and an additional 57 thousand tons in the three adjacent distriets. Uthamalingam (1980) provides alternative astimates based on the quantity shipped by railway and that purehased by the Salem Sago and Starch Merchants Association (Table 14). These are only about one-third of the above estimates. The .rail shipments obviously do noto include the starch consumed 10ca1ly or that transported by road and therefore provides only a minimum estimate of production and an idea of variation of production from year to year. The estimate based on per factory output: implies root utilization of 992 thousand tons in Salem and 228 thousand tons in the adjacent distr1ets, assuming the relatively high conversion rate reported in Tamil Nadu of 4:1. Most reporta suggest that food usage of the cassava root is relat1vely minima1 1n Taroil Nadu. The 1973/74 National Sample Survey reports an average annual rural consumption of cereal substitutes of 4.1 kg/year for the whole state. lt 18 probable that this figure includes only cassava but it is not certain what percentage would be root and what would be processed cassava. Sinee the only reported consumption 1n Tamil Nadu is for rural areas, it is probable that this figure only ineludes root consumption. This would 1mply a total food consumption of 125 thousand tons. The recapitulation of the consumption, together w1th an assuroed 10% wastage, gives a total figure of 1,514 thousand tons, whieh compares favorably with the production estimate of 1,682 thousands tons in 1978/79 snd 1;591 thousand tons in 1979/80. A small change in the starch conversion rate could aecount for any difference. The production and consumption data would appear to be more or less consistent, at least since the 1977/78 crop year. 13 Other States: For the sake of completeness, Andhra Pradesh is the only other state with anywhere close to a significant production volume. Production in this state was 88.2 thousand tons in 1979/80 and 171.0 tbousand tons in 1980/81. This volume is comparable to about 10% of the production of Salem District. Cassava 1s a ra1nfed crop in Andhra Pradesh and is principally grown in East Godavari District. The cassava root 1e used exclusively in s small, cassava pearl industry located in the district. ~ummary: A consistent set of produetion snd utilization estimates for the crop year 1977/78 are presented in Table 15. ·The disparicy in the market structure between Kerala and Tamil Madu 1s apparent from the difference in the weight of the freah human eonsumpt1on and stareh markets in the two states. In India as a whole starch 1s a far larger consumption fotID of casaava than 1e apparent by only focusing on Kerala. Of the starch produetion a large part is in turn consumed as human food in the forro of tapioca pearl. Having some idea of the different magnitudes of each market, eaeh will now be analyzed in more detail to evaluate the potential for absorption of increased cassava production. Cassava for Direct Human Consumption: Cassava as a direct food souree achieves substancial weight in only the food economy of Kerala Seate. As might be expected in rural economies where population pressure on land is high, per capita food consumption levels are low. About 70% of average incomes are spent on food, with the principal component being rice, on which 30% of total income 18 spent (Table 16). In the rural areas over 6% of average income is apent on just cassava. In such economies .food consumption is directly dependent on in come levels and as can be seen in Table 17, food calorie distribution is symmetric to income distribution. Average daily caloric intake 1s just over 2000 calories. Using the relatively gross standard of 2100 calories as the minimum daily requirement, Table 17 shows as much as 35% of the population in rural areas and 50% in the urban areas falling below minimum requirements. Because of the work and activity patterns of the poor in rural areas, calorie shortages can be considered to be chronic. Cassava plays a key role in the caloríe nutrition of the population of Kerala. Cassava i9 at least as important (National Sample Survey, 28th Round) or more important (Kumar, 1979) than rice for the low-income strata in rural areaa. Rice 18, however, the preferred food and consumpt10n . increases markedly with ineome. However, at least for the 81% of the population in the rural areas cassava consumption shows a slight increasing trend across income strata (Tabla 18). Even though per capita consumption levels are high, as compared to Indonesia for example, the National Sample Survey vould indieate some limited capacity to increase cassava consumption in the rural areas with increaaea in income; although with evarything else equal, mose of that ineraase in income would go to increased rice consumption. Because of the limited incomes in Kerala, a low-cost-per-calorie food 8uch as cassava playa a principal role as a supplement to the h1gher cost rice. A principal issue 1s whether promoting technical change in casaava , production, and the resultant: lower priees, will lead to bridging the ,¡ calarie deficit. In the major cassava producing district of Trivandrum 1 cassawa prices tend to be substantially lower and rice prices higher than in ot:her districts. The survey of Kumar in Trivandrum suggests that cassa9a consumption levels are substantially higher and rice consumption slight1y lower than the average for Kerala (Table 19). However, for the po orar income strata total calorie consumption is substantially higher than for the sta te average for this stratum. In areas such as the survey area where average annual consumption reaches 172 kg, there is probably not much potential for further increases in cassava consumption but changing the rice-<=B.ssava price relationship in other parts of Kerala would, on the basia of this very limited comparison, lead to increases in cassava consu.ption and increased calorle consumption. Shah (undated) has argued that tlattempts to increaae the production of low cost, high calorie fooda, with a view to bridging the calorie gap, by themselves may prove inadequate tl because preferencss for food qualities other than juat calories bias consumption even in the low income groups to more !:ostly foods. Food consumption patterns across incorne groups as described above would indeed confirm that food quality i8 important but as well that for the poor, where price dlfferences are sufficiently large, cassava can constitute up to two thirds of total calorie 1ntake. The central government has in part incorporated the quality argument in ita system of publ1c food distribution. The foodgrain d1stribution syatell has played a major role in the food economy of Kerala. since 1964, when food shortages in India led to food zoning and curtailment oí private interstate trade. The system depends on a comprehenslve system of ration or ísir price shops, at which consumera are given quotas for foodgrains and prices are set well. below open market prices. However, consumption requirements are well above the ration quota and consumers must purchase their additional requirements from the open market. The availability of ration rice has a marked influence on rice and cassava consumption patterns. A study by Ceorge (1979) found that consUllption of ration rice waa relatively constant across income strata (Table 6), although this finding 18 based on household income. Kurnar (1979) found that ration rice consumption increased with incorne when expressed on a per capita basls. However, whereas the hlgher lncome strata were able to complement this allotment with rice froro open roarket purchases and at the highest incoms levels from own production, the lower income strata suppleroented the ration rice with very high levels of cassava e onsUllption , most of which was purchased (George, 1979). Nutrition of the poor thus depended principally on ration rice allotments and cassava purchases, as was also found by Kumar. Wheat is also available through the ration shops but Ceorge (1979) found that "rural households consumed only a small quantity of wheat. When their rice quota was exhausted, consumera preferred to purchase cassava from the open market than wheat froro the ration shops. Wheat purchases from the ration shops accounted for only about one-third of the total wheat allotment for the total sample and were the lowest in the low income household" (p. 33) • G1ven thecpref~~jl!n..c'L..f.2I.~rice,. a principal determina!,t of the demand for cassava w!ll be ration rice allotments. The first factor to consider l5 i8 whether ration rice consumption i9 influenced by demand factora. Two studies (George, 1979 and Kumar, 1979) conclude that ration rice consumption is not influenced by dernand factors but purely by supplies available; that is, all that is available would be consumed. As levy procurement of rice within Kerala dropped to negligibl" levels, the ration system in Kerala carne to rely almost completely on allotmenta frem the Central Pool of the Food Corporation of India (FC!). Moreover, these allotments now account for over half of rice supplies in Kerala (Table 20), and whereas such allotments should introduce a certain stability in rice supplies, they are in fact, the major cauae .of variability in rice availability in the atate. The author knows of no study which analyzes the determinants of atate allócation of ration rice by the FCI, but obviously there are other criteria than just maintenance of per cap ita consumption levels over time. There is little choice but that cassava viII continue to be a principal component of a food strategy in Kerala and in particular cassava can be used to provide a certain flexibility in the operation of the food ration system in the state. The dried chip market A peeled dry chip, similar to gaplek in Indonesia, is produced in Kerala. The market principally provides an alternative outlet for cassava during the principal harvest period from December to April, which coincides with the dry season. The chips are principally produced and assembled in the northern districts, with Calicut, Trichur and Chsnganachery being the principal assembly centers. Data on the markets for cassava chips are virtually non-existent. What can be said is that this market is not as large nor as well-integrated as the gaplek market in Indonesia. Most consumers in Kersla have relatively direct access to fresh roots and most field observations would suggest a consumer preference for fresh over dried cassava. Tbe one, and relatively dated, source on processed cassava consumption suggests very limitad consumption levels, with an average annual per capita consumption of 9.5 kg of dried producto Indications are that the dried chip market for human consumption will remain very limited. As is apparent in Indonesia a well functioning dried chip market provides an element of price stability to the fresh root market, especially where the majar portion of planting and harvesting takes place at relatively restricted times of year. The chip market acts as a storage mechanism for cassava during the low season and provides a price floor during the peak harvest periodo In Kerala th. . other major market for cassava chips ia for processing into starch and glucose, especially glucose. Fresh roots produce a higher quality starch (Meuser, et • al. , 1978) but chips are used in the starch industry in Kerala because they are cheaper on a starch basis and help to maintain oparation outside the peak harvest season. However, if roots were.available at the price and quantity desired, the starch industry· would operate exclusiv ely on roots. This particular outlet then does not provide a certain demand on which to develop an expansive dried chip market. ~----~ ~- The other principal~option in developing a dried cassava market ie the export market. India exported limited quantities of cassava chips to Europe between 1957 and 1964. The largest export level reached in this period was 72 thousand tons in the 1958-59 crop year. Exports virtually ceased until 1977 when exports to the EEC were resumed. This reopenlng of export shlpments was brought on by a substantlal price fall in dried cassava ln 1977, together wlth a very large margln between domestic wholesale prices and import prices in Europe (Table 21). Exporta have continued at relatively moderate levels since 1977 (Table 22). However, levels of 20 to 30 thousand tons result in high cost shipping and allows few of the benefits of an export price floor to develop, especially the incentives for investment" in more efficient marketing and processing capacity. At thls stage Kerala does not have the production base to develop an effective export market and roeet domestic requirements, nor wilI India ever be in the position of belng a large exporter of cassava products. However, a signiflcant increase in yield levels could lead to further development of this nascent industry, which would in turn provide incentives for further market integration, the setting of a stable floor price, and in turn lower and more atable prices for fresh cassava for food. The starch market The market for cassava for starch production 1s divided between a ful1y integrated industry based on smal1-to4!ledium scale plants in Tamil Nadu and a relatively fragmented starch industry 1n Kerals consisting of two large-scale plants, 3 medium-scale and 50 small-scale plants. The principal constraint on expansion of this industry ia supply of raw material to run the plants. The industry in Kerala probably opera tes at no more than 50% capacity. Factories here must compete with cassava for the fresh market and during at least part of the year must offer a lower price for cassava roots than pertains on thefresh market, in ord,er to remain competitive with production 1n Tamil Nadu. Thus, in 1981 a major starch factory in Kerala paid 260 rupees/t for roots, which compared to farro level prices in Tamil Nadu of between 280 to 360 rupees/t and farro gate prlces for the fresh market in Kerala of 400 rupees/t (field observations, 1982). The farmer prlce would only cover variable production costs for the farmer and represents a price at whlch farmers would sell roots of low quality or where identification of other market outlets was a constraint. Further development of the starch industry in Kerala requires that prices in the fresh food and starch markets be brought closer in lineo Unlike the chip export market, the cassava root market for starch 1s already probably large enough to set an effective price floor, should that ever be nece5sary. As it i5, declining production trends and ris1ng cassava prices implies that the starch industry in Kerala will remain moribundo The cassava root market for starch in Tamil Nadu functions as a single, integrated market. The starch industry here, nevertheless, operetes at between 45 to 60% capacity. Competition in Tamil Nadu does not come on the demand side with alternatives but rather from the supply side, where cassava must compete with a substantial "number of crop alternatives for irrigated land. Root prices toO the farmer are in turn determined principally by the sale price of starch, since roots make up approximately 80% of the total cost of starch or sago production (Table 23). 17 !he cos~ and ope~ating structure of the starch and sago industry, show in Table 23, suggasts a relatively competitiva, small-to-medium scale industry where annual returns on fixed investmen~ of from 17 to 31% provide a normal return on investment, considering the general capital scarcity that characterizes the ludian economy. Wi~h further incraaaes in farro production capacity, there is little doubt that a dropping cassava price would motivate further investment in processing capacity. !he end market for sago and starch i8 not well documented. The market for both apparently is centered in the·more northern states. ·The end use of starch is principally in the textile industry, especially Bombay. Rere cassava starch competes with maize starch, which ie preferred over cassava starch, apparently because of the higher viscosity, and selle at a premium to· cassava starch. The cassava pearl or sago, on the other hand, ie used strictly in food uses and the larges~ marke~ appears to be Bengal, particularly Calcuta. Uses range from a festival food to a filler for rice. Ex-factory prices of sago in 1978-79 of 1.55 rupees/kg compare favorably to rice priees of 2.2 rupees/kg. The potential eonsumption of starch and sago in India i8 not known but traders knowledgeable about the industry suggest that demand is no constraint st forseeable production levels. Pricing aud market efficiency: Prica determination and market allocation betwaen competing uses are governed, at least in Kerala, essentially by factors which 1nfluence the demand for fresh cassava for human consumption. The starch, chip, and export markets essentially serve to set a price floor and absorb any snrpluses at this price. Becanse of the very marked seasonality of harvest such surpluses oecur seasonally during the year. as well as per10dically from year to year. Hecanse the fresh human consumption market makes up such a large part of total production - compared, for example, to Java - any changes in either cassava supply or fresh· root demand will create substantial instability in snppl1es going to alternatives markets. Due to this factor and the very severe constraint On expansion in production area, the development of these alternative markets has been very fragmented. Although cassava consumption and' prices are obviously influenced by rice availability and priees, there are no studies which measure the degree of this influence. Planning and investment in rice production, cassava ptopuction, and ration rice distribution in Kerala are critically dependent on such a study. Priee series provides the only data which shed light On the interaction between the rice and cassava markets and here several inexplicable trends beeome apparent. One special difficulty in analyzing price series 18 separating out the effeets oí inflation in th" general price level. Sinee the consumer budget i8 weighted so heavily by food purchases, the consumer price index will refleet changes in food prie"s more than other products. These tend to be somewhat volatile anyway but in India upto 1977 food zoning beavily restricted interstate trade in food grains. Food price levels thus varied by state and· using the consumer price index for India as a whole to deflate priees in any particular state will probably not be reflective of price inflation in that particular state. For this reason che consumer price index in Trivandrum was used to deflate all prices in Kerala. During the decade of the 1970's real, retail rice price rose till 1974 and than fell dramatically (Table 24). Retail cassava prices, on the ocher band, remalned relatively constant tbrough the period, resulting in rice becoming relatively cheaper to cassava. While tbe marketing margin for fresh cassava in Kerala ie proportionally low compared to margina in other countries, tbe margin has masked much higber variability in cassava prices at the farro-level (Table 25). At the farro-Ievel comparable, thougb not as marked, trenda to those that have occurred in the retail rice market have occurred. In particular, there is a falling real cassava price at the farm-level at a time when production was declinlng rapidly. Ibis would support a marked influence of rice prices and availabilities on cassava prices. The dominant issue then i9 what has been happening witb rice availabilities? Througb the decade of the 1970's rice production in Kerala was relatively stable (Table 20). The component of variability in rice supplies in Kerala was tbe availability of ration rice. What is inexplicable with tbe available data i8 the low rice prices in 1978 and 1979. Since food zoning and restrictions on interstate trade of food grains was eliminated in 1977, it is possible that there have been flows of rice iuto Kerala from other etates brought by private traders and sold on tbe open market. However, even the limited evidence on open market availabilities suggest that sucb supplies were not mueh ehanged in the years 1978 and 1979 and that eliminating food zoning has had no impact on rice supplies in Kerala. Rice priees in Kerala have been traditionally higher than in the other lndian states, and while the liberalization of trade flaws should bring prices more in line, the mechanism to do this has to be increased availabilities. Ibus, while it is nat elear why, deelining rice prices are putting a damper on cassava priees, that wauld otherwise be rising in response to declining production. Ibis has allowed cassava pricea to remain competitiva in the world market. To the extent that increased rice supplies can be assured, this would have the greatest impact on nutrition in Kerala. What 18 clear, however, i8 that there are no such assurances. Maintaining low priced cassava for the human consumption market provides a critical element of stability in foad supplies. ~~at is needad, however, is better integration with alternative markets which can handle surpluses when rice supplies are adequate. What thls requires is a larger production base and this can only be achieved with further increases in yields. Conclusions Cassava serves a major, if somewhat distinct, role in the agricultural economíes of Kerala and western Tamil Nadu. In Kerala internal rice production is stagnant and there 16 an increasing portian of the upland area being planted to higher value tree crops. Foad supplies thus rely critically on rice allocations from the central pool and more recently apparent, privately-traded inflows from outside the state • Hawever , in maintaining or improving the food intake and nutrition of the low incorne strata, the options are increases in rice rationing off-take or more plentiful and cheaper cassava. Since an increase in the poor's rice ration allotment implies an increase for everyone, cheaper-·cassava.·would target directly on the poor and would not involve subsidies from the public treasury. The design of a faod and nutrition policy in Kerala is heavily 19 dependent on the prognosis for rice production in India as a whole given that food zoning is a policy of the pasto Nor should policy makers appear insensitive by suggesting that the poor should just eat cassava. Pure pragmatism suggests that the calorie intake of the poor ia critically low and that cassava can be as cheap a means as any of increasing calorie intake. In Tamil Nadu, on the other hand, a potential growth industry, much like the caae of Indonesia, exists in the starch and tapioca pearl market. The industry is constrained by lack of raw material for processing and for farmers there is no restrictions on 'finding market outlets tor "their production. Prices are in most respecta relatively atable and any increases in yields will directly improve farmer incomes. The issue, then, is how much higher farm level yields can be raised in these two states over the relatively high level which farmers already achieve. Such increases will almost certainly depend on higher yielding varieties. The research of the CTCRI sugge5ts that there i5 acope for doing this in Kerala. An iseue which CTCRI is very conscious of is that the quality characteristics of these improved varieties shall have to rema in high, since cassava is essentially consumed in a fresh formo In Tamil Nadn, on the other band, there are no such restrict10ns, other than that the yield gap to be exploited there appears to be much smaller. India 18 probably the only of the major cassava producing countries in Asia where the only frontier for cassava to exploit is the yield frontier. lb le India: Trends in Area, Production and Yield for the Country and the Major Producing States, 1964-1981. Inola Kera la Tami 1 fiadu ~op Year Area Production Yield Area Produdlon Yield Area Productioñ Vie le ¡OOOha) (OOO tl ¡t/ha) ¡OOOha) (000 t) ít/ha! (OOO ha) ¡OOO tl ¡t /h¡ ~64-65 240.0 3,033.0 12.6 209.0 2,763.0 13.2 25.0 243.0 9.7 165-66 271.0 3,467.0 12.8 230.0 3,095.0 13.5 35.0 339.0 9.6 166-67 290.0 3,817.0 13.2 245.0 3.410.0 13.9 39.0 377.0 9.8 167-68 335.0 4,520.0 13.5 298.0 4,198.0 14.1 30.0 285.0 9.7 168-69 359.0 4,636.0 12.9 298.0 4,081.0 13.7 55.0 527.0 9.6 169-70 353.0 5,214.0 14.8 296.0 4,666.0 15.8 44.0 513.0 11.8 170-71 353.0 5,216.0 14.9 294.0 4,617.0 15.7 47.0 567.0 12.1 171-72 353.7 6,025.9 17.0 303.3 5,429.3 17.9 42.6 545.0 12.8 l72-73 363.2 6,317.4 17.5 304.8 5,629.4 18.7 50.0 629.5 12.6 173-74 368.2 6,420.9 17.1 306.4 5,659.5 18.5 51.7 681.6 13.2 174-75 387.6 6,325.9 16.3 317.9 5,625.1 17.7 52.7 564.9 10.7 175-76 392.0 6,638.3 16.9 326.9 5,390.2 16.5 50.1 1,115.8 22.3 176-77 385.8 6,375.0 16.5 323.3 5,125.5 15.9 48.0 1,128.2 23.5 m-78 358.3 5,688.3 15.9 289.7 4,188.6 14.5 52.8 1,310.3 .24.8 178-79 361.5 6,050.1 16.7 289.9 4,226.3 14.6 54.0 1,682.0 31. 2 179-80 365.3 5,952.2 16.3 290.3 4,223.6 14.5 58.1 1,591.4 27.4 180-81 320.8 5,868.1 18.3 243.3 4,097.8 16.8 53.3 1,539.3 28.9 181-82 310.2 5.267.4 17.9 241.8 4,073.0 16.8 42.3 1,324.8 31.3 ¡urce: "Bulletín on COJllllercíal Crop Statistícs" and "Agricultural Situation in India", Ministry of Agriculture. Table India: Annual Raíl Shipments of Starch and Pearl from Salem and Purchases by.the Salem Sago and Starch Merchant's Assocíatíon, 1970-1977. Raíl Shi prrents ~sociation Purchases Year Pearl Starch Pearl Starch (t) (t) (t) (t) 1970 52,589 39,553 N.A. N.A. 1971 55,171 28,987 N.A. N.A. 1972 41,133 41,488 N.A. N.A. 1973 22.249 41,102 N.A. N.A. 1974 18,871 42,822 N.A. N.A. 1975 44,774 45,827 N.A. N.A. 1976 36,394 30,656 38,605 29,583 1977 55,702 35,081 55,095 26,596 Source: Uthamalingam, 1980. Table India: Average Prices of Fresh Cassava Roots at the Fann, Wholesale and Retail Level, 1970-80. Farm-leve 1 11 Wholesale Retail Year Nominal Real - Nominal Real ]j Nominal Real ]j (Rupee/t) (Rupee/t) (Rupee/t) (Rupee Al Rupee/t) (Rupee/t) 1970 N.A. N.A. 209 386 300 550 1971 214 391 222 407 310 570 1972 235 406 240 415 320 550 1973 309 446 311 449 400 580 1974 384 423 397 437 510 560 1975 400 400 391 391 540 540 1976 398 449 391 441 550 620 1977 325 376 323 373 500 580 1978 316 353 326 363 490 590 1979 398 411 410 424 590 610 1980 N.A. N.A. 443 N.A. N.A. N.A. ]j Deflated by'consumer price index in Trivandrum, 1975 : 100 Source : Government of Kerala, "Statistics for Planning", Directorate of Economics and Statistics. Trivandrum, various years. Table India: Area under Principal Tree Crops in Kerala, 1970-80 Rubber í' Less than Crop Year Coconut Black Peper 2 has Total Cashewnut (000 ha) (000 tia) (000 ha) (000 ha) (000 ha) 1970-71 719.1 N.A. 68.5 20.31 N.A. 1971-72 730.3 116.3 71. 7 208.8 N.A. 1972-73 745.4 116.3 74.1 213.1 N.A. 1973-74 744.8 118.2 77 .1 217.5 103.2 1974-75 748.2 118.4 79.4 221.3 104.9 1975-76 692.9 108.2 81.9 224.4 109.1 1976-77 695.0 110.6 85.5 230.6 113.3 1977-78 673.5 108.3 88.4 233.4 127.0 1978-79 660.6 108.3 91.3 235.9 N.A. 1979-80 663.8 N.A. N.A. N.A. N.A. Source: Government of India, "Bulletin of Commercial Crop Statistics", Directorate of Economics and Statistics, Ministry of Agriculture, various years. Table India; Wholesale Prices of Cassava Chips at Kozhikode and Comparison with European Import Prices, 1970-79. Month 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 -------------------- Ru~ees/lOO kg ------------------------ Jan. 3D 40 35 45 56 55 55 60 55 80 Feb. 30 41 35 45 75 55 53 57 48 88 Mar. 32 45 38 52 70 55 52 55 47 98 April 36 44 47 62 75 55 80 62 55 110 May 37 40 47 70 85 55 80 60 58 N.A. June 40 48 50 62 80 57 88 55 65 N.A. July 40 48 60 62 73 55 100 50 70 N.A. Aug. 45 50 50 78 78 52 103 50 65 N.A. Sept. 50 50 N.A. 75 80 55 100 40 65 105 Oct. 50 52 N.A. 82 85 65 105 4D 65 120 Nov. 58 47 58 80 70 85 110 35 85 120 Dec. 43 45 52 68 70 85 80 50 90 110 Average 41 46 43 65 75 61 94 51 64 104 ~ $/t Equiv- alent 55 61 57 84 93 73 105 58 78 128 cif Rotterdam N.A. 66 74 87 149 124 118 108 101 164 ($/t) Source: Government of India, IIBulletin on Commercial Crop Statistics," Directorate of Economics and Statistics. Ministry, of Agricul- ture. various years. Table India: Characteristics of Starch and Pearl Factories in Salem District Tamil Nadu, 1978/79 11 Starch Pearl Sman Large Smali Large Root Input (t) 1,629.6 2,416.1 1,635.3 3,287.3 Starch Output (t) 431.6 652.8 411.8 822.0 Conversion Rate (%) 26.5 27.2 25.2 25.0 () '!- Average Operati n Peri od .135 144 175 184 A (days) 11 In Salem District there are 269 starch factories and 228 tapioca pearl factories Source: Uthamalingam, 1980 , \. Table India: Monthly Rural Consumption of Cassava and Rice by 1n come St ra ta Kumar Survey National Sample Surv! Open Market Income Strata Cassava Ration Rice Rice Total Rice Cassava Rice (Rupees/capita) ( kg/ capi ta) (kg/capita) ( kg/capita) (kg/capita) (kg/capita) (kg/capité 0-15 19.95 1.60 .69 2.29 6.27 1.88 15-24 17 .68 2.29 1.46 3.75 6.47 3.83 25-34 16.13 2.51 2.04 4.55 6.70 5.03 35-49 16.09 2.67 2.06 4.73 7.18 6.17 50-74 14.35 3.46 1.64 5.10 7.20 8.43 Greater then 75 1/ 4.19 3.55 2.35 5.90 7.16 12.08 Average 14.13 2.89 1.98 4.87 6.99 7.23 1/ For Kumar sample there are two observations on1y. Sources: Kumar, 1979; Government of India, 1973/74. Table India: Estimated Capacity and Output of Starch Plants in Kera 1a Production Plant Capacity Estimate (t of starch/day) (tjyear) Lekshmi (Quilon) 80 t 15,125 Tapioca Products (Trichur) 100 t 17,500 Hode Chemical Sago (Quilon) lOt 1,500 Pemba Starch (Quilon) lOt 1,500 50 small-scale plants 3 t 21,500 Total 57,125 Source: Report of the Sub-Committee of the Tapioca Market Expansion Board, Oepartment of Food, Government of Kerala, Trivandrum, 1972. Tab 1e . India: Monthly Per Capita Consumption of Cassava and Rice by Income Strata, 1973/74. Cassava Rice Income Strata Rural Urban Rural Urban (Rupees/capi ta) (kg/ ca pita) ( kg/ ca pi tal (kg/capita) (kg/capita) 0-13 5.04 1.96 13-15 8.33 0.20 1. 75 3.60 15-18 4.63 12.50 3.42 .1.67 18-21 7.60 3.23 3.18 2.95 21-24 6.49 3.05 4.34 4.23 24-28 5.14 5.59 4.98 4.06 28-34 7.49 3.06 5.06 5.60 34-43 6.48 4.10 6.05 5.59 43-55 7.79 4.04 7.26 7.81 55-75 7.20 4.73 8.43 7.32 75-100 6.86 3.24 10.44 9.90 100-150 7.35 2.02 11.88 8.81 150-200 11.16 1.65 15.37 9.63 Greater than 200 5.43 1.50 18.67 10.50 Average 6.99 3.64 7.33 7.23 Source: Government of India, "The National Sample Survey", 28th Round, National Sample Survey Organization, 1973/74. Table 3.30. India: Yield Oi5tribution from Crap Cutting Survey, Tamil Nadu, 1979-80 (287 farms) Yield Strata Percentage ( t/ha) Distribution 0- 7.5 13 7.5-15.0 14 15.0-2.2.5 16 22..5-30.0 2.5 30.0-37.5 16 37.5-45.0 8 45.0-52.5 5 52..5-60.0 2. 60.0-75.0 1 75.0-90.0 0.3 Average Yield - 24.5 t/ha Standard Deviation = 14.1 t/ha Maximum Yield = 84.2 t/ha Irrigated Yield = 27.4 Unirrigated Yield - 15.6 SOURCE: Unplublished results of crop cutting survey. Tamí] Nadu. Table 3.26. India: Consumption of Rice and Cassava by Income Strata and by Source of Supply, Rural Kerala, 1977 (kg/household/week) Annual Rice Cassava Household Total Own Open Total Own Open Income Consumption Ration Production Market Consumption Production Market (Rupees) (kg) (kg) (kg) (kg) (kg) (kg) (kg) Less than 600 8./¡0 5.65 2.75 12.90 O. I¡Q 12.50 601-1200 9.43 6.39 3.0/¡ 11.31 2.96 8.35 1201-2400 13.47 7.70 1.77 4.00 15.46 /¡. 13 11. 33 2401-3600 13.89 6.67 1. 11 6.11 12.66 /¡.33 8.33 3601-4800 12.00 4.90 2.00 5.10 6.70 4.50 2.20 More than 4800 13./¡2 5.14 5.71 2.57 3.29 3.29 SOURCE: George, 1979. Table 3.24Jndia: Calorie Consumptlon by Ineome Strata in Kerala, 1971-72 Per Capl ta Rura 1 Urban Monthly % Oistribution Per Capl ta % DIstrlbution Per Capl ta Expenditure of Housenolds Calorle of Housenolds Ca lorie (Rupees) Consumpt I on Cons umpt I on 0-15 3.1 893 3.3 953 15-21 5.9 1229 7.6 1079 21-24 4.6 1716 5.7 1575 24-28 8.5 1466 6.9 1490 28-34 13.0 1900 12.1 1787 34-43 9.5 2320 14.5 1989 43-55 15.6 2603 14.2 2289 55-75 18.6 2900 10.9 2700 75-100 9.2 3614 7.3 3060 Hore tnan 100 12.3 4293 17.6 3907 Average 100.0 2023 100.0 2103 Souree: Statisties for Plannlng 1980, Government of Kerala. Table 3.21.1ndia: Percentage Distrlbution of Farms by Size in Kerala, 1970-71. Size of Distribution Holding of Holding (ha) (%) Below 0.04 18.7 0.04 - 0.25 37.2 0.25 - 0.50 15.6 0.50 - 1.00 13.3 1.00 - 2.00 9.7 2.00 - 3.00 3.2 _ 3.00 - 4.00 1.4 More than 4.00 0.9 Total 100.0 SOURCE: Statistics for Planning 1980, Government of Kerala, 1980. Table India: Annual Costs of Production of Starch and Tapioca Pearl in Tamíl Nadu, 1978-79. Starch Ta~ioca Pearl Small Large Sma 1 large Cost Item Factory Factory Factory Factory (Rupees) (Rupees ) (Rupees ) (Rupees) Variable Costs Cassava Roots 465,611 690,303 497,227 989,237 Temporary Labor 25.294 39,,236 43.826 78,011 Fue1 5,060 11,492 El ectri city 4,292 7,624 4,687 9,240 Cocanut Oil 2,955 4,864 Gunny Bags 23,891 36,035 25,602 50,436 . lnterest on Working Capital 23,039 36,605 33,333 69,067 Total Variable Costs 542.121 809,803 612,689 1,212,346 Fixed Costs Pennanent labor 9.091 11,277 7,231 12.908 Office Overhead 2,171 4,181 2,040 3,825 Oepreciation Buildings 2,174 2,810 1,103 2,695 Machinery 6,832 10,285 5,003 10,611 Interest on Fixed Capital 15,937 22,910 13,295 19,618 Taxes 3,250 4,000 2.756 3,786 Total Fixed Costs 39,455 55,523 32,034 53,449 Total Costs 581,583 865,326 644,723 1,265,195 Annua 1 Output (tons) 431.6 652.8 411.8 822.0 Total Cost per Ton 1347 1326 1566 1540 Output Pri ~e' per Ton 1333 1333 1556 1555 Va 1u e of By Products pe r Ton 85 93 72 72 Source: Ulthamalingam, 1980 Table India: Oifferent Estimates of Per Capita Consumption of Fresh Cassava in Kerala. Annuaí Sample Samp 1e Per Capi' So urce Size Structure Peri od Consumpt' Kumar 43 househo lds Trivandrum District Feb-Sept.1974 171.9 Rural Only Bottom 50% of Income Strata George 100 househo lds Two Vi1lages Nov. 1977 114.7 Rural Only National Sample Survey 890 nouseholds Complete State Oct. 1973-June 78.3 Rural and Urban 1974 Tapioca Market unknown All but One Oistrict 1971 56.5 Expansion Board Rura 1 and Urban U.N.Dept. of Food Balance Tables 1961/62-1970/71 208.4 Economic and Social Affairs Govt. of Kerala Food Balance Tables 1974 276 Sources: Kumar, 1979; George, 1979; Government of India, 1973/74; Government of Kerala, 1972; U.N. Oepartment of Economic and Social Affairs, 1975; Government of Kerala, 1977. Table India: Labor Use in Cassava Production Systems, Tamíl Nadu, 1978-79. Irrigated Rainfed Activity Men Women Men Women (days/ha ) (days/ha) (days/ha) (days/ha) Preparatory Cultivation 27.2 11.9 Seeds and Sowing 15.2 3.6 6.5 5.3 Manuring 5.4 7.1 Irrigation 25.3 Weeding 96.7 91.9 Harvesting 30.6 28.1 Mi s ce 11 aneous "1.8 1.9 Total 103.7 161.6 53.5 85.0 Source: Uthamalingam, 1980. Table India: Availability of Rice in Three Major Markets in Kerala, 1970-81. Year Jan-Mar Apr-June July-Sept. Oct-Dec. Total (000 t) (000 t) . (000 t) (000 t) (000 t) 1970 21.0 10.7 5.5 4.4 41.3 1971 7.2 12.1 9.4 11.3 40.0 1972 25.7 25.7 15.3 15.3 82.0 1973 11.2 9.8 8.5 12.2 41.7 1974 8.6 9.6 8.4 4.7 31. 3 1975 4.2 8.3 11.3 4.5 28.3 1976 4.3 12.4 7.8 " 10.9 35.4 1977 12.6 12.5 11.7 9.7 46.5 1978 12.0 13.9 8.7 11.2 45.8 1979 8.1 10.6 5.5 7.1 31.3 1980 8.0 5.1 5.0 13.1 31.2 1981 10.2 8.6 3.3 24.9 47.0 Source: Government of India, "Bulletin on Food Statistics", Directorate of Economics and Stati~tics. Ministry of Agriculture, various years. Table India: Growth in Area Planted to Cassava in Kera la, 1920-1980 . Area Crop Yeár (000 ha) 1920-21 164 1925-26 170 1930-31 194 1934-36 175 1940-41 183 1944-45 197 1952-53 105 1955-56 222 1960-61 245 1965-66 260 1970-71 294 1975-76 327 1980-81 243 Source: Panikar et,al •• 1977 and Government of Kerala, "Statistics for Planning", Directorate of Economics and Statistics, Trivan­ drum, variou$ years. Table India: Cost of Production of Cassava in Tamil Nadu and Kerala, 1978-79. Tamil Nadu Kera la Cost Item Irrigated Rainfed Rainfed (Rupee/ha) ( Rupee/ha) (Rupee/ha) Variable Costs Preparatory Cultivation 273.0 180.4 466.6 Seeds and Sowing 220.5 222.0 221.1 Manures and Manuring 1,101.6 529.2 687.6 Irrigation 300.1 79.8 Weeding 477.6 228.2 349.5 Plant Protection 17 .0 Harvesting 237.7 177.5 200.6 Interest on Working Capital 274.1 140.4 212.3 Total Variable Cost 2,884.7 1,477.7 2,234.5 Fixed Costs Rental Value of Land 1,776.4 989.7 Depreciation 210.7 147.8 Interest on Fixed Capital 387.5 228.4 Total Fixed Capital 2,374.6 1,365.9 1,880.0 Total Costs 5,259.3 2,843.6 4,114.5 Yield (t/ha) 22.96 10.74 13.63 Variable Cost per Ton 123.9 137.6 163.9 Total Cost per Ton 229.7 . 2, 65.2 301.9 . Source: Uthamalingam, 1980; Hone, 1973. Table . India: Average Consumer Expenditure Pattern, Kerala, 1973-74 Rural [Jroan Amount Percent Aíñount Percent Item (Ru~eesl (%) (Ru(1eeS l (%l Cereals 18.14 32.8 18.10 26.3 Rice 17.70 32.0 17 .26 25.0 Cassava 3.53 6.4 1.67 2.4 Grams and Pulses 0.72 1.3 1.21 1.8 Vegetable Oil 1.12 2.0 1. 72 2.5 Milk and Dairy Products 1.82 3.3 3.93 5.7 Meat, Fish, Eggs 2.52 4.6 3.42 5.0 Other Food Items 11.75 21.2 16.69 24.2 Total Food 39.60 71.5 46.74 67.8 Fue 1 and li ght 2.97 5.4 "3.60 5.2 Clothing 2.63 4.8 2.55 3.7 Rent 0.10 0.2 1.26 1.8 Other Non-Food 10.05 18.2 14.78 21.4 Total Non-Food 15.75 28.5 22.19 32.2 Total 55.35 100.0 68.93 100.0 Source: Government of India, the National Sample Survey, 28th Round, 1973/74. \> ~ Table India: Produetion and Utilization of Cassava Roots %y State. 1977 /78. Domestic Utilization Human Consum~tion Animal State Produetion Export Fresh . Dri ed Stareh Feed Waste (000 t) (000 t) (000 t) (000 t) (000 t) (000 t) (000 t) Kerala 4189 22 2437 619 499 503 Tamil Nadu 1310 126 1162 ]j 131 Andra Pradesh 137 123 14 Other 52 47 5 India 5688 22 2610 619 1784 653 11 Ineludes 109 thousand tons of roots and chips imported from Kerala. Source: CIAT estimates Table India: Estimates of Production and Utilization of Cassava i~ Kerala, 1977/78 Fresh Root Es ti mate Conversion Estimate ____~ U~se~a~g~e ______________( t~) ________~ R~a~te~ ______~ (t~) _____ Human Consumption-Fresh 1,854,850 1 1.0 1,854,850 Human Consumption-Dried • 225,045 2 2.75 618,875 Starch 110,808 3 4.5 498,636 International Export-Chips 7,950 4 2.75 21,860 Inters tate Export-Chi ps 12,700 5 2.75 34,925 Interstate Export-Roots 75,000 6 1.0 75,000 Waste 502,630 1.0 502,630 Total Utilization 3,606,776 Production 4,188,600 Sources: 1 National Sample Survey, 1973/74; 2 Tapioca Market Expansion Board; 3 Kerala State Planning Board; 4 Renshaw, 1983; 5 Govern­ ment of Kerala, "Statistics for Planning"; 6 Estimate. Table India: Gassava Root Yield of Different Varieties in a Fertl1izer Trial Combinations [K Zhaof N ánd K20¡ Varieties 50:50 ·50: 100 50:150 i5:75 75:150 75:2~5 100:100 100:150 100-:200 -100 :2S-Ú- Mean H-165 22.67 23.01 22.88 24.24 22.84 26.47 28.30 25.08 23.87 27.93 24.73 H-2304 24.07 25.99 25.27 27.84 30.42 28.64 32.16 32.96 32.43 31.41 29.12 H-1687 19.29 19.04 21.47 19.62 20.13 22.96 26.05 26.39 25.31 25.02 22.53 M-4 15.18 14.76 15.66 16.95 16.10 15.83 18.62 18.66 17.48 18.62 17.79 Mean 20.30 20.70 21.32 22.16 22.16 22.37 23.47 26.28 24.77 25.74 Source: Central Tuber Crops Research Institute, Annual Report 1978-79, •Tr ivandrum. Table lndi a: lmports by the EEC of Cassava Chips. 1975-1980. Year Quantity (tons) 1975 O 1976 O 1977 7.949 1978 37,182 1979 26,799 1980 11,915 . \ Source: Renshaw, 1983 Table Indi a: Rice Production, Ration Rice Take-off, and Rice Availabilities in Kerala, 1971-1980. Rice 1 Rati on Card Total Production ..J Take-off Supp 1i es Year (000 t) (000 t) (000 t) 1971 857 844 1701 1972 892 874 1766 1973 908 764 1672 1974 830 786 1616 1975 814 539 1353 1976 879 937 1816 1977 828 1380 2208 1978 854 872 1726 ? " 1979 048 570 1418 ~ 1980 N.A. 812 N.A. 11 Rice production is on a milled basis by crop year. Source: Government of Kerala, "Statistics for Planning", and Government of India, "BúlJetin on Food Statistics." Table. India: Percent of Farm Production Commercialized in Various Districts of Kerala State, 1971 Percent District Commercialized Trivandrum 46.8 Quilon 32.2 Alleppey 33.9 Kottayam 28.5 Ernakulum 16.9 Tri chur 53.4 Palghat 77 .6 Malappuram 42.6 Kozhikode 38.2 Cannonore 23.0 Kera la 39.3 Source: Tapioca Market Expansíon Board, 1972 Table India: Constant1 Retail Prices of Rice and Cassava in Kerala. 1970-1979. Year Rice Cassava Rice/ Open Market/ (Rupee/kg) (Rupee/kg) Cassava Ration Rice 1970 2.87 .55 5.2 1.5 1971 2.78 .57 4.9 1.4 1972 3.04 .55 5.5 1.6 1973 3.47 .58 6.0 1.8 1974 3.84 .56 6.8 2.6 1975 3.53 .54 6.5 2.7 1976 3.02 • 62 4.9 N.A . , 1977 2.73 • 58 4.7 N.A . 1978 2.43 .55 4.4 N.A. 1979 2.33 .61 3.8 N.A. 1 Prices deflated by consumer price index in Trivandrum. 1975 = 100 Source: Government of Kerala. 1980; George. 1979. TRENOS ANO OISTRIBUTION OF CHINESE CASSAVA PROOUCTION ANO USE 1820 - 1984 Production trends and distribution No official national data seríes for cassava in the Peop1es Repub1íc have been published by Chinese authorities. It is possible to obtain estimated series from the Food and Agricultural Organization of the United Nations. 1 Such series are based on assumed annual increments in harvested area for most years and somewhat less regular, but a similar monotonical1y non-decreasing set of estimates for production. Yields appear to be derived from the rough area and production estimates by calcu1ation. The only figure among these which appears to have come from a Chinese source is the 3 mil1ion ton production figure círca 1980, provided unofficially as an undated estímate to the 1982 CIAT delegation by one of the agricultural science institutes visited in Guangdong. Earlier work 2 has concluded that the entire FAO series for root and tuber crops bears little relation to the aggregate series published sinee 1979 by Chinese statistical authorities. 3 It,is now a1so clear that the FAO l e .g. FAD, "Supply Utilization Tapes, 1984," Rome 1985; FAO, "Standardized Commodity 8a1ance Tape, 1984," Rome, 1985; and FAD, "Production Yearbook Tape, 1984," Rome, 1985. 2Bruce Stone, "An Examination of Economie Data on Cassava Production, Utllization and Trade,· a paper preparad tor the lnternational Center for Tropical Agriculture (CIAT), lnternational Food Policy Research lnstitute, Washington, D.C., August 1983. 3e . 9 . He Kang et al, Zhongguo Nongyebu [Ministry of Agriculture of China], (eds.) Zhongguo Nongye Nianjian 1980 [Agricultural Yearbook of China 1980] (Beijing: Nongye Chubanshe [Agricultural Publishíng House], 1980) and Zhongguo Guojia Tongjiju [State Statistieal Bureau], Zhongguo Tongji Nianjian - 1983 [Statistical Yearbook of China - 1983J (Beijing: Tongji Chubanshe [Statistical PublishinA House 1983. - 2 - series for cassava, per conflict with officially published series for one of the two principal growing regions and with scattered national estimates for individual years found elsewhere in Chinese publications. Since 1984, the FAO has taken account of sorne of the recent information in formulating current root and tuber erop estimates for publication in FAO Production Yearbooks. But much recent information has not been reflected in FAO series and additional work is required to obtain a reliable impressien of long- term trends fer individual crops, including cassava. According to Chinese sources, 4 cassava had be en introduced ioto China from South America via "nanyang" [the "South Seas" or Pacifie Ocean) by 1820, although it is not elear whether it entered Guangdong Provioce direetly from the West or whether it was introduced indirectly fol10wing regional cultivation in Sri Lanka, India or Indonesia. By far the main Chinese producing area is the extreme south, below the Tropic of Canear (23.5°N), especially Guangdong 4Liang Guangshang (ed.), Mushu Zalpai yu Liyong [Cassava Cultivation and Use] (Guangzhou: Guangdong Keji Chubanshe [Guangdong Scientific and Technical Publishing House]" 1981), author's preface and p. 4. Cassava is confirmed to have been ~rown in China for more than 100 years in Zhongguo Kexueyuan, Dílí Yanjiusuo, Jingji Dili Yanjiushi [Chinese Academy of Sciences, Institute of Geography, Economic Geography Research Room], Zhongguo Nongye Oili Zonglun [A General Treatise on China's Agricultural Geography), (Beijing: Kexue Chubanshe [Scientific Publishing House], 1980), p. 129. 1820 was also the introduction date mentioned during a spring 1982 delegation from the International Center for Tropical Agriculture (CIAT) and recorded in James H. Cock and Kazuo Kawano, "Cassava in China," unpublished trip report, CIAT, Palmira, Colombia, Jun~ 1982, p. 1. However, Mushu Zaipei yu Liyong clearly indicates that 1820 is the earliest record of cassava cultivation so far uncovered; the introduction date may wel1 have been earlier. I'I.IIVHH ",,~.., ..; H."'''U-UIVn¡H'j f\1;:'jIVII~ VI \.,111:: fCVPIC..., l\'Ct.lU'-', ..... V' ',."'''11.1 ,..>'- ..... ""'-"''''' IV' ..... ~"'V.·."""I . ~, I1, ) , ',V(j:''/ e L ..< ";;h~ ... 9 • 0-.",1." ".,,11.1..... ~\"1'\ L"'''9c ...n IDh Pr~..etlJt. .O~.'i . I ...........1 1.1 Ao.¡t....-. tovftl., ¡ ...,u: .... '~~ " ~ <:> o Wu6tfl9 CU·41" ViII"';:U o ! .,." .' A.. . to"Q_. ... Co6.7 17 .2 0.6 1962 (183.5/158.7) 18.2 1963 153.4 20.2 1964 ." 154.3 19.8 1965 <149 158.5 20.5 (-0.3) 1966 102.2 21.0 1967 70.3 22.0 1968 73.7 25.0 1969 * 124.7 25.9 1970 <201 145.6 24.7 1971 129.6 24.6 1972 167.3 124.5 24.6 1973 107.9 24.3 1974 100.8 * 26.8 1975 <223 131.9 21.8 1976 110.5 22.2 1977 74.6 22.3 1978 (470-530) <236 * 131.0 19.5 1979 156.0 17. O 1980 207.8 14.9 1981 (-350) (-200) 190.4 13.9 1982 S195 175.2 9.9 1983 S158 120.6 5.8 1984 S159 94.0 5.2 Notes: Empty data cells indicate that the statistical information is not available and do not denote zero values. Parentheses enelose rough estimates for .the indicated or nearby years. The applicable years for parenthesized estimates were not stated in the source. Other provinces where farmers grow cassava include Hubei and Síchuan, but sown area is minoro Taiwan Province is now normally not inc1uded in national aggregated statistics for the People's - 5 - Republic of China, although separate data entries for Taiwan are not unusual among PRC statistical compendia. Taiwan is probably included in the 1961 national figure, however. * These figures probably overestimate officially recorded plantings by 20-40 thousand hectare. See Table 7. Sourcas: Guangxi: Guangxi Jingji Nianjian Bianjibu [Guangxi Economic Yearbook Editorial Department) (eds.) Guangxi Jin~ Nianjian 1985 [Guangxi Economic 'Yearbook 1985J (Nanning: Guangxi Jingji Nianjian Bianjibu, 1985), pp. 531 and 593. The 1976 figure was confirmed in Guangxi Nongye Dili Bianxiezu [Guangxi Agricultural Geography Editorial Board] (eds.), Guangxi Nongye Dili [Guangxi Agricultural Geography] (Nanning: Kexue Chubanshe [Scientific Publishing House), 1980), p. 76. The lower figure for 1962 is from Liang Guangshang (ed.), Mushu Zaipei yu Liyong (Guangzhou: Guangdong Keji Chubanshe, 1981), p.9. Taiwan: Republic of China, Executive Yuan, Oírectorate-General of Budget, Accounting and Statistics, Statistical Yearbook of the Republic of China 1985 (Taipei: Republic of China, 1985), p. 281. The 1952-54 figures were added from: Republic of China, Oirectorate-General of Budget, Accounting and Statistics, Statistical Yearbook of the Republic of China 1982 (Taipei: Republic of China, 1982), p. 115. China and other Provinces: The "1978" figure is fram Zhongguo Kexueyuan, Oili Yangjiusuo, Jingji Oili Yanjiushi [Chinese Academy of Science, Institute of Geography, Economic Geography Research Laboratory], Zhongguo Nongye Oili Zonglun [A General Treatise on Chinese Agricultural Geography] (Beijing: Kexue Chubanshe, 1980), p. 129. The "1981" figure is from James H. Cock and Kazuo Kawano, ·Cassava in China", unpublished trip report, - 6 - International Center for Tropical Agricultural Research (CIAT), Cali, Colombia, June 1982, pp. 1-2. The 1961 figure is from Liang, Mushu Zaipei yu Liyong, p. 9. This source also stated that national cassava­ sown area remained around 5 mil1ion mu during the 1960s (300-367,000 hectares, assuming 4.5-5.5 million mu.) The figure for Hunan, Zhejiang and Jiangxi ~ombined was given as around 5,000 mu (333 ha.) in each year of the 1960s. Guangdong: The overestimates for Guangdong for 1965, 1970, 1975, 1978, 1979 and 1982-84 are from Table 7. A 1981 overestimate of 201 thousand hectares was also calculated. The 1979 and 1982-84 estimates are relatively close approximations. The 1965, 1970, 1975 and 1978 figures probably overestimate by at least 20-40 thousand hectares. See Table 7. The 1943 and 1972 figures are from Liang, Mushu Zaipei yu Liyong, p. 9 and the "1981" estímate is from Cock and Kawano, "Cassava in Asia,!' p. 1. seems to enjoy sorne very minor farmer cultivation in Sichuan, but probably not elsewhere within the experimental area. In fact, it is not yet clear from the estirnates of national, Guangdong and Guangxi cultivation assembled in Table 1, that cassava expansion efforts have resultad in significant increased plantíngs outside of those two provinces. In the absence uf a reliable national cassava production series, the best approximation would be to synthesize production series for Guangdong and Guangxi. Fortunately, complete 1950-84 series for Guangxi were published in 1985 (Tabla 2). These data, though not necessarily without flaws, provide the best understanding of year-to­ year movements in cultivation and yields. A glance at Table 2 will - 7 - lle 2. Cassava ~roduction, Area and Yield in Guangxi Zhuang Autonomous Region, 1950-1984 Production Area Yield (Grain Equívalent) (Fresh Root) (Graio Equivalent) (Fresh Root: Tons Tons (Heetares) T/Ha. T/Ha. .iO, 30,045 150,225 41,507 0.724 3.619 )- 39,365 196,825 37,567 1.048 5.239 ¡2 41,870 209,350 48,493 0.863 4.317 ;3 36,635 183,175 41,340 0.886 4.431 ,4 42,535 212,675 67,453 0.631 3.153 ,5 35,365 176,825 62,647 0.565 2.823 ,6 58,280 291,400 93,013 0.627 3.133 .7 91,000 455,000 104,320 0.872 4.362 ;8 165,205 826,025 132,567 1.246 6.231 ;9 140,330 701,650 118,840 1.181 5.904 iO 88,045 440,225 127,913 0.688 3.442 ,1 115,855 579,275 104,353 1.110 5.551 ,Z 189,Z60 946,300 183,547 1.031 5.156 ,3 152,335 761,675 153,433 0.993 4.964 ,4 160,225 801,125 154,307 1.038 5.192 ,5 167,835 839,175 158,520 1.059 5.294 ,6 84,435 422,175 102,220 0.826 4.130 ,7 173,715 868,575 70,300 2.471 12.355 ;8 162,120 810,600 73,667 2.201 11. 004 ,9 216,750 1,083,750 124,733 1. 738 8.639 O 235,990 1,179,950 145,600 1.621 8.104 1 211 ,295 1,056,475 129,613 1.630 8.151 2 262,270 1,311,350 124,480 2.107 10.535 3 206,545 1,032,725 107,900 1. 914 9.571 4 170,765 853,825 100,847 1.693 8.467 5 260,425 1,302,125 131,900 1.974 9.872 6 187,065 935,325 110,473 1.693 8.467 7 141,865 709,325 74,567 1.903 9.513 8 258,295 1,291,475 131,020 1. 971 9.857 9 312,645 1,563,225 155,993 2.004 10.021 O 481,215 2,406·,075 207,760 2.316 11. 581 1 484,280 2,421,400 190,387 2.544 12.718 2 468,255 2,341,275 175,173 2.673 13.365 3 326,680 1,633,400 120,640 2.708 13.539 4 241,180 1,205,900 94,001 2.566 12.829 es: Cassava production and yield data are often quoted in Chinese statistical sourees on a "grain equivalent basis". Sinee 1964, the conversion to "grain equivalence" for al1 root and tuber crops has meant dividing the fresh weight by five, although this wou1d undervalue eassava, sweet potatoes and taro relative to most cereal crops in terms of calories per unjt weight. lt js assumed that the production and yield data in the source for this table appeared in - 8 - "graill equivalent" formo The original data have therefore beell multiplied by five to calculate fresh root weight. Source: Guangxi Jingji Nianjian Bianjibu (eds.), Guangxi Jingji Nianjian 1985 (Nanning: Guangxi Jingji Nianjian Blanjibu, 1985), pp. 531-532 and 593. conflrm that the 35-year perlod encompasses considerable variation in both. During the 195Ds, some government-initiated efforts were undertaken to expand cultivation of cassava which was viewed as a crop capable of providing considerable bulk and caloric content per unit area. One cannot rule out the possibility, however, that a portion of the implied increase in cultivation reflected previously unregistere~ cassava areas eventually included in statistical coverage, especially during the formation of agricultural producers' cooperatives (1954-56) and the people's communes (1958). Elsewheré it has been demonstrated that most of the implied growth in total root and tuber erop area since 1952 is likely to be real, the actual figures remaining, in all probability, within about 5 percent (below) .. the official data. The considerable inerease in cassava area in 1958 parallels an aven largar reported increase for a11 root and tuber craps. While 1958 was a yaar of extreme statistical distortion, casting daubt on GBruee Stone, ItAn Analysis of Chinesa Data on Root and Tuber Crop Próducti órf;-u-TheChi na Quarter 1 y, September 1984, pp. 594-630. - 9 - the magnitude of the increase, the implied growth was no greater than that of 1956, mueh of which may have been real. 1958 was also ayear in which great efforts were made to increase foodcrop produetion by whatever means possible. Root and tuber craps, including eassava, W9pe carrectly identified as the easiest means to effect a short term leap in bulk foad praduetion. It is diffieult, however, to aecept the implied 1958 increase in average yield to an unpreeedented level, especially in view of the (except for sweet and white potatoes, more modest) expansion of area planted with other foad crops and maintenance of yields in that year. In sum, while it appears that the total Guangxi foodcrop data (excluding eassava) have been adjusted in the 1985 Guangxi Economic Yearbook for the statistical distortion typical of 1958 published materials, it is quite possible that those for ea5sava may not nave been, particularly in the yield category. The decline in 1959 area, however, followed by some recovery in 1960 are undaubtedly real, altheugh it is impassible te verify the exact figures. lnflated reports af miraculous grain production success in 1958 led authorities to increase area sown with ecenomic crops in 1959 at the expense of staples. 7 When the truth became clear (1958 had beeo a good, but not spectacular year), it was too 7Li Choh-ming, The Statistical System af Communis~ China (Berkeley: University of California Press, 1962); Kenneth R. Walker, Food Graio Procurement and Coosumption in China (Cambridge: Cambridge University Press, 1984); Nicholas R. Lardy, Agriculture in China's Modern Economic Development, Cambridge: Cambridge University Press, 1983. 10 - late to correet spring planting. Some compensation would have been made with 1959 fall planted cassava, however, and in 1960, in view of poor harvests for all fooderops the previous year. The yield decline in 1960 is consistent with widespread natural disasters throughout China estimated to be'the worst in the twentieth century. These were somewhat less severe in Guangxi than in some other provinces, but yields of other Guangxi food crops reportedly decline 'by a weighted average of 9 percent during 1960 and 1961. 8 Spring planted cassava, in particular,is subject to insect damage during the seedling period and in the fall, typhoon damage. The low area figure for 1961 is consistent with both poor statistical coverage during the period and significant rural dislocation associated with the 1960-61 famine throughout China which may have partial1y extended into Guangxi. The large increase in cassava area in 1962, followed by subsidence duríng the following few years is also explainable in terms of reaction to the 1960-61 famine. Geographic coverage may not have been consistent throughout the series. Qinzhou Special District was transferred from Guangxi to Guangdong in 1955, then back to Guangxi in 1965. Qinzhou includes the entire current Guangxi coast and extends north from the current provincial border to the Yu River, then angles southwest towards the 8Guangxi Jingji Nianji4n Bianjibu [Guangxi Economic Yearbook Editorial Board], Guangxi Jingji Nianjian, 1985 [Guangxi Economic Yearbook 1985] (Nan~ing: Guangxi Jingji Nianjian Bianjibu, 1985), p. 530. ' - 11 - border with Vietnam. In 1976, area sown with foodgrains in Qinzhou covered 461,333 hectares. Area planted with root and tuber crops in the western district of Guangdong circa 1957 (including Qinzhou Special District and Zhanjiang Prefecture) consisted of 28.3 percent of total area sown with foodcrops (excluding soybeans), a little less than 5 percent of which was planted with cassava and umaou potatoes. 9 These reports suggest that something on the order of 6 thousand hectares of cassava were transferred from Guangxi to Guangdong in 1955, then (potentially more extensive cassava area) back to Guangxi in 1965. This could explain the counter-trend movements of cassava area in the Guangxi series for 1955 and 1965. Data oscillations during the succeeding decade (1966-77) are less understandable as a function of nationwide economic developments and may be peculiar to cassava or to Guangxi. Hypotheses for explaining these oscillations Inelude the lagged effeet of earlier shocks echoed via the rotation system (see below) and perladic reclamation inltiatives. In Guangxi, cassava is often grown during the early years of a reclamation project in order to earn sorne economlc return before reclamatlon is complete. When the quallty of farmland construction and field preparation permits, cassava is often phased out to make way for more hl9hly valued trops. 9Bruce Stone, nAn Analysis of Chinese Data on Root and Tuber Crop Production," pp. 612-615. - 12 - The low planted area figures for 1967 and 1968 and, particularly, the high average yield estimates for those years are especially anomalous. Although fertilizer use accelerated during the 19605, widespread application to cassava as early as 1967-68 is very unlikely. One is consequently motivated to hypothesize about a statisti cal qui rk: e. g. independent pr,oduct ion and areaest i mates with the latter underestimated due to statistical confusion typícal of tbe early years of the Cultural Revolution period (1966-77). Even excluding 1967-and 1968, the data indicate a marked increase in yields from an average of 4.5 tons per hectare (1950-66) to 9.0 tons per hectare (1969-77) or 10.3 tOIlS per hectare (1969-84). Sorne of this increase per unit productivity is explainable in tarms of initiation of fertilizer application, and cultivation of cassava on state farms with plentiful access to fertilizers. But state farms in Guangxi occupied only 20 thousand hectares (1982) and large portions of this total were devoted to cultivation of grain crops and sugar cane. lO lt seems unlikely, therefore, that increased fertilizer use alone can ful1y explain this yield íncrease. In the absence of definitive information, what co~ld explain a sudden doubling of average yields in the mid-1960s? One hypothesis v/ould emphasize technical change. t~uch of the important selection and breeding work was undertaken in the late 1950s and early 1960s. lOZhongguo Guoj i a Tongj i ju, Zhongguo T()ngj i Ni anj i an 1983, pp. - 13 - The South China Tropical Crops Research Academy bred or selected many of the ~ell-known varieties under current production representing significant improvement in aggregate speed and quantity of root production auring the 1959-62 periodo The South China Agricultural Science Academy in Guangzhou bred or selected for multiplication and dissemimation, severa1 other higher yielding varieties during the 1957-62 period. 11 Particular attention paid to cassava during this period may also have produced important results in improving field cultivation techniques. Anather hypothesis would suggest that cassava cultivation on somewhat better land was initiated during this periodo The Cultural Revolutlon decade {1966-77} was marked by a pol icy of local self­ suffi ciency i f1 grai n product ion and es ca 1a ti on of quota de Ti ver i es. In sorne cases, quotas were specified in terrns of particular crops needed by the state. In other cases, quotas were specified only in terms of weight of staples leaving the choice of crops to each collectivity of farmers. Although farmer! received compensation for quota deliveries, prices were notoriously low, involving an implicit tax. Land taxes, arnounting to roughly 5-13 percent of output during this periad depending on location, were also payable in kind. Taxes and quotas were therefore obligations to be discharged with cOffimodities achieving the highest bulk yield per unit area. Altho~gh fresh weight of root and tuber crops was divided by 4 for these 11Liang Guangshang led.}, Mushu Zalpei yu Liyong, pp. 77-78. - 14 - accounting purposes through 1963, and by 5 thereafter, cassava may hava been cultivated and ave n fertilized by a wider variety of localities in South China with the express purpose of expeditiously discharging these obligations. 12 The determinants of variation during the final period (1978-84) are somewhat easier to identify with confidence. The steady growth in yields is almost certainly related to an increase in manufactured fertilizer nutrient app1ication. Although average application 1evels for cassava are not knownwith precision, nutrient application within China as a whole tripled between 1976 and 1984 and doubled between 1978 and 1984 culminating with an average rate of 120.6 kg./ha. of sown area. Efficiency of uti1ization a1so increased during the periodo Although .. the average 1evel in Guangxi was somewhat lower, it grew even more rapid1y than the national average between 1976 and 1982 (to 110.2 Kg.jha.), then stagnated in 1983 (112.4 Kg./ha.) and 1984 (109.7 Kg./ha.), paralleling yield progress in Guangxi. 13 12For further discussion of these Issues, see Bruce Stone, "China's 1985 Foodgrain Productlon Target: Issues and Prospects" in Anthony M. Tang and Bruce Stone, Food Production in the People's Republic of China IFPRI Research Report no. 15, (Washington, D.C.: International Food Policy Researeh lnstitute, 1980), pp. 147-149. 13Bruce Stone, "Chinese Fertilizer Application in the 19805 and 1990s: lssues of Growth, Balance, Allocation, Efficiency, and Response" in US Congrass Joint Economic Committee (eds.), China's Economy Looks to the Year 2000, vol. 1 The Four Modernizations . (Washington, D.C.: U.S. Government Printing Office, 1986, pp. 453- 496; and State Statistical Bureau, PRC, Statistical Yearbook of China 1985 (Hongkong and 6eijing; Economic Information and Agency, and China Statistical Information and Consultancy Service, 1965), p. 283. - 15 - Application of manufactured fertilizers to cassava is likely to be mueh below the average level for all crops in Guangxi except on state farms, but scattered survey reports 14 confirm that on farmers' fields near cassava research institutions in South China, yields which al'e comparable to the recent Guangxi- provinéial averages are only obtainable with fertilizer application, ur under good soil and climatie conditions atypical of most Chinese cassava growing areas. One of the survey respondents, however, also indicated that the cassava research in China had made significant progre!! in developing improved varieties and low-cost cultural practices a decade earlier. Yet the predominant varietias planted in the 1980s were among those selected (or bred) during the late 1950s and early 1960s (see below). The rise and fall in cassava area during tha 1978-84 period is attríbutable to a number of factor!, the most powerful of which has bee" the rise and fa" of opportunities for export to the European Community. With EC pressure on Thailand (the dominant and 101'1 cost supplier) to reduce exports during the late 19705, Chinese exports responded to the opportunity with rapid growth in 1979, 1980 and 1981 14"Delphi Survey for the Assessment of Potential 'Yields of Cassava" cireulated to eassava breeding institutions in China and elsewhere by J. S. Sarma, lnternational Food Policy Research Institute, 1986. The respondent who mentioned varietal and cultural improvement a decade ago \'las líu Yingjing of the South ·China lnstitute of Botany in Guangzhou. - 16 - (Table 3) before similar pressure eventually forced a deceleration beginning in 1982 (with 1981 fall sown cassava).15 Other circumstances contributing to this responsiveness involve changes in rural institutions since 1978-79: farmers have been allowed more control over cropping and management decisions, but are also afforded less market security from the government as a guaranteed buyer. At the same time, very poor locations typical of many Chinese cassava-grQwing are as have been released from tax and quota obligations, while the government, in response to substantial success in accelerat;ng national foodcrop production growth, bagan emphasizing higher quality in farm procurement items compared with the considerable previous period emphasis on cheaper bulkier products such as most roat and tuber craps and the lowest quality grades of cereal crops. These consideratians, coupled with the overall liberalization of econom;c activities in rural areas explains the fall in cassava area to a 1984 level below that typical of the pre- 1978 periodo The decline in sown area cuts aeross most graio crops throughout China, but is particularly nateworthy in proportional terms in the case af crops typically grown in poorer farmlands and characterized by low prices aod weak markets such as sorghum, white potatoes, bean craps and, no daubt, cassava (Table 4). In Guangdong and Guangxi, although unsuitable for such a warm moist climate, 15Bruce Stone, "An Analysis of Chinese Data on Root and Tuber Crop Production," pp. 623-625; Bruce Stone, HAn Examination of Economic Data on Cassava Production, Utilization and Trade in China," pp. 16-22. - 17 - Table 3. PRC Cassava Exports, 1963-1984 TotalCassava Oried Cassava Cassava Tapioca Cassava Starch Exports in To·European Share of Ee net Total Fresh.Root Community Only Cassava Imparts Exports Eguivalents (metric tans) (percent) (metric tons) (metric tons) (metric tans) (metric tans) 3 20,977 4 ·33,393 5 72,676 6 57,077 7 53,173 B 28,015 9 1,324 O 4,984 1 14,859 2 16,070 3 8,083 4 4,111 0.2 - 4,000 11,429 5 4,211 0.2: 4,000 11,429 6 7,253 0.2+ 7,000 6,500 2,000 60,657 7 999 0.0+ 1,000 2,000 11,948 3 1,327 0.0 1,000 1,000 7,403 ~ 51,449 1.0: 51,000 5,800 2,060 183,522 ) 335,989 6.9 336,000 20,500 2,500 1,067,070 1 606,589 9.1: 607,000 10,000 1,500 1,788,073 > 440,181 5.4_ 445,000 14,000 1,500 1,343,397 3 15,222 0.4 - 460,000 1,314,285 ~ 143,000 2.7 1,314,285 ~s and Sources: European Community data for dried cassava imports from China and other countries are liled from EUROSTAT and NIMEXE Analytic Tables for Foreign Trade (which are in clase !ement). Total dried cassava, cassava tapiocaand cassava starch export data are from Food Agricultura Organization of the United Nations, "Supply Uti1ization Accounts Tape, 1984," !, 1985. The fresh root equivalents of all cassava exports aggregated together appear in "Standardized Commodity Balance Tape, 1984" Rome; 1985. The 1983 and 1984 data must be ,rded as open to some question and may be revised in future compendia. ~ ",.' o 4. Area Sawn with Majar Cereals, Bean Crops, Roots and Tubers in China, 1976-85 $weet Other and Only Only Cereals \4hlte Sweet White & Bean Total Rice Wheat Corn SoybeanJL MI 11 et SOrJi¡hum Potatoes ,Jl.otatoes ~otatoe~ _erops Foodgrains (thousand hectares) 1976 36,217 28,417 19,228 6,691 4,501 4,329 10,366 lQ,994 120,743 1977 35,526 28,065 19,658 6,845 4,477 3,759 11 ,229 10,841 120,400 1978 34,421 29,.183 19,961 7,144 4,271 3,456 11,796 -6,800 -5,000 10,355 120,587 1979 33,873 29,357 20,133 7,247 4,173 3,173 10,952 10,355 119,263 1980 33,879 29,228 20,353 7,227 3,872 2,693 10,153 9,829 117,234 1981 33,295 28,307 19,425 8,023 3,888 2,610 9,621 9,789 114,958 ,1982 33,071 27,955 18,543 8,419 4,039 2,783 9,370 6,916 2,454 9,283 113,463 ..... co 1983 33,137 29,050 18,824 8,414 4,087 2,707 9,402 6,840 2,562 8,426 114,047 1984 33,179 29,577 18,537 7,286 3,797 2,384 8,988 6,426 2,562 9,136 112,884 1985 32,070 29,218' 17 ,694 7,718 8,571 108,845 $ources; Most data were converted from Chinese unlt figures or were calculated from data appearing in State Statistical Bureau (SSB), PRC, Statistíca,"Yearbook of China 1985 (Hong Kong and Beiding; Economic lnformation and Agency and China Statistical Information and Consultancy Service Centre (CSICSC) 1985), p. 253. 1985 data were added from SSB, PRC, China; A Statistical Survey in 1986 (Beijing; CSICSC, 1986), p. 37. 1982-84 figures for sweet potatoes and for white potatoes are from He Kang et al, Zhongguo Nongye Nianjian Bianjl Weiyuanhui [Chinese Agricultural Yearbook Editorial Committee] (ed.), Zhongguo Nongye Nianjian 1983 [Agricultura1 Yearbook of China 1983] (Beijing: Nongye Chubanshe [Agricultural Publishíng House], 1984), p. 40; He Kang et al, Zhongguo Nongye Nianjian 1984 (Beíjing: Nongye Chubanshe, 1985), p. 88; He Kang et al, Zhongguo Nongye Nianjian 1985 (Beijing: Nongye Chubanshe, 1986), pp. 147-148. The estimates for sweet and white potatoes in 1978 are from Bruce Stone, nAn Analysis of Chinese Data en Root and Tuber Crop Productíon,· The China Quarterly September 1984, p. 628. - 19 - wheat had been cultivated for import substitution purposes. W;th relaxation of this uneconom;c emphasis on wheat, sown area declined in the two provinces. Less drastically, area sown with several other food trops, such as paddy, sweet potatoes, sorghum and millet, also fell in favor of economic craps, especially sugarcane (Tables 5 and 6) . After 1979, is it possible to confirm that the trends indicated for Guangxi are representative nationally? Even without national data, the addition of series for Guangdong would provide a reasonable proxy. Unfortunately, cassava series for Guangdong are unavailable, but a very rough approximation may be discerned from Table 5. The left-hand column is comprised of figures quoted for Guangdong specifícally. The center column is derived from data appearing in the 1984 and 1985 Guangdong Statistical Yearbooks. These data are not estimates of cassava area ~~, but are formed by deducting data for sugar cane, peanuts, sesame, jute, kenaf and tobacco from figures for total area planted with economic crops. The estimates in parentheses to the right more ,closely approximate cassava plantings inasmuch as area sown with all oil crops, all fibers, and medicinal herbs have also been deducted from the "economic crop" area along with sugarcane and tobacco on the basis of recent Agricultural Yearbook of China volumes to arrive at the residuals. During the recent decade at least, cassava has been classified as an economic crop in production statistics, rather than as a foodcrop, and the calculated residual should be predominantly 20 - .". ...... "" .... ~ ,.::..; ...... o ;:;g .. - ~ .. ~~~~=:~ f:';"'!! ; ~ 2 ;; :; L .- :1 :e"" ........ ""'".."" 1, !;.~: ~~~....:t~ ... !!.f;!? :;- -;:: 2 ~ - ~ rr~ .!.~~ -;;;;:~;. ;~~ ! t! .... g:~ .. <:>!2:;; .. ~ ~ ..: s: ..... ..: . ::to- "" "" - .'5 ~- ;;'0'; t. ~t -l! j' o _ ,. :o ..... r-. -o .......... .". ..... ~.. c. ... i~;:.'::~; -........ ~ < < - ~ ... " ~~ ;; . .c>- - 9 ::; ... :~ .. ,!~~~ r; ". ..... ..,, "--""-- ;g~!S~!:: ~~":;~ g:;: ~ :;;". ~:1:: ~:'.:.; ir.:;: :i .;. ~., % - ~!: ,i:: ,"~ q ~~~ !! &< ~ !i.: ~~a·~ :~ ~~ - ~~ « !l:~~:i ~ €Y' '" , .. ,~ f; ~ '-.!: ~1 ~ ,~ FJi!~ 21 ~ s::? ;i! :: .... tí ~~..~ :c !!;: ..... 1 ~ j ;~;o ~r::;:5! .... -;:~ ... ~l!!rJ~ ~ ~;~~~;; !:!i":"; ~ -<>;~ ~ » q~~~ ;.:;e;""~ ~~ :.;~ ~~!7! ~ :;;:,;:! ~i:~~ :~;;~_~N ~~ci'¿zl ~ ~.!: :;;J¿~":; ~¡;;. x " ~ ... ~~."~ í.~f~ ~~~~ .... ! ;:.;;;; ro ~SS:~!! ;:::~;::. ~ :s:;~ :! ~ .J~-:."" ~ ~ >- ~ i~ "I~J ;. !: ~ ~~I:I !2 - comprised of, but should overestimate area planted with cassava. The estímate in the right-hand calumn is derivad by deducting publíshed Chinese estímates for area sown with cassava in Guangxí (1961), Taíwan (1961), Fujian (1961), Yunnan (1960), Guizhou (1961) and Hunan, Zhejiang and Jiangxi (circa 1960s) from a published 1961 national figure. The calculated figure substantially exceeds the residual-based overestimates of cassava area in Guangdong for surrounding years in a periad when cassava area in other Chinese provinces was undoubtedly amall. These data are eVidently in conflicto An examination of 1950s Chinese material provides an impress;on that 19505 cassava area in Guangdong was greater than that implied by the residual-based "overestimates· in the center column of Table 7. Guangxi cassava area in 1957, for example, was around one-quarter of all Guangxi farmland plantad with root and tuber craps. If the same proportion were relevant for Guangdong, 1957 cassava area would total more than 300 thousand hectares. But whereas 36.21 percent of Guangxi root and tuber crop production consisted of crops other than sweet potatoes, this figure was only 13 percent for Guangdong, and included cassava, taro, white patatoes and "mao· patatoes, primarily the first two categories. 16 Still, 1957 Guangdong cassava area could easily have been in the range of 100-200 thousand hectares. 155ee ~at~ and Chlnese saurees eited in Bruee Stone, "An Analysis of Chinese Data on Root and Tuber Crap Productton," pp. 609- 616. - 23 - ¡ble 7. Estimates of Area Sown with Cassava in Guangdong Province, 1943-1984 Residual-based National estimate Guangdong Cassava estimates of minus Guangxi, Yunnan, area estimates in "other" economic Fujian, Taiwan, Guizhou, Chinese sources crops in Guangdong Hunan, Zhej i ang & Jiangx i (thousand hectares) 143 33.4 152 25 /57 57 361 -240 162 25 l65 149 170 ~ 201 372 167.3 l75 223 n8 236 179 (215) 180 237 181 -200 (201) 182 243 (195) 183 188 {l58} )84 206 (159) )urees: Data appearíng in the left- and right-hand columns are based on Table 1 exeept that the Taiwan Provínce figure deducted along wíth those from other provinces from the national estímate for 1961 (10,000 ha.) was taken fram the same source as the natiana1 fígure, Liang Guangshang (ed.} Mushu Zaipei yu Liyong, p. 9. Data appearing in the centar column are based on data from Guangdongsheng Tongjiju [Guangdong Province Statistical Bureau] (ed.), Guangdongsheng Tongji Nianjian 1984 [Guangdong Province Statistical Yearbook 1984] (Xianggang: Xianggang Jingji Daobao Shechuban [Hong Kong Economic Reporter Publishing House], 1984), pp. 113-114; and Guangdongsheng Tongj i j u, Guangdongsheng Tongj i Nianjian 1985 [Guangdong Province Statistical Yearbook 1985) (Xianggang: Xianggang Jingji Daobao Shechuban, 1985), pp. 107-108. Sown area data for sugarcane, peanuts, seSame, jute, kenaf and tabacco were deducted fram total area sown with economic crops. Data for rapeseed and other oilcrops, other fibers, and medicinal hel'bs have a1so been deducted from the figures appearing in parentheses on the basis of Zhongguo Nongyebu [Chinese Ministry of Agricu1ture], fhongguo Nongye Nianjian, 1980, 1982, 1983, 1984 and 1985 (Beijing: Nongye Chubanshe [Agricultural Publishing House], 19B1, 1983, 1984, 1985 and 1986). - 24 - During the 1950s, cassava was treated explicitly as "shulei" [including both tuber crops and tuberous roots], which in turn were classified as "liang~hl" [staple food crops], occasionally as part of "miscel1aneous grains." By the mid-1970s, however, it Is clear that cassava was excluded from "shulei" and "liangshi" statistícs and incorporated as a sub-category or as a residual within "jingjí zuowu" [economic crops]. The transltion date has not been clearly determined, although 1964 and 1976 have beensuggested as candidates. 17 In view of the trends exhibited for Guangxi in Table 2 and the foregoing discussion attempting to resol ve the conflict implied in Table 7, it seems likely that the 1950s economic crop statistics appearing in the Guangdong Province Statistical Yearbooks, though recently published, are unlikely to have been adjusted for inclusion of cassava; hence the center column cannot be used as a proxy for cassava are a for the 1950s nor probably for 1962. From 1965 onward, however, these residuals may well provide the best indication of trends in (though not exact estimates of) Guangdong cassava area, since cassava is likely to dominate the category. It should be noted, however, in view of economic liberalization since 1979, that the divergence of this residual series and actual cassava area is likely to have increased, especially since the decline in export opportunities in the early 1980s. 17op . cit., pp. 600-604. - 25 - Unfortunately, despite the availability of an official cassava series for Guangxi and a rough approximation of trends for Guangdong, it is still not possible to be definitive about national trends for China. It is clear that cassava was planted on less than 100 thousand hectares in the mid-1940s, rising quickly to perhaps around 250 thousand hectares by 1957 and 355 thousand hectares (excluding Taiwan) by 1961 during the famine. Total plantings on the Chinese mainland probably subsided to roughly 300 thousand hectares by 1965 and were cer~ainly not much lower in J972 when plantings in Guangdong and Guangxi alone totalled 292 thousand. Official are a sown with cassava in the two southern provinces seems to have risen to 370 thousand hectares in 1979, perhaps peaking in 1980 at 410-420 thousand hectares, subsiding,to 390 tha and 370 tha in 1981 and 1982 and plummeting to 275 tha and 250 tha in 1983 and 1984. But whether cassava area rose appreciably outside of these two southern provinces since the early 1960s is not clear. The (undated) total of 350 thousand hectares given to the CIAl delegation by Chinese cassava breeders in spring 1982 would imply that it has not, while the (undated) Institute of Geography estimate (around 500 thousand hectares) published in 1980 suggests either considerable expansion into other provinces or more aggressive estimates of non­ field cultivation. Barring the unlikely event of relatively even distribution among other mentioned provinces, official1y recorded plantings of 120-190 thousand hectares outside of Guangdong and Guangxi implied by the Institute figure and the provincial estimates - 26 - would surely have be en mentloned by the breeders or in cassava­ related publications, while the 350 thousand hectare figure, though purportedly ineluding an estlmate for eassava on private plots, does not even appear to caver probable plantlngs in the two southern provinces. Part of the problem is that eaS5ava area is undaubtedly more difficult to estimate than that of most field craps, since considerable proporti~ns are grown on prívate ~lots, on narrow strips adjacent to roads and fields, on hilly and incompletely cleared 1and not yet or normal1y considered farmlands, and on tiny corners not even counted among private plot statístics. There is even sorne 111egal cultivation: under trees on state rubber plantations, for example. 18 The Institute of Geography figure probably incorporates a more aggressive estímate, based on sorne survey evidence of these kinds of plantíngs which in large part elude offíeial statistical coverage. All that can be claimed with near certainty is that national cassava planting reached another major peak in the late 19705 or early 1980s, and then declined rapidly with the subsidence of opportunities for international trade, increasing liberalization of rural economic activities and a probable cut back in the government's role in cassava marketing. 180p • cit., p. 621. - 27 - National production trends are even less discernible. The only available figure for recent production is 3 million tons provided to the CIAT delegation in spring 1982, 19 a1though like the 350 thousand hectare figure provided at the same time, it may wel1 be an underestimate. The best indication of national yield trends is undoubtedly the Guangxi series in Table 2 with some reservations about a few of the years 5uch as 1967 and 1968. The national average implied by the f~gures given to the CIAT de1egation is 8.6 tons per hectare, suggesting that average yields in Guangdong and elsewhere are lower than in Guangxi. But this comparison, too, cannot be taken too literally, since the four to five tons per hectare 1981 Guangdong average suggested. by such an exercise implies too great a divergence between Guangxi and Guangdong, particularly in view of greater general availability of fertilizer in the latter provinee. Within these two southern provinces, some of the principal cassava-growing areas can be identified. The first record of Chinese cassava cultivation was in 1820 in Gaozhou County, part of Zhanjiang Prefecture in southwestern Guangdong. 20 Gaozhou is not a coastal county and earlier cultivation is entirely possible. In the 1950s, there is continued record of cassava in Zhanjiang Prefecture, where uplands constituted 27.5 percent of cultivated land, a greater 19James H. Cock and Kazuo Kawano, "Cassava in China," unpublished trip report, Internatienal Center fer Tropical Agriculture, Palmira, Colombia, June 1982, p. l. 20Liang Guangshang led.), Mushu Zaipe! yu Liyong, p. 4. 28 - proportion than in othar Guangdong Prefectures. Suixi County and tha Zhanjiang city suburbs (where uplands comprised 12 parcent) in the center of the prefectura, and Xuwen County on tha southern tip of the Leizhou Peninsula are mentioned in 1950s literature on cassava, but the erop may haya baen grown more generally throughout the grain deficient Leizhou Península and in the uplands adjacent to the Jianjiang Plain where "miscellaneous grains" (80.9 percent of which were root or tuber crops) comprised 44 percent of staple foodcrop production in 1955. Throughout the Zhanjiang Prefecture and enclosed municipal areas, root and tuber crops (valued at one-fourth fresh weight) constituted only 28 percent of staple erop production which occupied 95 percent of sown area. Sweet potatoes were the principal root erop, however, with cassava and "mao" pota toes comprising a little less than 5 percent of root and tuber erop production. 21 But cassava cultivatlon clear1y was not llmlted to southwestern Guangdong In the 1950s. There is a1so record in the Economic Geography of South China (1959) of eassava and taro being grown in the mountainous uplands surrounding the Sui and Xi River Va11eys in West Central Guangdong, notably Huaiji, Guangning, Sihui, Gaoyao and Oeqing Counties, al1 in Zhaoqing Prefecture. Cassava was not speeifically mentioned in the discussion of Hainan Island, but has 21Sun Jingzhi (ed.), Huana" Oichu Jingji Oili [Economic Geography of South China] (Beijing: Kexue thubanshe [S¿ientific Publishing House], 1959). Translated in Joint Pub1ications Research Service, August 24, 1969, no. 1'4954, pp. 137-138 and 178-179. ~Jhen these statistics were gathered, the region ineluded the Qinzhou Special Oistrict encompassing known cassava-growing areas such as Hepu Gounty and the Beihai suburbs. - 29 - been grown there at least since 1912 when a well-known Malaysian variety was introduced into Dan Xian rubber plantations. According to 1951 statistics, roots and tubers accounted for 38.5 percent of grain consumption in plainsareas of the Island and 69.8 percent in hilly districts, paddy rice providing most of the remainder in both cases. 22 In Guangxi, cassava was generally distributed in the Xunjiang and Liujiang Valleys (east central Guangxi) characterized by relatively barren, drought-prone land. Yet yields of 7.5-15.0 tons per hectare were cited.· It was used as food, feed and to produce starch for cotton yarn, in the city of Wuzhou in east central Guangxi on the Guangdong bordar where Guangxi's first starch factory was opened in 1952. Cassava was also widely planted in southeastern Guangxi and along the southern coast, especially Hepu County and the suburbs of Beihai on the southeast coast. But although Beihai and Wuzhou remained major centers, by the mid-to-late 1950s, cassava starch factories and consequently expanded cassava cultivation had spread widely in the Autonomous Region including Ningrning in the southwest, Barna Yaozu Autonomous County toward the northwest and Wurning in the center of the Region. 23 In Yunnan, cassava cultivation in 1960 was recorded in Hekou Yaozu Autonomous County in the south 220p . cit., pp. 137-138 and p. 201. See details of varietal transfer below. 230p . cit., pp. 258 and 333-334; Guangxi Jingji Nianjian Bianjibu, Guangxi Jingji Nianjian 1985, p. 192. - 30 - along the Vietnamese border, in Dehong Daizu Jingpozu Autonomous Prefecture in the west along the Burmese border, and elsewhere. 24 By 1972, Zhaoqing Prefecture had taken over as the principal cassava growing region of Guangdong, accounting for 57 thousand hectares or 33.9 percent of the provincial figure for that year. Zhanjiang Prefecture was next with 33 thousand hectares or 19.5 percent. The remaining 77+ thousand hectares were distributed throughout Guangdong, including Hainan Jsland and Shaoquan, Meixian, _ Shantou, Foshan and Huiyang Prefectures. Sorne of these secondary regions increased cassava plantings rapidly in the late 1970s. Cassava area in Meixian Prefecture for exarnple, in the northeast corner of.the province. grew from 10,800 hectares in 1977 to 40,000 hectares in 1978. 25 Jn spring of 1982, a delegation of cassava breeders trom the International Center for Tropical Agriculture (CIAT) visited a number of cassava growing areas in Guangdong, including Baisha County and Haikou Municipality on Hainan Island, three state farms in Zhanjiang Prefecture and Dongguan County (Huiyang Prefecture) on the Pearl River Delta. Some impression of area trends on the Delta can be obtained fram statistics for Dongguan. Cassava plantings declined from 8,600 ha. (1957) to 4,600 ha. (1977) with much of the decline occurring in the 19705. Cassava area then fell even more rapidly to 24Liang Guangshang (ed.), Mushu Zaipei yu Liyong,. p. 9 251' b1'd . - 31 - 3,157.4 ha. in 1978, then 3,100 ha. (1981) and 2,816.8 (1982). But on the other side of the Delta in Taishan (Foshan Prefecture), cassava was not grown on a large scale until recently. And Fucheng Commune (within Dongguan County) cassava area fell from 500 to 367 hectares between 1980 and 1981, but recovered to 434 ha. in 1982. 26 Vields observed by the CIAT delegation were generally in the 6 to 8 ton/ha. range, but 20-25 tons/ha. was claimed for sorne state farms and experiment stations. 27 Average yields for Dongguan County on the Delta were 11.73 tons/ha. in 1978 and 15.76 tons in 1982. Fucheng Commune within Dongguan County claimed around 15 tons/ha. in 1980, 14.43 tans/ha. in 1981 and 17.75 tons/ha. in 1982. 28 In Guangdong generally, with 1200-1800 mm of annual rainfall, yields on farmer's fields with poor soils have been estimated by one Chinese breeder to fall typically between 5 to 7 tons per hectare and between 10 to 13 tons under good climatic conditions and soil conditions. Throughout Southern China (800-2000 mm/yr annual rainfall) yields are estimated by another breeder to be 5 to 9 ton5 per hectare on poor soíls and 15-30 tons/ha. (avg. 20 tons/ha. ) under good conditions. Without fertilizer or irrigation, however, poor soil yields were reported to be 3 to 6 tons/ha. (average 4 ton!) and for good soils 26Cock and Kawano, "Cassava in Asia", op. cit. The 1957, 1977 and 1981 figures for Dongguan County are from p. 13. The 1978 and 1982 data, the Fucheng Commune data and the impress;ons fer the 1970s and for Taishan are from Prof. Graham Johnson, Dept. of Anthrepology and Sociology, University of British Columbia, correspondence, Sept. 19, 1983. 27Cock and Kawano, "Cassava in China", p. l. 28Graham Johnson oo. cit. - 32 - with good weather 12 to 18 tQns/ha. In Zhaoqing and shaoiuan Prefectures (1450-1700 mm/yr. avg. rainfall) farmers' yields without fertilizer and irrigation were reported by an agronomist specializing in cassava to average 6.4 tons/ha. under poor conditions and 11.2 tons/ha. under good conditions. With fertilizer but without irrigation, these averages rose to 11.69 tons/ha. and 19.7 tons/ha. with ranges of around 4 tons/ha. Average yields on research stations run 2 to 10 tons per hectare higher than those quoted abo ve for farmers' fields.- 29 These data in sum would seem to suggest that most cassava in Guangdong is grown on poor land, especial1y uplands and until recentl y, rare 1y r.ecei ved much fert il i zer. Tata 1 cassava area has fallen during the past decade or so on better lands such as those typical of the Pearl River Delta (with scattered temporary exceptions due to the short-lived EC export opportunitles) leading to sorne decline in the average quality of farmland growing cassava. Thís decline has been more than counterbalanced by the increase in fertilizar applicatlon to cassava in recent years such that average yields (though not necessarily total production) have Inerelsed sharply. The higher cassava yields on state farms and for private and cooperative farmlng In the Pear1 River Delta locations líka 29Delphi survey responses sent to J.S. Sarma (IFPRI) for Shaaquan and Zhaoqing Prefectures by Huang Xi of the Institute of Drought Grain Craps, Guangdong Province Academy of Agricultural Sciences, Guanzhou, June 28, 1986; for Guangdong by Liu Ylngjing of the South China Institute of Botany, Chinese Academy of Sciences, Guangzhou, June 30, 1986; and for South China Academy of Tropical Crops Research, Dan Xian, Hainan Island, June 20, 1986. - 33 - Oongguan County are partially explainable in terms of greater access to (and more attractive relative prices for) manufactured fertilizers, as well as to often better soil and higher standards of agronomy. But an additional important factor relates to varietal adoption. An especially sma11 portion of cassava grown on state farms and on the Delta is likely to be utilized for direct human consumption, so there is little reason for managers and farmers to cultivate the lower yielding sweeter varieties characterized by low cyanlde and higher protein content, as well as greater overa1l pa1atibility (see below). The argument is at least partia1ly ~ relevant for Zhaoqing and Sha09uan Prefectures, which are becoming one of Guangdong's major regions for processing industries utilizing cassava, and, for similar reasons, east central and southern Guangxi, historica1ly among the principal cassava-growing areas within the Autonomous Region. Cassava production systems: Cassava in China is grown both extensively and in small plots and scattered p1antings. Extensive cultivation is most notable on, but by no mean s confined to state farms, and is principally associatedwith starch production, the domestic animal feed market and exports. Outside the state farm sector, with the formal dissolution of the communes in favor of the household production responsibility system, it is safe to assume that extensive cultivation has declined somewhat since the early 1980s. However, - 34 - Graham Johnson has pointed out 30 that rural reforms have, in some instances, strengthened rather than weakened cooperation in South China, so it cannot be assumed that extensive cultivation in the old cooperative sector has disappeared. Since the formation of agricultural producers cooperatives (1954-56) and the people's communes (1958), collective lands constituting the vast majority of Chinese farmlands, have been cu1tivated communa11y. However the 54 thousand-communes have normally not been the principal cultivation unit. More often smaller units, the 719 thousand brigades, or most commonly, the 5.6 million production teams have cultivated as cooperative groups. A production team normal1y consistect of around thirty farm families (an average of 139 people) that pooled usual1y contiguous land and shared cultivation responsibilities. 31 The principal farm unit varied geographically in sfze, but by the late 1970s averaged around 8.6 hectares in Guangdong and 8.9 hectares in Guangxi, and certainly les! in the very denseiy populated Pearl River Delta of Guangdong. 32 30Graham E. Johnson, "The Production Responsibility $ystem in Chinese Agriculture: Sorne Examples from Guangdong "Pacific Affairs, vol. 55, no. 3 (Fal)) 1982, pp. 430-449. 31Zhongguo Guojia Tongjiju [State Statistical Bureau of China], Zhongguo Ton9ji Nianjian 1983 [Statistical Yearbook of China 1983] (Beijing: Tongji Chubanshe [Statistical Publishing House], 1983), p. 147. 32ibid., p. 148; Di1i Yanjiusuo, Zhongguo Nongye Dil! Zonglun, pp. 77-79. - 35 - Since the early 19805, however, cultivation of collective lands is no longer a communal responsibility but has been delegated to several specialized households. Normally, it is the particularly skilled farmer who is entrusted with responsibility for farming collective lands .. But in relatively advanced communes or in suburban areas, non-agricultural activities with higher income earning potentíal attract the most able workers. -Aside from collective lands, individual farm familres maintain private plots of normally 0.03-0.05 hectares which are used primarily for famíly productíon of food items, especially vegetables and livestock products (and consequently fodder for the latter). Although no estimates are available for.cassava cultivation on such lands, the importance of cassava as a swíne feed, the considerable importance of swine in the livestock economy of South China and the dominance of family-owned and managed swine within the swine husbandry sector, suggest that prívate plot cultívation of cassava in South China is not trivial. In addition to formally established private plots assigned to each family, there appears to be cultivation of cassava on an even more fragmentary basis: on narrow stfips adjaeent to roads and fields, on steep hillsides and other areas not formally counted among eultivated lands and illegally in eeonomic forests, reclamation areas and other lands managed by the state. The latter may be distinguished, however, from planned eultivation on sueh lands by the State Farm and Reclamation Bureau. While land is being eleared and - 36 - reclaimed, cassava is often gfown as an intermediate crop for a few years until it is discontinued when field transformation progre ss allows cultivation of the principal crop.33 Finally, cassava is planted as a field crap on state farms. There its cultivation is especially extensive and is characterized by high standards of agronomy and abundant applicatian of modern input!, particularly fertilizers. Visitors interested in cassava are often brought to state farms to view extens~ve cultivation and high yields, but state farm plantings remain a small proportion of total cassava area. Cultivated area on state farms in Guangdong varied between only 60 and 64 thousand hectares from 1981 to 1984, and remained at 20 thousand hectares in Guangxi. In 1984, state farm sown area in Guangdong was only 86,900 hectares or less than 1.8 percent of th~ provincial total, of which 72,200 hectares were planted with cereals, beans, sweet and white potatoes, oilcrops and sugarcane, leaving a residual of 14,700 hectares which could have be en planted with cassava, vegetables, green manure, other fodder crops or other southern industrial crops such as sisal hemp. In Guangxi, state farm sown are a was only 17 ,400 hectares or less than 0.5 percent of the regional total of which the residual category including cassava 33Bruce Stone, "An Analysis of Chinesa Data on Root and Tuber Crop Production," The China Quarterly, September 1984, p. 621; Liang Guangshang (ed.), Mushu Zaipei yu Livong, p. 36; Bruce Stone, "An Examination of Economic Data on Chinese Cassava Production, Utilization and Trade". - 37 - comprises but 3,300 hectares. 34 Thus private and collective plantings dominate cassava area in China. Available international data on cassava utilization in China is unreliable, but it is clear that animal (especially swine, but also cattle, fish and silkworm) feed is associated with each of the cassava production systems. Exports and starch production as well as less traditional industrial and processing uses are assocíated with co 11 ect i ve product i~D and the state farms, wh-i 1e di rect human consumption is associated with private production and the collective sector in poorer areas. Machine cultivation is associated with a portion of the extensive plantings between 100 m and 300 m above sea level. Between 300 m and 1,000 m, cassava is grown in rotation with dryland crops as far as 30oN. Most cassava in China is unirrigated, but the climate provides adequate moisture in most years and locations. This is especially true in the south where fall-planted cassava is common. 35 Cassava is cultivated year round in South China, with the principal plantings concentrated in spring and fall. The planting material may be either fresh1y cut stakes or stored material. Storage is practiced by cutting long stakes which may either be 1eft in the sun in bundles or placed under trees. Cuttings are fairly 34China Agricultural Yearbook Editorial Board, China Agricultural Yearbook 1985 (Beijing: Agricultural Publishing House, 1986), pp. 114 and185-186. 35liang Guangshang led.), Mushu Zaipei yu liyong, p. 36. short (10-15 cm) with mínimal selection. Plantíng is fairly deep (up to 10 cm and horizontal). Germination varies considerably by location but is frequent1y very poor and strands are not uniformo Land preparation is generally acceptable and is done manually, by draft animal or tractor-dl'awn impl ements. 36 Spring cassava (e.g. in the Guangzhou area) is typically planted between January and March and harvested in the fall, after at least 8 months especially from October, although for fodder purpases, cuttings may be taken continuously over an extended periad of time. The spring and surnmer seasons considerably aid leaf and stem growth of spring-planted cassava and fal1 arrives optimally for starch formation. Yields of spring-planted cassava tend to be large, but are less reliable since typhoons in fall occasionally cause damage. Furthermore, low temperaturas in spring extend the budding and sprouting period and thus the risk of insect damage. But spring- plantad cassava fits well into South Chinase intercropplng and l'otation systerns, facilitating the achievernent of as many as three crops par year, including one of cassava. 37 Fall- and winter-planted cassava is common in the most tl'opical areas with harvests starting tha following fallo Tha peak period for both planting and harvesting is September to November. Fall-plantad 36Cock and Kawano, "Cassava in China," p. 7. 37The discussion of spring- and fal1-planted cassava is primarily from material appearlng in Liang Guangshang (ed.), Mushu Zaipei yu Liyong, pp. 10-11 and 33-34. - 39 - cassava ís practicable from around Gaozhou County (21 0 56'N, Zhanjiang Prefecture, Guangdong Province) s~uth, where temperatures average about 22.7°C annually and the lowest average January temperatures exceed 15°C. These areas also enjoy 1304-1718 mm of rainfall per year and 1941-2455 hours of sunlight, higher than more northerly regions, especially during the winter, thereby províding more hospítable conditions for fall planting. Of course, fall-planted and spring-planted cassava are not mutually exclusive. Qíjing Brigade, for example, in Díanbai County (within the coastal zone lying ~long the South China Sea well to the south of Gaozhou), planted 25 thousand hectares of cassava in 1972, approximately one-third fall­ planted, two-thirds spring-planted. A principal advantage of fall-planted cassava is the potential for avoiding typhoon damage. This ís particularly ímportant on the Leizhou Península and Hainan lsland. Insect damage to the sprouts is also lower sínce cricket populations decline rapidly in fall and the sprouting period is collapsed, with sprouts and roots beginning within a week after planting. Fall-planted cassava can be more conveniently linked with sericulture, sínce leaves are provided more opportunely, without influencing root yield. With the longer season, cassava planted in fall facilitates fuller utilization of production capacity in local .starch factories, and is convenient for on-farm livestocK development. The principal drawbacks are the slower winter growth and the inconvenience of the longer season for rotation and multiple cropping. Thus even in the far south, if the cropping intensity is hi h, cassava is apt to be lanted in spring. Hith virtua11y a11 cassava north of 22°N and an important portion of the remainder p1anted in spring, the m~jority of cassava in China is 1ike1y to be spring-planted. The Chinese are we1l aware of the necessity of rotation and intercroppin~ for continued cassava cultivation. They estímate that yields decline by 20-30 percent in a second consecutive year of cassava cultivation, and by 30-40 percent for three consecutive years. 38 The CIAT de1egatíon noted, howeve~, that cassava is grown as a monocrop in some areas. 39 South Chinese rotation systems are complex and varied; those including cassava are no exception. Figure A presents notable 2-year through 6-year rotatian systems for cassava and other dryland foad crops. In newly reclaimed areas, cassava is often grown for one or two years among jade eassia (Chinese cinnamon), mountain apricot, bamboo, tong oil, tea oi1, rubber trees, or in other economic forests. Chinese literature points out the importance of rotation of eassava with green manure erops in eeonomic forests to avoid erosiono Cassava is norma1ly the principal erop in a smal1 number of exceeding1y poor loealities and a very few state farms. As Table 5 and 6 indieate, the most important erop in South China is unquestionab1y paddy rice comprising 63 percent of sown area in Guangdong in 1984 and 59 percent in Guangxi. Paddy fie1ds occupy 63 38Liang Guangshang (ed.), Mushu Zaipei yu Liyong, p. 40. 39Cock and Kawano, ·Cassava in China,· p. 8. - 41 - Figure A. Cassava Rotation $ystems in China 2-year systems cassava - uplarid rice, $weet potatoes cassava - peanuts, sweet potatoes spring peanuts, fall-planted cassava - fall harvested cassava, spring soybeans 3-year systems cassava - sugar cane - sugar can e cassava - peanuts, wheat - upland rice, sweet potatoes 4-year systems cassava- mung beans, sweet pota toes - sugar cane - sugar cane 5-year systems peanuts, wheat - upland rice, sugar cane - sugar cane­ sugar cane 6-year system cassava - sugar can e - sugar cane - soybeans, sweet potatoes - upland rice, radishes - peanuts, sweet potatoes Notes and Sources Liang Guangshang (ed.), Mushu Zaipei yu Liyong, p. 40. In Cock and Kawano, Cassava in Asia, p. 8, the authors noted that cassava was often grown with legume craps, predaminantly peanuts. - 42 - pereent of cultivated land in Guangxi and are similarly dominant in Guangdong. Sweet pota toes are second in order of planted area in Guangdong and, combined with white potatoes, totalled 10 pereent of sown area. Peanuts (6 percent) and sugar cane (5 percent) rank third and fourth, probably followed by eassava at around 3 percent. Soybeans, maize, bast fibers and tobacco are also grown, and until its de-emphasis in recent years, wheat area exeeeded cassava plantings. In Guangxi, maize is seeond at 11 percent of sown area, followed by soybeans and sweet potatoes (5 percent eaeh), s¡¿gar cane and peanuts (3.5 percent each) and green manure crops as a group (2.5 percent). Cassava at 2.1 percent is slightly below vegetables and melons as a group. When cassava area peaked in 1980, its share was 4.3 percent, ranking fifth behind rice, maize, soybeans and sweet potatoes and higher than al1 economic crops.40 Yields Most available information on cassava yields was provided in the section on production trends and distribution. In that section it was suggested that the considerable increase in average yields during the latter 19605 (Table 2) was due to varietal improvement and to some extent, improvement in cultural practices, while yield growth since the late 1970s has be en principal1y the result of increased fertilizer application to cassava, complemented by some improvement in varieties and cultivation techniQues. Mean cassava yields throughout China (- 8.6 tons/ha in 1980) approximate the average for 40Ta ble 5 and 6; China Agricultural Yearbook 1985, pp. 114-126; and Dili Yanjiusuo, Zhongguo Nongye Di1; Zonglun, pp. 77-79. - 43 - the rest of the world, but are somewhat higher than mean yields in the remainder ef Asia. Mean yields in Guangxi (13.1 tons/ha. 1981-84 average), however, are somewhat higher than the international average, and the highest yields from field cultivation in China (average 20-25 tons/ha with a maximum ~f 30 tons/ha.or more) are comparable to the very highest yields in the world. 41 But Chinese cassava is also grown on poor soils with no fertilizer or irrigation where average yields have been characterized in the 3 to 8 ton range. The average figures cited aboye suggest that those poor conditions are more typical of Chinese cassava cultivation than the state farm or Pearl River Delta private and cooperative farming experience. However, survey results suggest that even on poor soils without irrigation, fertilizer application can increase yields en both research stations and operating farms by an average of at least 6 tons per hectare. Yield differences among farms are due not on1y to differences in so11 fertility, climatic conditions, adopted varieties and applied fertilizers, but to substantial differences in management as wel1. Farmers in sorne areas use unselected planting materials giving very poor stands and low yields. On private plots, management varies more than on col1ective lands within a single vicinity, but the level of agronomy is often fairly high. 42 41 ibid ., p. 1 and 8; Delphi Survey responses; and correspondence from James H. Cock, June 24, 1983; Table 2. 42Cock and Kawano, "Cassava in China"; correspondence from James Cock, June 24, 1983. - 44 - Among the responses of three Chinese cassava breeders surveyed, low yield potential of existing varieties and unavailability of fertilizers were both listad by each respondent as important constraints on farmers' yields. But the survey results also suggest that output marketing prob1ems, storage and processing difficu1ties and general lack of production incentives may restrict application of labor and fertilizers to cassava in sorne areas. 43 A1though there is consiaerable variation in the quality of tultivated varieties, China has several popular varieties, such as South China 205, providing reasonably high and stable yields. It is the provisional conc1usion of one international breeder that, like Thailand in the recent past and Malaysia currently, rigidly selected CIAT clones cou1d outyíe1d the best Chinese cultivars only slightly. Thís contrasts with Indonesia and the Philippínes where the best local varieties are more easily dominated. 44 Poor fertilizer response and inadequate extension were listed as a secondary constraint on yields as was inadequate rnoisture in sorne areas. The 1982 CIAT delegation noted that fertilizer applications were not generally linked to soi1 analyses or recommendations made on the basis of experimental results. Each of the surveyed breeders appeared to agree that pests and diseases were re1atively unimportant 430e1phi Survey results 44Kazuo Kawano, "Tri p Report to Chi na (18-24 January, 1986)," unpublished trip report provided in carrespondence fram Kawano, April 14, 1986. - 45 - in limiting cassava yields. The 1982 CIAT delegation a150 found that although pests and diseases were not chemically controlled, they appeared to be of very low incidence and harvest losses from 5uch sources were concluded to be mínimal. The most commonly observed disease was Cercospora leaf spots and during the dry months Tetranychus mítes are reported to be a problem. 45 Costs of production and labor utilization The 1982-CIAT delegation was told that labor use varied from 100 man days par hectare with mechanical land preparation to 270 days without machines, and total production costs were estimated at $550 US per hectare. 170 days may be somewhat excessive for manual land preparation, but althou%h the total of 270 days per hectare is higher than in sorne Asian countries it is not unprecedented. The total cost figures are likely to have come directly fram the product~on accounts of one or more Guangdong state farms where workers are paid set wages, or from a sma11 sub-group of more prosperous cassava-growing col1ectives which happened to ha ve kept good records and where yields are high. Most of the implied cost per man-day of around $2 US would be labor. A project prospectus for an agricultural credit application t6 the World Bank involving cassava cultivation implied a return to labor of $1.25 US per day. Much of the labor involved, 45Cock and Kawano, ·Cassava in China· . 7. - 46 - especially where cassava is fertilized, is for hand-weeding since herbicides are not used. 46 Much of the non-labor costs on state farms would consist of fertilizer application. The highest per hectare application rates encountered by the CIAr delegation in 1982 were 20 tons of organic manures, 375 kilograms of superphosphate (45-68 kg. of P205) and 150 kilograms of muriate of potash (37.5 kg. of K20).47 Such ratas are likely to have existed~only on state farms with plentiful access to fertilizers and/or few alternativa uses. Implied per hectare reta;l value of this level of manufactured fertilizer use alon~ would have totalled $ USo 48 On collective lands with plentiful access to fertilizers, use of manufactured products is less lavish but organic manure use with associated high labor requirements is very substantial. In Fucheng Commune of Dongguan County on the Pearl River Delta, average yields of 21-22.5 tons per hectare on 400 hectares of cassava were achieved with 225 kilograms of ammonium sulfate per hectare. But in addition, three organic manure applications were undertaken involvlng total per hectare use of 3 tons of swine and cattle manure, 3-4.5 tons of human night soi1, and 15 tons of green manure (primarily legumes) mixed with 22.5 too s of soi1. On the Huashan State Farm in Lingshan County, Guangx; per 46Ibid., pp. 7-8; correspondence from John Lynam, CIAT Cassava Program, December 22, 1983; Stone, HAn Examination of Economic Data on Chinese Cassava Production, Utilization and Trade," pp. 6-9. 47Cock and Kawano, "Cassava in China,· p. 7. 48 - 47 - hectare applications of 255 kilograms of ammonium sulphate and 15 tons of organic manure yielding 19.62 tons per hectare were estimated to provide 141 kilograms of nitrogen, 79 ki10grams of phosphoric acíd and 180 kilograms of nítrogen. 49 One of the 1986 Chinese survey respondents provided a combined per hectare estímate of farmer fertilizer use on poor soi1 cassava lands in Guangdong of 150 kilograms, associated with average yields of only 5 tons per hectare, while another respondent, based on Hainan lsland (Guangdong), implied that no manufactured fertilízers were used on cassava by farmers regardless of so;l condítions. 50 -It is very unlikely that much ferti1izer has bean app1ied to cassava on distant co11ectives and private p10ts. This is due to low farmgate cassava prices, a weak cassava market in many areas (see below) and to the higher prices and difficult access associated with fertilizar purchase unless such purchase is linked to sales to government procurementorganizations of farm goods ;n particular sta te demando Prívate plot production of cassava employing household labor and without manufactured fertilizer use, coulct be conducted for purposes of home consumption and hog feed at very low implied return to labor. However, with the low yields associated with most production, such returns could be well under $1 US per day, and may have been sustainable only as a function of Chinese labor market 49Liang Guangshang (ed.), Mushu Zaipei yu Liyong, p. 86. 50Delphi Survey responses. - 48 - restríctions. With increasíng liberalization of economíc activities in the 1980s, labor opportunity costs have risen substantially in suburban and wealthier rural farm areas. As export epportunities have declined, these healthy ecenomic movements have undoubtedly worked against cassava cultivation in such areas. Opportunity cests would be less affected in poorer and more distant farm areas, but the state's declining marketing role ís less apt to be vigorously replaced by prívate market development in such areas. Technology develoDment Publicatíon of Liang Tingdong's Zhong Mufanshu Fa [Cassava Planting Methods] in 1900 was a benchmark in the initiation of a formal process of cassava technology improvement in China, which could span time and space. As indicated in the first section, cassava spread to Fujian and Taiwan in the 19205, roughly 100 years after its first known cultivation in neighboring Guangdong. Introduction in Hunan and Jiangxi in the early 1940s may have be en the first example of deliberate trans-provincial dissemination by Chinese scientific institutions. The Peoples ~epublic agricultural science establishment gave attention to cassava as a bulky, relatively dreught-resistant crop which could be grOl'm on poor sol1s and still provide growth in available calories per unit of farmland, with some advantages in yield stability, Alternatively it could al$o furnish raw materials for industry. This orientation toward bulky cheaper fooo items and industrial crops was wel1 11ithin a tradition established early in the - 49 - history of most socialist governments and still continues to distinguish the pattern of food production and availability, although to a decreasing extent over .time, in the Soviet Union, Eastern European countries and North Korea as well as in China, Vietnam and other socialist nations more suited to cassava production. 51 Although dissemination of cassava was emphasized throughout the 1950s, broadening cultivation in the two southern provinces, and injtiating it in Zhejiang and Jiangxi, cassava researc~ began to show results in the late 1950s. Between 1957 and 1962, the Agricultural Science Department's Grain Crops Laboratory of the South China Academy of Agricultural Science in Guangzhou (23°S'N) selected 10 varieties from a pool of 30 for dissemination, at least six of which have been extensively cultivated, including Zajiao [Hybrid] no. 4 and 'ftnni Xiye [Indonesian thin leafJ, exhibiting 11 pereent and 23 percent yield improvements over widely planted Hongweizhong [Red Tail Variety], and Mianbao Mushu [Bread Cassava], Zajiao no. 1 and Nanwan Mushu [South Bay Cassava], yielding 70-86 percent of Hongweizhong, but exhibiting other desirable characteristies su eh as superior edibility, higher starch rates and/or yield stability. Although breeding objectives for cassava have broadened eonsiderably since the 19505, higher root yields and improved edibility remain as central 51Shigeru Ishikawa, "China's Food and Agrieulture: A Turning Point," Food Poliey 2 (May 1977), p. 93; Bruce Stone "China's 1985 Foodgrain Production Target: lssues and Prospects," in Anthony M. Tang and Bruce Stone, Food Production in the Peooles Republic of China. Res ar 5 (. Table 9. Cassava Root Nutritional Content (percent) Water Starch Soluble Variet~ Content Rate Sugar Protein Fat Fi Mianbao Mushu 101 [Bread Cassava 101] 64.0 29.2 1.29 0.61 0.20 O Naomi Mushu 102 [Glutinous Rice Cassava 102] 63.0 29.0 2.15 0.81 0.20 o Ma 1a i huang 103 [Malay Yellow 103] 63.2 31.3 1.46 1.09 0.15 O Wenchang Hongxin 104 [Wenchang Red Heart 104] 62.4 30.5 1.26 1. 55 0.21 O Maoming Baíxin 105 [Luxuriant & famous White Heart 105]60.6 32.6 _ 1. 54 1.04 0.13 O Hainan Hongxin 211 [Hainan(Island) Red Heart 211] 67.0 26.8 1.85 0.50 0.21 O e;¡ Huguang pñingjing 210 [Huguang Green StemJ 57.6 36.8 1.23 1.40 1.14 O Hongweizhong 201 [Red tai1 variety 201J 71.0 23.7 2.22 0.59 0.32 O IInni Xiye 202 [Indonesian Thin Leaf 202] 65.4 27.7 2.03 0.73 0.13 O 1inni Daye 203 [Indonesian Big leaf 203J 66.0 28.2 1.69 0.92 0.14 O Nanyang Qingpi 204 [South seas Green skin 204J 66.0 28.8 2.87 0.60 0.17 O Nanwan Mushu 205 [South Bay Cassava 205J 66.0 28.1 1.85 1.13 0.17 o. Huanan 206 [South China 206] 59.0 35.6 1.93 0.99 0.16 O. Huanan 207 [South China 207J 64.8 29.6 1.00 0.88 0.12 O. Zijingzhong 208 [Purple stem variety 208J 70.1 21. 5 3.43 0.47 0.19 O. Fanyu Zijing 209 [Fanyu {County)Purple Stem 209J 61.8 23.0 2.02 0.86 0.15 o. Average of al] varieties 64.2 28.8 1.86 0.89 0.17 o. - 51 - Sources: Liang Guangshang (ed.), Mushu Zaipei yu Liyong [Cassava Cultivation and Use] Guangzhou: Guangdong Kezhi Chubanshe [Guangdong Scientific and Technical Publishing House], 1981), p.l08. foei of the Chinese breeding program. 52 South China 201 is also known as Hongweizhong or Dongguan Hongwei [Dongguan Red Tail]. A high yielding cultivar with high cyanide content, it is the most popular variety for flour production. Cultivated on plains, hilly tracts and mountainous uplands, this variety covers 70-80 percent of cassava area in many Guangdong and Guangxi Prefectures. It is also experimentally cultivated in the Yangzi Val ley. South China 202 orYlnni Xiye was introduced from Indonesia in 1956 by the South China Agricultural Science Department in Guangzhou. It typically outyields Hongwei by a sma11 margin, but has the highest cyanide content of popular varieties and is thus also used in processing industries, primarily for flour and starch production. Plantings are concentrated on the Aoxi State Farms. There has also been successful experimental cultivation in Nanjing. South China 205 or Nanwanmushu was the shortest of the sixteen leading cultivars testad and is famous for withstanding the August 17 typhoon in 1963. lt combines"yield stabillty with hlgh potentlal, 52Liang Guangshang (ed.), Mushu Zaipel yu Liyong, pp. 10 and 77. Much of the succeeding discussion on varieties and Institutions is based on po. 77-80 and Table 9, with a few additions from Cock and Ka\~al'!o, "Cassava in Asia." - 52 - and ís good for flour and especíally starch productíon where it significantly outperforms other popular varíeties. As Table 9 G indicates, Huguang ~ingjing [Huguang Green Stem] or South China 210 and South China 206 have by far the highest starch rates per unit weight, but Nanwanmushu's respectable rate coupled with higher yield potential make it a clear leader in starch per unit of harvested al'ea. Following Nanwanmushu, South China 206, 207, andYInni Xiye feature the highest starch content per unit area. South China 205 is an internationally recognized cultivar with similar characteristics to those of the Vassourinha variety of Brazíl and the Phjlippines. The greatest area of Nanwanmushu concentration is Zhongshan, Dongguan and other counties in the Pearl River Delta, but it ;s planted widely throughout Guangdong. South China 101 or Mianbao Mushu is also known as Malaihong [Malay Red] sinee it was introduced onto rubber plantations in Dan Xian from Malaysia in 1912. The variety combines yiald stability with iow cyanide content and reasonably high yield potential, and is racognized as China's best tasting cultivar. Plantings are concentrated on Hainan Island, especially in Dan Xian, Wenchang, and Baoting Counties, but bread cassava is also grown in most areas of Guangdong, and has be en experimentally cultivated in Hebei Provinee, farther north than any other variety (39°20'N). ¡ts eharacteristics are relatively similar to those of Aipin Valencia of Soúthaast Asia. South China 104 or Wanehang Hongxin [Wenchang Red Heart] is tha highest yielding variety among tha better tasting (sweeter) cultivars. Jt has tha highest protein content of the 16 leading - 53 - varieties, also features low cyanide concentrations, reportedly outyields Mianbao Mushu by 22 percent, but is not typically preferred to the latter for direct consumption. South China, 104 is planted predominantly in Wenchang and Qiongshan Counties on Hainan Island with little cultivation elsewhere. Among other palatable varieties, Maoming Baixin [Maoming White Heart] or South China 105 from Maoming Municipal Area near Guangdong's Leizhou Peninsula, and Nuomi Mushu [polished glutinous rice cassava] or Sauth China 102 are worthy af mention. Both outyield Mianbaa Mushu by 10-11 percent, with substantially greater superiority in more northern areas. Both are sweet, and low in cyanide content, with South China 102 lowest of the sixteen prominent varieties. A variety known as 6068 is also famaus far its excellent eating qualities and is planted on around 10,000 hectares despite its modest yields. In sum, the South China Tropical Crops Research Academy concentrated not only on selection and dissemination of cultivars featuring higher and more stable root yields and improved edibility, but has focused breeding attention in combiníng those characteristics, and initiated research on starch contento By focusíng on faster, as opposed to strictly higher root yields, the Academy a1so brought to cassava breeding in this early period, the beginnings of a quintessential1y Chinese orientatíon: . breeding to fít rotational patterns and multiple cropping sequences. With the catastrophic famines of 1960-61 centered in North China and the Yangzi Valley, efforts to spread cassava cultivation northward intensified considerably. The focal institution in tnis effort was the Zhejiang Province Sub-tropical Crops Institute in Pingyang (27°38'N). Between 1962 and 1964, the institute introduced 31 varieties from Guangdong, Guangxi and Fujian including Hongwei, Nanwanmushu, Inni Daye, Shibei~ingjing [stone tablet green stem] and Zajiao nos. 1-6. But as Table 10 indicates, there has been experimental cultivation much further north, although the South China Troplcal Crops Research Academy has indicated that good growth and yields are consistently obtained only up to around 26°N, which cuts across southern Hunan, Guizhou, Jiangxi and Fujian. Aside from the above-mentioned institutions, sorne cassava- related research is reportedly conducted in each of the provinces within which cassava has be en introduced. In South China, other relevant institutions are the Guangxi Province Asian Tropical Craps Research lnstitute in Nanning, the South China Crop Research Institute and the South China Institute of Botany within the Chinese Academy of Sciences, the Institute of Drought-Resistant Grains and the Upland Grains Department in the Guangdong Agricultural Science Academy, and the South China Agricultural Callege, all in Guangzhou. However cassava research is not reputad to be a significant current focus of any of the Guangzhou institutions. Cassava research and deve10pment in China is increasingly shifting its focus from the original narrowly defined 90a15 of - 55 - Table 10. Results of Cassava's North Migratlon Cultivation Experíments Planting Harvest Total Growing Fresh Root perimenting Unít Location Varíety Date Date Days Yield (N latitude) (tons/ha. ) rthwest Agricul­ ral Scíence ademy A, B Apr.25 Nov.25 216 33.0 bei Dashahu Farm A,B,D Apr.21 Nov.22 216 18.75-30.0 hui Provínce ops Institute 31°53' B Apr .12 Nov.3 206 20.325 njing Botanical stitute A,B,C Apr .15 Nov.5 205 23.25-24.45 ina Root and ber Illstitute A,B May 6 Oct.24 172 37.5-45.0 aanxi Province ains Crops Inst. A,B May 7 Oct.23 170 5.775-17.775 andong Province ops· Institute A Apr.15 Oct.24 193 22.5 da (Dalian) A,B May 6 Oct.23 171 12.75-19.5 1 Farm bei Province restry Science stitute A,B Apr.21 Oct.24 187 37.5-45.0 tes: A= Naomimushu [Glutinous Rice Cassava] B= Mianbaomushu [Bread Cassava] C= Inni Xiye [Indonesian Thin Leaf] D:: Malaihuang [Malay Yellow] Jrces:Liang Guangshang (ed.), Mushu Zalpei yu Liyong [Cassava Cultivation and Use] illgzhou: Guangdong Kezhi Chubanshe [Guangdong Scientific and Technical Publishing Jse], 1981), p. 26. - 56 - improving yield and edibility. The main improvement efforts still inelude edibility, but a150 emphasize cultivation teehniques, especial1y eassava's relation to other erops in various systems, and the eombined development of cassava and non-erop rural aetivities. Breeding objectives a1so inelude early planting, early ripening and rapid maturity goa1s, as we11 as disease resistance, high yields, and high starch and protein content. 53 Research and development 90a1s related to cultivation techniques feature improvement in rotafion synergies, seasona1 cultivation, intercropping, and achievement of two or even three ripenings per year. Bean erop and eassava rotations and intercropping are of particular interest as techniques for developing soi1 strength. The 1982 CIAT de1egation ob5erved that cassava was often intereropped with graio 1egumes in more intensive1y cu1tivated areas and estimated that yields of both crops were probably reduced by only 15-30 percent resulting in re1atively efficieot land use with good soi1 conservation properties. 54 Sinee 1979, non-crop agriculture has been emphasized in China, partial1y correcting for the substantia1 pre-1979 stress on food crops, especially stap1es. Consequent1y a recent goal for cassava development has been to integrate cassava with forestry, animal husbandry, sericulture, aquaculture and rural side1ines for 53Liang Guangshang led.), Mushu Zaipei yu Liyong, p. 10. 54Ibid.; correspondenc! from James H. Cock, Cassava Program Director, CIAT, June 24, 1983. - 57 - cooperative production. Investigation of additional and even novel industrial uses is also of increasing interest. Survey respondents among Chinese cassava breeders and agronomists 55 appeared optimistic about the potential for growth in farmers' yields during the next 4 and 14 years. Respondents were instructed to base their assessments on existing varieties and those currently under development, but their estimates differed considerab1y. They were also optimistic about the prospects for increasing-that potential via a doub}ing of research expenditures related to cassava, with the most conservative assessments provided by the representative of the institution where most research on cassava ís conducted. In his view, farmers' yields on poor soils could increase from currently 3-6 tons per hectare to 4-8 tons by 1990 and 5-9 tons by 2000 or 5-10 tons and 6-12 tons respectively with a doubling of research expenditures. With good soi1 and climatic condítions, farmers' yields could increase from currently 15-30 tons/hectare with fertilizer, to 18-35 tons by 1990 and 20-40 tons by 2000 or 25-35 tons and 35-45 tons with a doubling of research resources. It is clear that yields can improve, especial1y in Guangdong, vía greater access to manufactured fertilizers, analysís and extension related to its optimal use, and to proper selection of planting materials. 'ertilizer pricing, distribution and analytic systems are undergoing considerable structural change in China. 55Delphi Survey responses. - 58 - Preper resolution of remaining and newly emerging difficulties will be instrumental in achieving yield progress through growth in fertilizer use. 56 It also appears that there may be some limited potential exploit~ble with further international exch~nge of genetic materials. 57 State farms are technological leaders in cassava cultivation, though not for most staple crops, and careful selection of planting materials and quest fer improved cultivars are evident on state farms. Yield-progress on several state-farms in recent years has allowed continued profitability of cassava cultivation despite declining prices. This means that new improved varieties can move rapidly into full scale production in China. What may be called for are institutional links which can bring state farm developments ínto the prívate and collectíve economy more expeditiously. A new variety must undergo regional testíng for three years. The results are presented to the provincial seed commission which may then recommend the variety to seed production companles for multiplication. Work on intercropping and rotational systems is something Chinese researchers do particularly well and is llkely to lead to sorne further improvements. Sorne of these may not immediately 56For detalls see Bruca Stone, "Chlnese Fertilizar Applicatlon in the 1980s and 1990s: Issues of Growth, Balance, Allocation, Efficiency and Response" in U.S. Congress Joint Economic Committee (eds.), China's Economy Looks Toward the Year 2000, vol. 1: The Four Modernizations (Washington, 6.c.:· U.S. Government Printing Office, 1986), pp. 453-496. 57Cock and Kawano, ·Cassava in China"; Kawano, "Trip Report to China (18-26 January, 1986). - 59 - increase cassava yields per se but may improve theattractiveness of planting cassava and thus arrest its decline in area. What is singularly missing for cassava, as well as for many other crops, is socio-economic research in cassava areas, particularly poorer ones. Lack of agro-economic data and analysis for assessing constraints l1miting farmers' yields is recognized by the South China Tropical Crops Academy.58 Finally, with the reduction in export opportunities and the cu~tailed government role in marketing, development o~demand and market institutions are of particular importance for continued expansion of cassava production and use. These issues will be undertaken in the following sections. MARKETS AND DEMAND A synthesis of production and utilization As indicated above, production statistics for cassava in China are highly fragmentary, except for Guangxi Zhuang Autonomous Region for which data are complete, though even for Guangxi, questions of reliability and comparability remain. Utilization data, however, are almost wholly unavailable, with the exception of the international trade data compiled from European Community Analytic Tables for Foreign Trade appearing in Table 3. Government procurement data for cassava assuredly exist, but have not been made available in Chinese 580e1phi SUI'vey response from Tan Xuecheng, breeder. - 60 - statistical compendia on marketing and trade. Production data from cassava flour and starch factoríes as well as from other industrial processors are certainly generated, but are not of sufficient importance to appear among national statistical series in the relatlvely detalled Guangdong Province Statistical Yearbooks, and the Guangxi Economic Yearbook 1985, although the latter contains a single column of discussion of the starch market in which cassava is mentioned. As a regionally concentrated crop, cassava has not turned Up among pUblished results from national farm surveys. Even liang Guangshang'~ cassava-specific publi~ation, Mushu Zaipei yu Liyong [Cassava Cultivation and Use], provides "ot a single statistic on aggregate utilization. In the past, it has been clear that FAO estimates of cassava use were all based on constant percentages of estimated production. 59 For example, the FAO Supply Utilization Accounts Tape 1981 evidently incorporated the following percentages: feed use (25 percent), waste (5 percent), foad use (67 percent), processing (3 percent), use for tapioca (70 percent of processing), starch use (30 percent of processing).60 Since the production series was mechanically generated from virtually no statistical base, the utilization series were inevitably unreliable, even if the percentage shares were roughly correcto Conversely, regardless of the accuracy of the production estimates, the utilizatíon shares have assuredly not been 59Bruce Stone, "An Examination of Economic Data on Chinese eassava Production, Utilization and Trade,· pp. 13-22. 60Food and Agriculture Organization of the United Natíons,· ·Supply Utilization Accounts Tape 1981," Rome, 1982. - 61 - constant over time, with feed and processing use incl'easing in importance, at the expense of direct human consumption. Moreover, shares for feed and processing would exceed the shares implied by the 1981 Utilization Tapes even for the 1960s. 61 As an examination of Tables.ll and 12 will reveal, FAO utilization series for China are now generated in a more complicated fashion, but historical production, are a and yield figures are identical to those appearing on the older tapes. Aside from the international trade series which relates well to, and is probably based on the EC Analytic Tables for Foreign Trade, FAO series are still generated from an extremely weak statistical basis which probably consists of no more than the partner-country trade data and the single production figure c;rca 1980, provided to the 1982 CIAT delegation. In these recent FAO series, such as ·Supply Utilization Accounts Tape 1984," released at the end of 1985, unprocessed feed is set at 10 percent throughout the 1961-83 period and waste is dropped from 5 percent on previous tapes to 3 percent for the entire periodo Direct food consumption estimates have become trended values declining from 72.0 percent of production in 1962 to 67.0 percent in 1979. (Table 12). Processed uses have becorne monotically non-decreasing trended values beginning somewhat arb•i trarily at 15.0 percent in 1962 and rising to 20.0 percent in 1979, of which dried cassava (chips and 6IStone , "An Exarnination of Economic Data on Chinese Cassava." This paper was provided to both ClAT and the FAO Statistical Division's Basic Data Unit in 1983 and provided part of the basis for subseouent adiustments. - 62 - Table 11. FAO Estimates of Chínese Cassava Production, Area, and Yield, 1961-1984 Harvested Area Production Yield 1982 Ta[!e 1984 Ta[!e 1982 Ta~e 1984 TaEe 1982 Taee 1984 Taee (lOaD hectares) (1000 metric tons) (tans per hectare) 1961 80 940 11.750 1962 85 1000 11. 765 1963 85 950 11.176 1964 90 1000 11.111 1965 90 1100 12.222 1966 95 95 1100 1100 11.579 11. 579 1967 100 100 1200 1200 12.000 12.000 1968 120 120 1400 1400 11. 667 11. 667 1969 130 130 1500 1500 11.538 11. 538 1970 140 140 1600 1600 11. 429 11.429 1971 150 150 - 1800 1800 - 12.000 12.000 1972 160 160 1900 1900 11. 875 11.875 1973 170 170 2000 2000 11.765 11. 765 1974 170 170 2000 2000 11. 765 11.765 1975 180 180 2100 2100 11. 667 11. 667 1976 180 180 2200 2200 12.222 12.222 1977 190 190 2200 2200 11.579 11. 579 1978 200 200 2300 2300 11. 500 11. 500 1979 200 200 2500 2500 .12.500 12.500 1980 226 226 3000 3300 13.274 14.602 1981 236 230 3120 3500 13.232 15.217 1982 235 3600 15.319 1983 240 3800 15.833 1984 Source: FAD, "Supply Utilization Accounts Tape, 1981," Rome, 1982; FAO, "Supply Utilizatlon Accounts Tape, 1984," Rome, 1985. - 63 - Table 12. FAO Estimates of Chinese Cassava Production and Use, 1961-1983 Production of which: Feed Waste Food Processed of which input to: Chi ~s & Pe 11 ets Ta~ioca Starch ¡1000 tons) 1961 940 94 28 668 140 90 20 30 1962 1000 100 30 720 150 100 20 30 1963 950 95 28 66(1 160 110 20 30 1964 1000 100 30 699 171 120 21 30 1965 1100 110 33 756 201 150 21 30 1966 1100 110 33 740 217 160 22 35 1967 1200 120 36 807 237 180 22 35 1968 1400 140 42 959 259 200 24 35 1969 1500 150 45 1014 291 230 26 35 1970 1600 160 48 1099 293 230 28 35 1971 1800 180 54 1246 320 250 30 40 1972 1900 190 57 1330 _323 250 33 40 1973 2000 200 60 1384 356 280 36 40 1974 2000 200 60 1380 360 280 40 40 1975 2100 210 63 1467 360 280 40 40 1976 2200 220 66 1519 395 300 50 45 1977 2200 220 66 1519 395 300 50 45 1978 2300 230 69 1606 395 300 50 45 1979 2500 250 75 1675 500 400 55 45 1980 3300 330 99 1466 1405 1300 60 45 1981 3500 350 105 1545 1500 2000 65 45 1982 3600 360 108 1512 1620 1500 75 45 1983 3800 380' 114 1606 1700 1700 78 45 Notes and Sources: FAO, "Supply Utilization Accounts Tape, 1984," Rome, 1985. To reach quantities of processed products, extraction rates of 35 percent for chips and pellets (dried cassava), 22 percent for tapioca, and 18 percent for starch are applied in FAO data . • - 64 - pellets for feed, either for domestic use or export) starts at 2/3 of the processed amount in 1962 and rises to 80.0 percent in 1919. Cassava input to starch production beglns at 20.0 percent of the processed amount in 1962 and declinesto 9.0 percent in 1979. The absolute quantities in FAO data form a step function, remaining constant for five-year perlods, then ~ncreasing by 5 thousand tons in a single year, then remaining constant again for five years. Cassava input to tapioca production comprises the remainder, with absolute quantities rising in similar monotically non-decreasing fashion, but with shares declining slightly to 11 percent by 1919. FAO data appear in other formats, but the statlstical base, or lack thereof, remains the same. For example, the "Standardized Co~modity Balances Tape 1984" (Rome, 1985) includes series for availability (production minus exports), food (direct food consumption plus cassava input to tapioca processing) and "other uses" (waste plus eassava input to starch processing). Bacause of the mass i ve i neraase in exports in 1979-81, the post 1,979 FAO seri es exhibít sorne peculiaríties. Drled eassava input on the "Supply Utilization Tape" inereases from 20.0 pereent to 42.6 percent of production from 1979 to 1980 (Table 12), f~r example, and the program synthesizing these serias generated large negativa numbers for "other uses" in 1980 and 1984 on the "Standardized Commodity Balance Tape." Neverthéless, these series represent some improvement in credibility over the 1981-82 tapes. The waste percentage has been lowered (to what is probably the mínimum parametric value used by FAO). The estimated production shares of processed cassava hava been - 65 - raised very substantial1y and exhibit a rising trend including slightly rising, then stagnating absolute quantities for starch production, and a massive acceleration in dried cassava to parallel the appearance of lucrative export opportunities in the 1980s. Food uses exhibit a plausible declining share of cassava production, and the FAO trade data now includes the overwhelmingly important movements in the dried cassava trade since 1979. But it must be remembered that there is no actual statistical basis for these utilization shares save a very indirect one based on the foreign trade data, and all series are essentially derived from the almost wholly unreliable produetion estimates. Of eourse, it is mueh easier to critieize than to suggest superior alternatives since little quantitative ;nformation from China is available. But it may be reasonable to suggest that several of the improvements since the 1981-82 tape did not go far enough. China has developed a considerable reputation for low foad waste. As others have previously indicated, this reputation may be somewhat exaggerated. 62 But with a large proportion of the cassava crop allacated to same-farm animal feed and high labor application per hectare, one may reasonably expect that at least cassava waste in China is quite low. The 1982 CIAr delegation observed that the primary use of cassava was as animal feed. Of course, their sample was biased toward moreproduct i ve farms. though they vi s ited some very poor 62e.g. Vaclav Smll, "China's Food: Availability, Requirements, u l' ,1 , - 66 - communes where cassava was the principal human food source. Visiting any of the state farms immediately biased the sample on such a brief trip. Based on Table 1 and other figures provided above, state farm cassava plantings could not haya exceeded 3.5 percent of Guangxi cassava area in 1984, although probably totalling 5-10 percent of production. In Guangdong, the proportians cauld be slightly higher, but state farm cassava is clearly a minar share af the total. However, the CIAT delegation found cassava primarily grown for animal feed on communes as well as on state farms. According to the extensive surveys (also biasad toward more productive farms) canducted by Nanjing University students supervisad by John Lossing Buck between 1929 and 1933, 18 percent of the output of svleet pota toes (generally a food preferred by Chinese to cassava) was employed as animal feed in the regían. The proportion was almost half in the more productive areas of eastern Guangdong. Only 60 percent of the taro erop was used for human food. 63 Sinca the 19305, swine stocks and graio and sugar production have increased more rapidly than the human population in the region (Table 13), and per capita incomes have increased. Oilseed and soybean production has declined in Guangxi, but in Guangdong, prod~ction increased at about the rate of popuJation growth over the 5-decade period given that included 19305 figures are somewhat prone to overestimation. Cattle stocks declined over the 1970s in Guangdong but due to their smaller numbers and diet preference for leaves and grasses over roots, this 63John Lossing Buck, Land Utilization in China (Atlas and Stuqy) (Nanking: Nanking University, 1937), Atla~ pp. 82 and 98. - 67 - Table 13. Growth Indices for Human Population, Livestock, and Grain, Sugarcane, Peanut and Soybean Production in Guangdong and Guangxi, 19305-1984 1979-84 Average Guangdong Guangxi Guangdong Guangxi (1952-1957 avg.=100l (19305=100 1 Human population 162 al 181 174 221 Swi ne stocks 280 til 257 Cattle & buffalo stocks 74 el 261 Small ruminant stocks 15 el 310 Foodgrain production 171 - 181 178-199 205-249 Sugarcane production 246 691 1631 Peanut production 285 di 138 168 69 Soybean production 182 -;'1 156 469 Cassava production 757 Notes: ~I 8ased on a weighted average of midyear figures for 1954 and 1957 to approximate a midyear 1955 figure. 1979-84 data are year-end figures. ~I Based on a mídyear 1955 figure. A weighted average of midyear 1953, midyear 1955 and a year-end 1957 is slightly 10wer. El Based on year-end 1984 and 1957 figures. ~I Based on 1953-56 average. The index number based on 1957 alone is 199. ~I Based on 1952-56 average. The index number based on 1957 alone ís 94. Sources: Bruce Stone, "An Examination of Economie Data on Chinese Cassava Productíon, Utilization and Trade," paper prepared for the International Center for Tropical Agriculture (CIAT), IFPRI, Washington, D.C., August 1983, Table 11. Data have been supplemented from Guangxi Jíngji Níanjían Bianjibu, Guangxi Jingji Nianjian 1985, pp. 519,530, 532 and 594; and from State Statistical 8ureau, PRC, Statistical Yearbook of China 1983, 1984, and 1985. - 68 - decline would ha va less effect on the allocation of the cassava root itself than would the sw;ne stock growth rateo According to a 1980 survey of 15,914 households, an average of 94.4 kilograms of meat (mostly pork), 35.6 kilograms of "grains" and 126 kilograms of "vegetables" were produced on private plots. Al though hog feedi ng regimens in Ch; na have been concentrate-poor historically, the fattening process would still require around 82 kilograms of concentrate per hog and the requirement has been rising with greater peasant autonomy, adjusted purchase price structure, and growing acceptance that extremely concentrate-poor diets are uneconomic. 64 In Guangdong and Guangxi, a sizable proportion of this concentrate consists of cassava, taro and sweet potato. Of the three, cassava would be the erop with the highest proportion allocated for feed. One may conclude that even for domestically utilizad cassava, 20-25 percent (for "feed use" plus "dried cassava") from 1961-79 is probably too small a proport;on for feed and the trend must have be en rising more rapidly over the period than assumed by FAO. When one considers that from 1980-82 dried cassava exports must have constituted 30-60 percent of what the 1982 CIAT delegation was told was national production, and that exports may still exceed 30 percent of annual output, even the current FAO feed proportions of 50-55 percent ("dried cassava" plus "feed") may be too low. 64See Stone, "China's 1985 Foodgrain Production Target," pp. 99-103. The 1980 survey appeared in Xinhua [New China News Agency), news bulletin, June 16, 1981. - 69 - Table 14. Oevelopment of Starch Production in South China, 1952-1984 Number of Starch Required Proportion of Total O~erating Factories Production Fresh Roat Cassava Out~ut Guangxi Guangdong Guangxi Guangxi Guangxi (metric tons) 1952 1 282 (-1,500) (-1) 1959 12,275 (-68,000) {-lO) 1962 29 1972 56 -10,000 (40-60,000) (3-14) 1983 284 59,400 (-242,500) (-15) 1984 240 49,000 (-200,000) (-17) Notes and Sources: Fi9ures in parentheses are calculated estimate5. The FAO extraction rate of 18 percent was used for the 19505 data to calculate fresh root equivalent, assuming also that all Guangxi starch was produced from cassava. (Actually small amounts of corn are also used.) For later years, an extraction rate of 24.5 percent was used based on the statement that starch content of dried cassava is more than 70 percent (Guangxi Jingji Nianjian Bianjibu, 1985),[Guangxi Economíc Yearbook Editorial Board], Guangxi Jingji Nianjian 1985 [Economic Yearbook af China 1985] (Nanníng: Guangxi Jíngji Nianjian Bianjibu, 1985), p. 192). If the FAO-adopted drying factor of 35 percent is used, this implies a starch extraction rate of more than 24.5 percent which 15 p05sible, especially in view af substantial cassava selection and breeding in China for high starch contento The 1982 CIAT delegation observed extraction rates'of 25-29 percent with 5-10 percent residues for animal feed (Cock and Kawano, "Cassava In China,· p. 8). It is not clear why the FAO-adopted extraction rate for tapioca (22 percent) is higher than for starch and exhibits as much as a 4 percent difference since tapioca production normally follows from starch production thereby achieving a very slightly lawer extraction rate (correspondence fram John K. Lynam, Cassava Program, Centro Internacional de Agricultura Tropical (CIAT), December 22, 1983.) The proportion allocated to starch production ts probably also consistently underestimated by FAO. Data assembled in Table 14 suggest that if the Guangxi record can be taken as representative of both southern provinces, utilization of ca55ava for starch production during the 19605 and 19705 canstitute not 10-20 percent of al1 cassava used for processing as assumed by FAO (2-3 percent of production), but closer to 10 percent of total production, and - 70 - potential1y higher in severa1 low production years. Assuming the adopted extraction rates and the Guangxi series are rough1y correct, and that starch produced from raw materíals other than cassava was indeed very minor in Guangxi, then the starch industry claimed more than 15 percent of fresh root production in the Autonomous Regíon in 1983 and 1984. The proportion for Guangdong ís probably somewhat lower but appears to be rising at presento All in a11, íf forced to estímate, current utilization of Chinese cassava might run 60-65 percent for feed (Including "dried cassava" plus fresh feed, exports and domestic use), 15-20 percent far the starch industry, 2-4 percent for tapioca production and as 1ittle as 1-3 percent far waste, 1eaving somewhere around 10-20 percent far direct human cansumption. As suggested in earlier papers and as FAO seems to accept, it ís quite possible that the 3 million ton circa 1980-81 production figure is an underestimate, but the production trend for the last few years is almost certaínly downward. The Guangxi starch production figure listed somewhat arbitrarily for 1972 is based on the statement that starch production in Guangxi remained at around 10,000 tans during the 19605 and 1970s (Guangxi Jingji Nianjian 1985, p. 192). Most data in the table appeared in ibid. The number of starch factories operating in Guangxi in 1962 and in Guangdong ;n 1972 are from Liang Guangshang (ed.), Mushu Zaipei yu Liyong [Cassava Cultivation and Use] (Guangzhou: Guangdong Keji Chubanshe [Guangdong Scientific and Technical Publishing House], 1980;r, p.~ The proportion of total Guangxi cassava production was calculated from data appearing in this table and in Table 2. - 71 - Cassava for direct human consumption The previous seetíon has concluded that cassava for direct human consumption probably comprises only 10-20 percent of current production. There appear to be four principal categoríes of direct human consumptíon of cassava in China: consumption related to ethnic mínorities where cassava has a traditional dietary role; consumption related to forest cultivation in remote areas; consumption associated with exceedíngly poor and/or risk-prone farming areas; consumption related to particular cuisine and especially seasonal preparations. These four categoríes are not mutually exclusive but seem to characteríze the dírect human consumption demand for cassava. Little recent ethnographic information on minorities in South China seems to be available, but taro and cassava are known to be important food items among the Yao minority in northern Guangdong. 65 The Mao people of Thailand are also habitual consumers of cassava. Mao people in South China were likewise reported to eat cassava and "mao" potatoes during the 1950s. 66 Even among Han Chinese (93.3 percent of China's populatíon) hame-processed cassava flour is aften used as a thickener in southern Chinese soups and in making special cakes at festival times such as New Year's Eve in Fujian, for example. 67 65Buc k, Land Utilization in China, lAtlas), p. 98. 66S un Jingzhi (ed.), Huanan Dichu Jingji Dili; State Statistical Bureau, PRC, Statistical Yearbook of China, 1985, p. 195. 67Cock and Kawano, "Cassava in China," p. 11; State Statistical Bureau, PRC, Statistical Yearbook of China, 1985, p. 195. - 72 - Poorly developed and poorly integrated markets are almost a defíníng characteristic of developing countríes and Chína ís no exceptíon. In China, market development was further retarded by a number of factors. First, for a thirty year period, civil war and World War 11 combined to destroy normal market activity in many areas of China.. Although Guangdong and Guangxí were spared to a much greater extent than North China, the Northeast and the Yangzi Valley, they were not unaffected by war, and nearby cassava-growing provinces su eh as Yunnan and Hunan were dírectly involved, as was Fujian, located díreetly across the straits from colontal Taíwan. For example, transport vehicles and draft animals were purchased or commandeered for the war effort. War time inflation sent marketing back to a semi-barter era and credit facilities were severely affected. In the 19505, eonditions stabilized but the government soon began to take over large segments of marketing activities. With grain erises in 1953 and 1955 and the difficultias the government was experiencing with procurement of foodstuffs for cities, graio trading became a state monopoly in 1954, and by 1955 ea eh unit of land in China was assigned a fixed quota of (usually) grain to be dalivered to state purchasing organizations at 1011 fixed prices. Taxes were also paid in kind but graín delivery obligations díd not end there. After retaining a provincially determined per eapita quantity to meet immedíate food, feed and saed needs of rural farms and households, and even after tax and quota oblígatíons were met, 80-90 pereent of all "surplus" graín was also to be sold to the state. Not only was prívate grain trading il1egal and most grain in excess of a modest - 73 - standard fer home consumption soaked up by government purchasing organization, but private traders were designated as class enem;es. The state, for its part, was having enough trouble providing for urban and army consumption, as well as reserving one-two million tons per year to export for foreign exchange. For the most part, only relatively prominent rural areas experiencing natural disasters received relief grain. More remo te and most very poor areas were left on their own without access to grain supplies from the outside. Afte~the famines in 1960-61 and especially during the Cultural Revolution period (1966-76), this situation was institutionalized as a policy of local self-sufficiency with disastrous implications for gains from specialization and trade, and for exceedingly poor risk- prone areas historically dependent on trading and non-agricultural activities to garmer enough to eat. With procuremant problems persisting, the government further restricted non-farming activities and made migration illega1 in order to limit the state's urban obligations, but thereby binding many farmers ave n more closely to poor and risk-prone agriculture. 68 68See Bruce Stone, "Relativa Foodgrain Prices in the People's Republic of China: Extractive Rural Taxation Through Public Monopoly." In John W. Mellar and Raisuddin Ahmed (eds.), Agricultural Price Policy for Developing Countries (Baltimore: Johns Hopkins University Press, 1987); and Bruce Stone, "Chinese Socialism's Record on Food and Agriculture," Problems of Communism, vol. 35 no. 5 (Sept.-Oct.) 1986; pp. 63-72. See also Tang and Stone, Food Production in the Peoele's Republic of China; Kenneth Walker, Foodgrain Procurement and Consumption in China (Cambridge: Cambridge University Presl, 1984); and Nichalas Lardy, Agricultura in China's Modern Economic Development (Cambridge: Cambridge University - 74 - lt is not difficult to imagine that with this institutional framework, eassava, at least in the south, had a particularly important role to play. Cassava was an ideal crop for insuring minimum levels of consumptíon because ít ís a relatively drought­ resistant, stable yielding, easily stored crop, providing high calorie levels per unit area, and performs well relative to alternative crops even under poor agronomic practice and soil condítions. As a crop cultivable on forest lands and hillsides, ít was also ideal for sustaining reclamation teams in remo te areas. With the rapid increases in South Chinese rice production during the past decade (Table 5, 6 and 13), the 19805 legalization of prívate grain trading and guaranteed state food delíveries for areas concentrating on the production of econQmic crops, cassava's special institutionally-induced importance has be en declining. However, cassava is still grown in exceedingly poor areas in South China for essentially the sarne reasons: food security and easy proviston of needed calories under inoptimal conditioos. rt should be emphasízed, for example, that seven counties in Guangdong and eight in Guangxl averaged per capita c011ective distributed incorne in 1977 of less than 50 yuan ($20-25 U.S. at concurrent official rates).69 While this category excludes important incorne sources such as prívate plot and side11ne production and sorne in-kind payments frorn collectíve work, it is indicative of the amount of cash available for farrners 69Nongyebu Renmin Gongshe Guanliju [Ministry of Agriculture, Bureau of People's Commune Management], "Yijiuqiqí zhi Yijiuqijlunlan Quanguo Qiongxian Qlngxing" [The Condítion of the Nation's Poor Counties, 1977-1979J Xlnhua Yuebao [New China Monthly], no. 2, 1981, pp. 117-120. - 75 - from their principal assets in very poor localities. 70 The number of counties falling below this lowest benchmark increased to 11 in Guangdong in 1978 but declined to 7 in 1979 (in Guangxi, 8 in 1978 and 6 in 1979). In Guangdong, the very poorest regions appear to be in the northeast, such as Wuhua and Longchuan Counties, and on Hainan Island in the South, including the known cassava are a of Basuo (Dongfang County). In Guangxi, such counties seem to be clustered in the north and west: for example, Du'an Yaozu Autonomous County. Luocheng, Donglan and Napo Counties, as well as Barna Yaozu Autonomous County wh~e cassava is known to be widely cultivated. 71 But with the exception of the exceedingly productive Pearl River Delta, no part of South China can be excluded as a region where dil'ect consurnption of cassava is not important Jor sorne segment of the poorer rural population. Areas were cassava is an important direct calorie source need not be remote. Even within the Haikou Municipal Area on Hainan lsland, 11 percent of cultivated area in the Yong Sing Township, for example, is planted with cassava, two-thirds of which is consumed directly as a staple. 72 This is because only 4 70Distributed collective incorne averaged around two-thirds of the total including private plot and sideline income during those years, according to a State Statistical Bureau (SSB) survey of 10,282 households (Zhongguo Guojia Tongjiju, Zhongguo Ton9ji Nianjian, 1981, pp. 431). But this may have excluded in-kind distributíon of productíon from collective 'ands. For a fu11 discussion of Chinese distribution data and its problems, see E.B. Verrneer "Income Differentíals in Rural China," The China Quarterly, vol. 89 (March) 1982, pp. 1-21. 71Nongyebu Renmin Gongshe Guanliju, "1977-1979 Quanguo Qiongxian Qingxing," Xinhua Yuebao, no. 2, 1981. 72Cock and Kaw'ano, "Cassava ,'n Ch1' na, "p p. 10 - 11 . - 76 - percent of the farmed area is suitable for rice cultivation, the remainder being rocky hillsides upon which fruit tree horticulture is being attempted. Cassava planting provides an economic hedge against heavy market dependence. The Starch Market What little quantitative information is available on starch production in Guangdong and Guangxi has been recorded in Table 14. Historically, a significant share of financing for capacity construction and an important_share of sales deliveries have been associated with overseas Chinese, especially in nearby Hong Kong and Macau. In 1952, the Wuzhou Charcoal Industry started Guangxi's first starch factory (Jiulian Crude Starch Factory, later renamed the Wuzhou Municipal Starch Factory) with financia1 assistance from the government and from overseas Chinese. Its "sanjiaopai" [Triangle Brand] cassava starch was exported from Wuzhou in east central Guangxi to Hong Kong, Macau, Southeast Asia, Japan and the Middle East. Since the mid to late 1950s, Beihai in the far south, Bama Yaozu Autonomous County in the northwest, Xijiang Farm in the east, Wuming Overseas Chinese Farm in central Guangxi, Ningming Overseas Chinese Farm in the southwest and other farming areas 'set up fixed scale factories. 73 The designation "Overseas Chinese Farm" is an indication that overseas Chinese financial resources are involved in the commune's development. -------_._- 73Guangxi Jingji Nianjian Bianjibu, Guangxi Jingji Nianjian 1913i, p. 192. - 77 - In Guangdong, cassava starch production may have begun even earlier, but at least by the early 1970s, 56 factories had be en set up in the province and "hongpai" [Red Brand] cassava starch from the Dongguan Flour and Starch Factory on the Pearl River Delta was sold widely in Southeast Asia and Eastern Europe. 74 During the 1950s, 1960s and 19705, it seems that production econom;es and the price structure concertedly favored cassava as a raw material for starch production since despite the provincial self-sufficiency imperatives for the period, Guangdong and Guangxi exported starch not only to ~ong Kong, Macau and foreign countries, but to other Chinese provinces as well. With liberalization of rural econom;c activities since the late 19705, small scale starch ~rocessing plants have been established, especially as township and village enterprises. By 1983, the total number of starch factories in Guangxi had increased sharply to 284, though with combinad fixed assets of only 25 million yuan. 75 But either production economies no longer so clearly favored the use of cassava as a raw material, or cassava production in other provinces was expanding to meet their demands for starch. This combination of overdevelopment of production capacity and loss of part of the interprovincial market brought about a contraction in the South Chinese starch industry in 1984. In Guangxl, the number of enterprises declined by 17 percent and production fell by 16 percent (Table 14). Hpwever, part of this decline may be due to intensified 74Liang Guangshang (ed.), Mushu Zaipei yu'Liyong, p. 9. 75guangxi Jingji Nianjian 1985, p. 192. - 78 - competition from nearby Zhaoq;ng and Shao~an Prefectures in Guangdong where starch production has been increasing rapidly.76 A variety of industries use cassava starch in China, the most traditional being the cotton yarn industry whích províded demand for the fi rst Guangxi factory in Wuzhou. 77 But the Wuzhou and Beí ha i factoríes have expanded and diversified to use cassava starch as a basis for glucose production. In 1984, Guangxi produced 7,800 tons of glucose, primarily for the candy industry. 80 percent of this total was produced in the Wuzhou-and Beihai factoríes, the latter exporting to Hong Kong, Thailand and other countries. The Wuzhou factory has also initiated tria' production of denatured starch and, with purchase of technically superior equipment from Japan, has increased its extraction rate by more than 5 percent. 78 In Guangdong, tha Dongguan Factory has also diversified and now produces glucose, brewer's yeast and wine. 79 As early as 1972, it exported cassava-leaf starch to Japan, and, to England, large quantities of glucose, partial1y based on millet as well as cassava. 80 In Shaoluan and Zhaoqing Prefectures, in addition to 760elphi survey response: comments by Huang Xi, agronomist, Institute for Dryland Grain Crops, Guangdong Province Academy of Agricultural Science, Guangzhou, June 28, 1986. 77Sun Jingzhi, Huanan Jingji Dichu, pp. 258 and 333-334. 78Guangxi Jingji Nianjian 1985, p. 192. 79Correspondence from Graham Johnson, Professor of Anthropology, Department of Anthropology and Sociology, University of British Columbia, Vancouver, September 19, 1983. 80liang Guangshang (ed.), Mushu Zaipei yu liyong, p. 9. - 79 - cassava starch factoríes, a number of other processing industries have be en established which utilize cassava, including a monosodium glutamate factory, molasses plants, breweries and feed-processing plants. 81 810elphi survey response from Huang Xi, June 28, 1986. INDONESIA Production trends and distribution Cassava was introduced into Indonesia through Portuguese trade with the Spice Islands but did not become well established as a major crop unti1 the mid to late 1800's. The spread of cassaVa WaS promoted by the Dutch as a famine reserve. Also, by the turo of the century the Dutch had developed a large csssava starch industry on Java directed towards export, which also provided incentives for expansion oi cassava production. By the mid-1960's area sown to csaaava on Java resehed a peak of 1.4 million hectares and has sinee declined (Table ). Sinee 1975 eassava area on Java has been relatively stable at an even one million hectares. Cassava are a on the off-islands remained statie through the 1960's and early 1970's. Only in the later part of the 1970's has area in the off-islsnds shown a significant inerease, due to the transmigration projeets and the expansion of the gaplek trade and stareh on Lampung. The dlstribution of eaaaava production in Indonesia to a 'large extent corresponds with the dlstri~ution of population. A~out 70% of the eassava is produced on Java. Java i5 followed by Sumatra, which aecounts for a little over 10%. The rest of the production i5 distributed throughout the other islands (Table ). Cassava ls thus grown throughout Indonesia, almost wholly in upland areas. Cassava has established itself as a major palawidja erop in Indonesia. Over the deeade of the seventies eassava produetion grew at annnal rate of 2.7% per annum in Indonesia. However, this productlon growth was marked by very different rates of growth between reglons. On Java eassaVa produetlon grew at an annual rate oí 1.8%, whl1e off-Java the growth rate was 5.2%. Even on Java growth occurred only in Central and Eastern Java, while produetion was stagnate in Western Java. By far the most rapid rate of growth occurred in Lampung on Sumatra, where produetion grew at a 12.2% annual rate, tripling in the spaee of a decade. The fas ter rate of growth on the off-islanda than on Java would be expected, partieularly given the severe land constrain on Java versus the other islands and the policy to settle populatlons on the outer islands. The 1. 8% growth rate in productlon on Java in the 1970' s was due to i decline in area of 0.9% per year and an annual increase ln yields of 2.8%. Historically, ylelds on Java had been static at a little over 7 tlha slnee the 1920's (Roche, 1982) and only sinee 1973 have yields levels shown s conslstent rising trend. The natural question is what are the faetora that have preeipitated this relatively sudden and rapid rise in yields? A corrollary, however, would be the identificatian of the factora that have kept yields an Java mueh lower than other majar produclng eountries in Southeast Asia, that 18 about half the yleld levels in India and Thailand. The intensity of productian systems on Java and the favorable agro-clima tic conditions would suggest similar or higher yleld potential. Production growth on the auter islands durlng the 1970'5 showed a distinctly different pattern to that on Java. Ibe principal factor responsible for the 5.2% produetion growth rate was the 3.2% annual 1 See Roche (1982) for a diseussion of faetors contributing to declining aTea planted to cassava. 2 expansion in area. This is similar to the population growth rate off-Java of 3.0% in the 1971-80 periodo However, most of this expansion was concentrated on Sumatra, and particularly in Lampung. Area and production expansion thus appeared to be related more to expanding infrastructure and market possibilities than to expanding population. However, expanding area vas not extensive in nature, sinee csssava yields as vell rose at arate of 2.0% per annum on the onter islands. Thus, trends in cassava production in Indonesia over the past decade have been favorable, particularly given the severe land constralnt on Java where the bulk of the cassava 15 produced. Nevertheles5, cassava production on the outer lslands is growing mueh faster, due in part to the unexploited land resourees there. This creates something of a dichotomy in planning further expansion of cassava, which, as will be seen In the succeeding analysis, is reinforeed by other major differenees in both production and utilization between Java and the outer islands. Cassava produetion systems Cassava production systems in- Indonesia, unlike other -major cassava producing countries in Asia, are complexo Complexity in this case introduces diversity and aeross Indonesia 'there is substantial variatian in production systems based on agro-climatie conditiona, land availability and market access. Unfortunately there has be en only one major attempt to study in depth some of these production systems, and thus, this section ..,i11 by force of necessity principally summarize the research of Roche (1982) in his analysis of cassava cropping systems in three regions of Java. Because of the differences in land/labor ratios between Java and the outer islands, production systems on Java will be considerad independently of tbose off-Java. The eomplexity of cassava production systems derives from intercropping and rotation systems and double-cropping with rice in certsin land types. Becauae median farro aize of Java ia only 0.4 heetares, farmera seek to optimize returns to this limited resouree. Over haH of cassava grown on Java in intercropped (Table ), with the principal intercrops being maize and upland rice, and in West Java legumes such a peanuts snd soybesns. In certain areas clase to urban areas where fresh market prices are sufficiently high, cassava in monoculture will follow rice on irrigated land particulary, where there is not suffieient water for a second rice crop. Finally, sltbough cassava will in most cases not complete for land with rice, it will have to compete for labor and capital resourees, so that appropriate timing of cassava cultural praetices is a major factor in production systems. Agro-climatie conditions, particularly rainfall distribution soil type, and soil fertility together with irrigation availability are determining factor s in the choice of eassava cropping system. Rainfall i8 adequate for cas,sava all over Java but in eertain rainfed areas is limiting for other erops. Thus, as rainfall reliability declines from west to cast (Figure ), eassava production tends to be concentrated more in the eastern part of Java and on the island of Madura (Figure ), even though cassava is grown tbroughout Java, apart from the irrigated areas of the northern plains. 3 Soil type, topography and the eroded state of so11s define the other major constra1nt on adaptat10n of upland crops. Soils with major fertility, acidity, or toxicity problema, such as Ultisols, are principally found on the outer islands. The principal constraints on Java arehighly eroded, unterraced hillsides. Such areas tend to be concentrated in the south-central coastal zone, an area where cassava production is most highly concentrated. Whereas rainfall distributlon principally affects tlming and whether one or two intercrops can be planted, land type determines the range of crops that can be grown. At the extreme where soils are highly eroded, eassava ls the crop of last resort. In general, as soil and rainfall eonstraints become more severe, first legumes leave the intereropping system, followed by up1and rice, and final1y maize, 1eaving Cassava as the sole crop on highly eroded soils. Where soi1 and rainfall are not 1imiting, al1 of these crops can be included in one aystem, as shown in Figure However, in general upland rice is the principal intercrop in the wetter, western part of Java, whi1e maize ia the principal intercrop in the central and eastern regions. In most systems the land is prepared before tne start of the heavy raiñs, in general around October or November. Tbe upland rice and/or maize are p1anted and after establishment in two to four weeks cassava i5 planted. Where s011 conditions are not 1imiting, this system provides effective ground cover until caasava reaches full canopy and th1s aids in controlling erosion under the high rainfall cond1tions of Java. The resource structure of the systems vary substantially (Table ) • Labor use i8 in general high even in those areas where bullocks are used in land preparation and lnter-row cultivatíon. Fertilizar use tends to be higher in the more productive land types, princlpally because more responsive crops are planted in the intercrop system and re1atedly such systems probably give the higher marginal return to fertilizer use. Cassava yield levela thus vary substantially between systems. Over 70% of caasava ia planted in the majar rainy period from September to January (Figure ). This introduces two principal constraints on cassava production systems. First, this coincides with the major rice planting season, which creates competition for labor resourees. Second, the crop must be harvested and the land cleared by the start of ths next raina. Where cassava is dried into gaplek, the harvest must be earlier to take advantage of the dry season. In those systems were cassava follows a rice crop, timing is crucial sines the crop has six ta eight manths before harvest. Nevertheless, the longer maturity of the eassava complements the harvesting pattern for rice (Figure ). The major portion of the cassava harvest oecurs in the June-October period after the principal rice harvest, insuring a more stable supply of carbohydrate sourees. This tends to coincide with the dry period, so that cassava roots can be processed into gaplek where markets for fresh cassava are not: assured. Roche (1982) presente evidenee which suggest t:hat cassava continues to grow and add root weight during the dry seaeon -- this wou1d not be the case were soil moisture limiting. Farmers t:hus face a trade-off between timely harvest for either gaplek drying or early land preparation and eventual cassava yield. Where eassava principally supplies starch factories or urban 4 markets, there ie a demand tor more continuous supplies of roots. However, this 1s only possible where rainfall is sufficient to support the intercropping system during most of the year, such as in West Java, or "here land types are suited only for pure stand cassava. In general, providing for more continuous supplies of cassava roots i8 heavily constrained by ralnfall distribution and the complexity of the cropping system on the small farms of Java. Moving from Java to the outer islands, the factors "hich determine cassava production systems change dramatically; rainfall distribution, soils, farro size snd markets a11 change quite significant1y. Cassava production systems on the outer islands are best considered independently of those on Java. Tbe initia1, striking difference 19 in rainfall distribution. In general the outer islands have a more continuous supp1y of rainfall than Java (Table ). On Sumatra, Kalimantan, and, to a slightly lesser extent, Sulawesi, the major portion of area i8 suitable for continuous cropping, as compared to only 20% of the area of Java (neglecting_the irrigated areas). Interestingly, par capita production of ca$sava in Indonesia is highast in those areaa -- Java and Nasa Tenggara -- where there is a significant part of the area "ith constraints on water availability during the year (Table ). Soils, in general, also vary markedly between Java and the outer islands. Whereas rainfall is not as limiting on the outer islands, 80ils in thase araaS impose much more severe constrains on cereal and 1egume crops, although not on cassava. The soils are in general ultisols, baing quite acidic, of a lo" fertility status, and occasionally having relatively high levels of exchangeable aluminium. Because of these soil problems together with the erodability on slopes, much of this land area has been classified as marginal for cereal and legume crops. Cassava, however, is well adaptad to these soils; but, continuous cropping of such soils requires appropriate crop and soi1 managsment to maintain productivity levels. Cassava production systems on the outer islands have in many "ays besn conditioned by the dictates of the transmigration schemes. Before the advent of the transmigration achemes, much of cassava on the outer islands was grown in a shifting agricultural system. Su eh a system was very extensive, particularly sinee the abandoned fields returned to "alang-alang" (Imperata cylindrics) rather than the original forest fallo". The transmigration schernes super1mposed a fixed farro size structure over the original shifting system. Farmers "ere in general given 3.5 hectares to exploit, and apart fram the tampung area, the settlement areas "ere chosen where the 80ils were not ultisols. Farmera, however, could not effectively utilize the "hale 3.5 hectares. On the oue hand, labor intensive cropping patterns "ere brought from Java to an area where labor needs relied solely on family availability and there "as no bullock power. On the other hand, infrastructure was lim1ted and there "as no effeetive market, even "ere surpluses to be produced. Until sufficient infrastructure was developed, such as happened on Lampung, there "as little incentive to SO" over 0.6 to 1.0 hectares, sufficient to meet family food needs. 5 Cassava provides a certain production without purchased inputs and for this reason eassava has been crucial in meeting the food needs of newly arrived settlers in the transmigration projects, at least until rice paddies ean be established in those are as where rice production is feasible. On the poorer soil areas cassava remains in the eropping pattern. Cassava in the outer islands 1s grown only on rainfed soils and usually in association, either with maize and upland rice or in the establishment of tree crops or between the rows of shorter tree cropa 1ike coffee. It is tree cropa that are becoming the major cash erops on the outer island8, and it i8 only in Lampung where caasava has so far carved out a place as a primary cash crop, first as gap1ek for export and< currently for starch. Even though rainfall is relatively well distributed farmers still prefer to plant upland rice and maize during the montha with the highest rain fall, so that there continues to be sorne seasonality in eassava production. Eecause of this seasonality and the history of plantation systems in Indonesia, cassava plantation systems have also been developed on the outer islands. These have usually been developed in conjunction wieh large-seale starch plants, of which there are at least eleven in Lampung (Nelson, 1982). There i8 little informarion on these systems. Ihere i5 substantial mechanization, even in the harvesting of roots. Mclntosh and Effendi (1979) suggest that after opening new land, yields are high the first year but decline over time. Fertilizer ia used only after the third or fourth year or the land ls left fallow, and new land 1s opened up. These plantation syatems provide continuity oí supply, but the factories depend for most of their needs on small-scale production systems. Cassava production systems in Indonesia, as compared to other producing countries in Asia, are characterized by considerable diversity, depending on rainfall, land type, and market, and a fair degree of complexity, due to the intensive nature of su eh small size farms. Focusing on just a single crop Buch as eassava would fail to define the determinants of the system. Improv1ng productivity of eassava wi11 necessari1y have to focus on improving the productivity of the whole cropping system. Yields: Yields of cassava in Indonesia in 1980 averaged 9.7 t/ha, compared to average yields of 13.1 t/ha in Thailand snd 18.3 t/ha in India. S011s and ra1nfal1 are probably on average better in Indonesia than the other two countries. Labor and input use are in general on a par with India. These comparisons would tend to imply that apart from var1ety cropping systems in Indonesia have a substantial affect on cassava yield. Probab1y three principal factors are influencing yield: plant density in intercrop systems, delayed planting of eassava in the intercrop system, and a shorter growth cycle. Zandstra (1978) has shown a decline in cassava yield with delayed planting of cassava in intercropping rice and maize. Planting cassava is delayed from 3-4 weeks (Roche, 1982) to two months (Mclntosh and Effendi, 1979) after the planting of the rice and maize. Such systems tend to ! incraase the rice yield and deerease the cassava yie1d. Plant densit1es f also vary in these systems, particularly 1f a second crop 1s to be 1ntercropped after the rice and maize harvest. In such cases plant densities are as low as 4,500 planta/ha. On the other hand, in the eommon 6 rice-maize-cassava system the cassava population can be maintained at 10,000 plants/ha. Depending in part on variety, trials in general show very little response to increased plant population after 10,000 planta/ha (Wargiono, et. al., 1979). Finally, there i8 substantial evidence to suggest a trade-off between early harvest and yield. Nevertheless, Roche (1982) among others has shown that intercropping systems are more productive than monoculture cassava. The issue again arises as to what has been responsible for rising yields of cassava, which then leada to the question of what is the potential for raising yields in these systems. Roche suggests that increased fertilizer use has been the principal factor. Since the early 1970's there has been steady development of fertilizer marketing channels, first for irrigated and then for upland areas. Moreover, there has been a policy oí subsidizing the price of fertilizer. Application of fertilizer on cassava has thus steadily increased over the 1970' s (Table ) • Nevertheless, average application rates only stand at little over 20 kg/ha, well below applieation rates on cther upland cropa. Yet, sinee cassava 1s often intereropped with upland rice and maize, cassava is also benefiting froro ehe increased appliea~ions to these crops. The other avenue to increasing cassava yields would be to favor cassava over other crops in the system. Farmers can make marginal adjustments in planting dates, harvest dates, spacing, or density of the 1ntererops to inerease cassava y1elds, in many cases at the expense of yields of other erops in the system. However, if anything eaasava prices have decl~ned moderately in relation to the prices of the other upland crops (Rache, 1982) over the decade, prov1ding little incentive to favor cassava over other cropa. The only other incentive would be improved market aceess. With the rapid expansion in starch production, both at the household and the factory level, more stable market conditions may have developed, resulting in a decrease in risk of marketing the perishable root. Unfortunately, there is no evidence to support the intensificatian of cessava within upland cropping systems. The other major characteristic of cassava yields in Indonesia 1s their variation between systems. Aggregate statistics suggest relatively similar yields between regions but Roche found average cassava yields varying from 2.3 t/ha to 19.5 t/ha, depending on the system. The variability depended in part en rainfall conditians, management, and intercropping system but seemed to be most related to land type. Yields "ere lawest on eroded hillsides and highest on the level rainfed soils or, in the dry season, bunded land, even though in the latter the growth period was very short. The yield range was further widened because fertilizer tended to be applied to the batter soils. Increesing yields wil1 in large part depend on adapting technology to different land systems, a principal feature of IRRI's cropping systems research methodology (Zanstra, et. al., 1981). Costs oi production and labor utilization· Compared to other countries in Asia, labor use in casseva production systems in Indonesia i5 high, in general double or triple per hectare labor inputs in most other countries. This reflects the very low land/labor ratios en Java, on the one hand, and the more complex cropping systems, on che other hand. Nevertheless, even in monoculture cassava systems where 7 bulloks are used in land preparation, labor input exeeeds 200 mandays/ha (Roche, 1982). Even more striking i8 the fact that labor input off-Java remains high. In a survey by Hambrect (personal communication), labor input in Gedony Tatson district in Sumatra averagsd 354 mandays/ha, of which 61 wsre for pseling and drying into gaplsk. Even on ths off islsnds labor intensity of the production systems i5 not radically sltered. Labor thus forms a major component in costs oí production; however, the proportion varies markedly with the ipherent productivity oí the land system. On the eroded hillsides of Gunung Kidul labor is praetieally the only input, while on the level rainfed soila oí Kediri, labor costs are higher than Gunung Kidul but still form less than half oí total variable costs (Table ) • Higher levels of purchssed inputs sre applied to ths more productive land systems, so that naturally higher yields are achieved with higher per hectare costs. Arriving at apure costs of production for cassava in Indonesia i8 complicated by the intercrops in the system and the costing of farroer owned resources. Using only monoeu1ture systeros where possible -- although Roche has shown intereropping systems to be more proíitable --, a fu11 costing of sIl inputs at tbeir market value shows that casssva systems in general are not even eovering variable eosts. Although cash incomes are positive (Roche, 1982), returus on own faetors are in general less than the market priee. Certainly at average output priees of 20 rupees/kg of roots there is no profit that can be attributed as a returo to land. On Sumatra farro prices can be as low as 9 rupees/kg. Clearly, the opportuuity cost of farro resourees can be well below market rates. This is quite ·logieal in systems where subsistenee ueeds have a high priority, where there is substantial underemployment in labor markets, and where land, though having a high scarcity value, 1s usually merely sufficient to meet subsistenee requirements. The very high labor inputs thus are not neeessarily translated into high labor costs, aud together with subsidized fertilizer priees and the additional income froro intercropping, cassava output prices can often fall below implicit production costs aud sti11 remain a relatively stable part of the cropping system. This can be seen in the relative stsbility of cassava in the Java eropping system over the last several decades, even though caS$ava i8 more of a cash crop than a subsistenee crop. '. Teehnology development Since the constraints on cassava yields are both not fully understood and vary substantia11y aeross Indonesia, !' research program to devalop yie1d-increasing, cassava technology needs both a clase linkage to farmer production systems and a quite extensive testing system. Moreover, raising cassava yields wi11 have to be done within intercropping systems, snd it will not be possible to heavily sacrifice yields oí other crops in increasing cassava yields, especially that of upland rice." Finally, yield potential wil1 be heavily circum8eribed by elimatie and 80il eonditions. so that any yield gap analysis will have to be defined in terms of Iocation and land system. Such a research focus requires a certain critical level of resourees, yet reseaeh resources for palawidja cropa have traditionally been limited, 8 as most resources have been devoted to rice. Agricultural research ie relatively centralized in Indonesia and comee under the responslbility of the Agency for Agricultural Research and Development (AARI). AARD i8 divided into seven major research centers, of which casssvs comes under the Central Reseerch Institute for Food Crops. These central research institute are in fact a coordinating body for a set of regionally based research centers, of which there are seven under the Central Research Institute for Food Crops. Cassava research in Indonesia is centered in the Root Crop Improvement Program, which is under the Bogor Research Institute for Food Crops. There is some consideration of plans for decentralizing research decision-making and making the seven research institutes semiautonomous, which could mean that cassava research could be done in many these institutes. However, currently caSsava research ls centered at Bogor, which focuses on more basie researeh. Thus, all of the cassava breeding research is done at Bogor. Agronomic research and advanced selection of clones are done at some of the other research centers. Cassava tec~nology development in Indonesia in the postwar period has principally focused on varietal development snd fertilizer trials. Two varieties, Andira 1 and 11, were released in 1978. Adira 1 has a lower HCN content, shorter maturity, higher starch content, and about the same yield potent1al (35 t/ha) as Adira 11. Adira 1 i8 apparently grown quite widely on Lampung (Roberto Soenaryo, private communication) but its adoption on Java has not been widespread. Understanding why farmers have not adopted Andina 1 could offer valuable insights into whether the problem is the variety or its extension. Clearly, in lndonesían cassava systems yield is only one criterion among many that wil1 motivate farmer adoption. Roche (1982) argues that the most immediate avenue to increasing cassava y1elds is through a cOmbination of the Adira 1 variety and appropriate fertilizatíon. In the longer term more finely tuned varietal development together with integrsted fertilization, rotation, seed management, and intercropping practices designed for homogenous land systems w11l probably be the principal means to achieving significant increases in caSSava yields. Certainly the objectives will be a stable, continuous cropping system in upland areas with cassava as a significant component. Another ,consideration ia "hether a distinction should be made in a cassava research strategy for Java versus the outer islands. This issue, to a large extent, ,,111 depend on land policy and the ava11ab11ity of labor-saving technology. Currently cassava and other food crop production depend on the very labor-intensive, production systems developed on Java. Farmers usually cannot utilize all the land allocated to them because of the lack of labor and/or tenant markets. Most research to date has focused on further intensification of intercropping systems, with focus en the particular so11 constraints of the outer islands. The issue i8 whether higher incomes could be schieved with more labor intensive use ·of land vis-a-vis less intensive labor use but cultivatlng more la~d. Because of the agrocllmatic conditions and this labor constraint on the outer islands, tree crops have become a principal farmer alternative. Cassava in sorne systems i8 intercropped with coffee, clove or oil palm, unt!l tree establishment. Smal1 land allotments, movement to tree crops, 9 and lack of less labor intensive soil preparation and weeding practices will thus 1imit cassava production increases to moderate area growth and yield increase. Policy has thus dictated similar lines of research for cassava on the outer islands as on Java. MARKETS ANO DEMAND A synthesis of production and utilization . Explainig the sources of increased cassava production in Indonesia provides only haH of an analysis of the cassava economy in the country. Increasing production implies increasing consumption, and a complete analysis requires an evaluation of sources of demand growth-Indonesia provides in many respects an example of a well integrated cassava economy, in that the multiple uses of cassava are fully exploited. Before studying the sources of increased cassava utilization, the consistency between the production and consumption estimates are first reviewed. The supply and distribution estimates are based on data for the year 1976 and the estimates are broken down for Java and the outer islands. Two other estimates of cassava supply and distribution exist; one is the food balance sheets for Indonesia put not by the Central Bureau of Statistics and the other is an estimate by Laurian Unnevehr (1982) for Java only. These estimates will be used as a point of reference in developing the supply and distribution estimates. Food uses are a dominant form of utilization of cassava in Indonesia. The most systematic estimates of cassava consumption patteros comes from the periddic National Socioeconomic Expenditure Survey (Susenas) -- see Dixon (1982) for a discussion of the structure of the surveys. The 1976 survey (Susenas V) found an average per capita consumption of 21. 6 .kg of fresh roots snd 8.0 kg of gaplek on Java and 34.2 kg of fresh roots and 3.8 kg of gaplek on the outer islands. This resulted in an average for Indonesia as a whole of 26.2 kg of fresh roots and 6.4 kg of gaplek 01' an average of 45.4 kg of cassava on a fresh equivalent basis. A standard rate for converting fresh roots to gaplek is more complex in Indonesia than Thailand becauee roots are peeled and gaplek is not dried to a standard percentage. This introduces peeling 10s8, moisture content, and dry matter content ·as variables in the determinatíon of the conversion rateo Field observations suggest a peeling 10ss of 20% (Unnevehr, 1982), which ls in accord with standard percentages of peel to root weight of 15 to 20% found at eIAT (Rupert Best. private communicat1on). Moisture content of gaplek is apparently highly variable. Field observation by Unnevehr suggests levels as high as 25%. Studies at CIAT (Rupert Best, private communication) have found problems of heavy fungal growth on cassava chips with higher than 18% moisture, even after one week. Drying to moisture levels of 20% 01' above, the storage life of cassava i8 not. substantíally extended, unless there are alternative means of centrol to fungal growth. Unnevehr did find relatívely high losses in gaplek storage, but only after relatively long periods. What average moisture content of gaplek i5 at the point oí consumption remains somewhat of a question. So also, does the average dry matter content of cassava roots. !U Dixon (1982) and Unnevehr (1982) both employ a conversion rate of roots to gaplek of 2.5 to l. Assuming a 20% weight 10ss due to pee1ing, gap1ek at a 25% moisture content imp1ies a dry matter content of 27.5%, while at 20% moisture, a 30% dry matter content is imp1ied. These dry matter percentages are we11 within the normal range and may even be somewhat on the 10w side when compared to different genotypes eva1uated at Bogor. A 2.5 to 1 conversion rate is then probab1y a reasonab1e balance between root dry matter and gap1ek moisture contento The 45.4 kg average 1eve1 of cassava consumption from the expenditure surveys compares to an estimate from the food balance sheets of 76.0 kg per capita. Food consumption in the food balance sheets is estimated as a residual, after a11 other uses.have been deducted. The discrepancy between the two estima tes is significant and provides the first indication that there may be some discrepancy between production and consumption estimates. Gap1ek is not only used directly for human consumption but is a1so exported and Unnevehr (1982) found some gaplek being mil1ed into f10ur by who1esa1ers and used in bakery products. Gaplek exports from Indonesia are high1y variable and in 1976 exports, particular1y from Java, were especial1y low. A five year average around 1976 is therefore used as a normal export 1eve1. Cassava flour is assumed to be produced on1y on Java and Unnevhr's estimate is used. Starch is a major uti1ization form in Indonesia and a1though it principal1y goes into food uses, starch consumption is not inc1uded in the human consumption estimates. Utilization of cassava as starch comes from starch production estimates. The most rigorous eva1uation of these estimates is provided by Ne1son (1982) for the years 1973 and 1979. Geometric growth rates are used to interpolate a 1976 estimate. Animal feed provides the on1y other possib1e end use of cassava. Roche's (1982) survey of cassava production systems suggested no feeding of fresh roots to anima1s. Given the limited importance of swine, the dominance of ruminant animals and their ability to utilize 10wer cost feedstuffs, and cassava's role either as a cash or food crop, any on-farm feeding of cassava roots wou1d be expected to be 1imited, a1though there are not reports to confirm this assessment. Incorporation of gap1ek into ba1anced feeds is al so thought to be 1imited, given that market channels for gaplek are directed principally to export. Unnevehr in her study of gap1ek marketing channels mentions no movement of gaplek into, what is in many respects, a very limited feed concentrate industry. The assumption wi11 be made then that any use of cassava in animal feed is limited. These data then lead to the supply and utilization estima tes in Table Without even considering a waste component, there is a very close correspondence between production and consumption estimates for ~he outer islands. On the other hand, for Java production estimates are significantly higher than consumption estimates by almost 2.9 million tons. Assigning all the difference to waste is not justified given the intensive nature of production systems, the close integration with markets, and because of the very limited incomes, the tendency for both farmers and, midd1emen to be very conscious of 10ss. In marketing channels for fresh roots Unnevehr reports losses of around 8%. The more significant 10sses 11 occur in the atorage of gaplek from the main production period for consumption in the period of high rice prices. Unnevehr reports losses in this context of from 10 to 20%. Applying 8% losses to marketed cassava and 15% to all gaplek for human consumption -- the lower moisture content and better storage facilities would militate against such losses in the export trade -- yie1ds a 10ss figure of .36 million tons. This leaves 2.5 mil1ion tons unaccounted for on Java. The discrepancy i8 too large to attribute just to underreporting in the different consumption estimates. Moreover, conversion rates of roots to processed product have consistently been assumed to be on the high side. A tendency to overestimate production somewhat would therefore seem to be implied, with no particular reason to suggest whether yields or area or both are being overestimated. All factors considered there i8 probab1y a firmer basis for accepting the consumption estimate over the production estimate. The supply and distribution analysis suggests a significant difference in utilization· patterns between Java and the' outer islands. On Java utilization forms are fair1y balanced between fresh roots for human consumption, gap1ek and starch. On the outer ie1ends, on the other hand, fresh root consumption is by far the largest consumption form, a not snrprising fact given the lack of infrastructnre and the focns on subsistence consumption. The other major characteristic of cassava utilization patterns in Indonesia is its diversity, particularly in relation to other Asian cassava producers. Indonesia heavily exploits the mu1ti-use characteristics of cassava, with major markets for fresh human consumption, starch and gaplek, both for human consumption and exporto Understanding how caSBava production is allocated to these various markets, each with relatively different growth potential, may aid in developing similar market structures in other countries. Cassava for direct human consumption The food economy of Indonesia is based on rice. While less preferred than rice, cassava, nevertheless, i8 the seeond most important carbohydrate source according to Susenas data (Table ) but sti11 makes up no more than 10% of average calorie intake. The successful extension in irrigated areas of the high yielding rice varieties resulted in increasing per capita availabi11ties of the grain during the decade. Trends in cassava consumption are more difficult to interpreto The food balance estimates follow production trends and suggest a distinct increase in consumption since 1973; on the ocher hand, the Susenas estimates suggest more or 1ess stable consumption over the decade (Table ) . What 15 clear i5 that cassava continues to maintain a secondary but yet important role in the Indonesian food economy, with this importance lying more in distribut10n of cassava consumption rather than in aggregate averages. Cassava i5 consumed principal1y in the form of fresh roots and gaplek, w1th these two forms being prepared in a variety of forms in the home. There is a marked regional variation in consumption patterns of both fresh roots and gaplek. Although per capita consumption leve1s for cassava are the same for Java as the outer islands, fresh consumption is much more important off-Java, probably due to the less seasonal nature oí root production and the greater difficulty in drying. Gaplek consumption i5 12 concentrated in the eastern part of Java, where soi1 and rainfall are more marginal (Figure ), while fresh consumption on Java is relatively more evenly distributed. The importance of cassava in the diet and the relatively ubiquitous distribution of fresh root consumption implíes that qualíty characteristics cannot be sacrificed in a varietal development program. !be locus of cassava consumption is very much in the rural sector, due not only to the bulk of the populatíon residing in rural areas but also to the much higher per cepita consumption of casseva in these areas. !bere i8 a significant change in consumption of non-preferred staples between rural and urban areas (Table ) • Gaplek and maíze are rarely consumed in an urban setting and yet are quite important in rural areas. Fresh cassava consumption, while higher in rural areas, nevertheless i8 still at significant levels in urban areas, even given the problema of marketing such a periahable commodity. Unnevehr (1982) estimates that in rural areas about two thirds of fresh cassava and one-half of gaplek are subsistence consumption. Counting urban consumption, only 37% of fresh cassava thar is utilized for human consumption is marketed. Probably the most important component influencing the distribution of cassava consumption ie income. Gaplek consumption shows a consistently declining trend w1th 1ncome (Table). Gaplek 19 a non-preferred food, principally consumed by the poor. Fresh CassaVa consumption, at least in rural areas, increases markedly with increas1ng income at low levels of lncome, levels off at medium income levels, and declines slightly at high lncome levels. !be overall tendency is for total cassava consumption to decline with incorna. Approximately 40% of· the population in Indonesia consumes less than 1900 caloríes per day (Tabla ). This group 18 obviously constrained by income in the amount of food which they can purchase and thus must make more use of cheap calorie sourceS. The poorar income groups, princ1pally in the rural areas, substitute cassava and maize for the more expensive, but more highly preferred, rice (Figure ). Cheap cassava allows the lower income segments of che population to achieve a higher calorie intake with their limited food budget than they would have been able to achieve with just rice. Cassava is thus a potentially key commodity in polic1es focusing on nutrition and the related issue of rice import management. !be role of cassava within an overall nutrition policy follows from an analysis of demand parameters. Estimates of income elasticities by Dlxon (1982) show that among the poorer income strata there ls a significant increase in cassava consumpt:l.on, both as fresh and gaplek, >lith inereases in incorne (Table ). Such changes in cassava consumption could come from real increases in income or from changes in the rice price, ainea expenditure on rice makes up such a larga part of the consumer budget. Substantial substitution between caloric staples would be expected depending on relat:ive pr1ces and in ·fact, elasticity estimates suggest substantial responsiveness to price changes. Timmer (1980) reports a cross 13 price elasticity of fresh cassava with rice ~ 0.77, showing a very marked effect of rice prices on cassava consumption . Cassava's role in the Indonesian food economy, while not central, is nevertheless critical to the support of that proportion of the population facing a risk oí not meeting their calorie needs from riee supplies. This population is essentially defined by low incomes and in years of poor rice status can be put further at risk by rising rice prices. The government's po11cy has been to try to maintain stable rice prices and this task is vested in the government grain marketing agency BULOG, which attempts to stabilize rice prices through rice imports, and to a certain extent through wheat imports. BULOG was aided in this effort in the last decade and a haH by the widespread adoption in the irt:igated areas of the high yielding rice varieties. Nevertheless, rice imports have almost consistently exceeded one million tons and have occasionally almost reached two million tons. At these levels Indonesia can account for as much as a third of the world export market, having a pronounced affect on world rice prices and, therefore, the foreign exchange costs necessary to meet import requirements. As the benefits of the new rice technologies start almost certainly, to plateau, Indonesia will be faced with even higher import requirements in a world rice market that i5 very thin. To resolve this dilernma, Indonesia has increasingly turned to wheat imports, which are both cheaper and a minor percentage of the world market. However, Indonesia has on the whole failed to consider the potential role of the secondary staples, cassaVa and maize. Total consumption of both of these cornmodities has essentially been sta tic over the past decade and a half, implying a deelining contribution to total calorie consumption, since rice consumption has risen dramatically. Since there are real supply-side constraints on meeting nutritional objectives with rice, since the locus of wheat consumption is principally in urban areas, and since cassava and maize are aIread y important staples for the rural poor, a strategy to inerease production of these crops at lower prices (that is, technical change) would contribute directly to increased calorie consumption of the most vulnerable pupulation. By integrating cassava into overall food policy, BULOG would have considerable more flexibility in managing rice imports and prices. However, because of the overal1 inelasticity in food demand for cassava, this flexibility ia dependent on diversifying end markets. That is, diversifying the end uses as the production base expands not on1y provides a certain market stability far farmers but as wel1 ensures alternative food supplies when rice i8 in short supply. The starch market Starch is the largest single market (on a root equivalent basis) for cassava in Indonesia. A cassava starch industry has existed on Java since 2 Dixon (1982), on the other hand, could find no significant cross price elasticities but based his estimation only an Java, whereas Timmer's was based on Indonesia as a whole. harvests, their nutritional 14 the turn of the century. Prior to World War 11 and independence, this industry was based principally on plantations and was geared principally to export. The recovery from the da1l!age incurred during the war induced a shift from foreign to domestic ownership, which in turn entailed a shift frem export to domestic markets. Indonesia ls currently the largest producer of cassava starch in the world, and essentially all the production i8 destined to domestlc markets. Unlike oth:F countries in Asia, there i8 v1rtually no production of starch frem maize • The structure of the cassava starch industry i8 characterized by great diversity (Table ). Starch factories are spread throughout Java and Sumatra, but with a particular concentrat1on in We8t Java. Location of the starch industry i8 pr1marily dependent on access to a ready water supply, to a sufficient concentration of root production, to adequate transport infrastructure, and to non-seasonality of root supply. These factors have until recently given the edge to ;lest Java as the center of starch production. However, as transport infrastructure has improved on Sumatra, particularly in Lampung, starch production has expanded rapidly. This has been enhanced by the less seasonal supply oi roots on Lampung. From virtually no produetion in the early 1960's, the starch industry on Lampung has expanded rapidly. especially in the 1970' s, to become the seeond largest starch-producing province after West Java. Diversity is also a characteristic of the scaIe of processing. Rudimentary, household processing techniques co-exist with large-scale, capital intensive factories, with a significant range of plant slzes between these two extremes. Nelson (1982) has recently analyzed the economics of starch production in Indonesia. At 1980 priees a11 precessing modes wsre found to be profitable (Table ). The large milIs were found te be most profitable, but only because the tax incidence was much less than on household production and medium-scale factories. To motívate investment the government has instituted tax holidays for three to six years for large-scaIe firms. This, together with a subsidy on diesel fuel and exemption from import duty for imports of processing equipment, give a distinet advantage to insuring the profitability of the large seale planto However, from a social point of view, Nelson finds that the household production genera tes bath the highest level of social profit as well as the most employment. Nelson further reports that household stareh productien has expanded rapidly in the 1970' s, motivated by the inereased capacity utilization fram the introduction of mechanical graters. The few figures on starch suggest that production has increased rapidly through the 1970 f s (Table ) • This growth "as charaeterized by sign1ficant inereases in household production en Java and very rapíd growth of large-seale processing on Lampung. The starch market was both large and growing, provídíng quite strong demand for cassava roats. Root productien, at least on Lampung, responded accordingly. 3 A single stareh/corn oil plant, Indocorn, is operating in Indonesia. It principally relies on maize imports for its operatian and was nat in operation in 1984. 15 The factore that were driving this increased demand for cassava starch are less well documented. Concensus seems to exist that the largest end use for starch is as krupuk, a crispy wafer consumed as a snack food. Nelson reports that this industry takes as much as 65% of total starch production, while the rest goes into other food proeessing industries (15%), the textile industry (10%), and glueose production (3%). !he on1y complementary data comes from the SUSENAS consumer budget surveys. !he 1976 survey reports an average annua1 per eapita consumption 1evel of starch of 6.9 kg on rural Java and 0.7 kg in urban areas of Java (Duon, 1984). Assuming only 2.0 kg in the rural areas off-Java and the same level of consumption (0.7 kg) in urban areas off-Java, leads to a total starch consumption as food of 587 thousand tons, based on 1980 population estimates. !his figure is 89% of the total starch production figure estimated by Nelson. On Java more cassava (on a root equivalent basis) is consumad for food as starch than as fresh roots. Moreover, Dixon (1984) suggests that the SUSEl'IAS data significantly underreport starch consumption, sinee the starch equivalents of direct purehases of krupuk and bakery goods are not included. Dixon estirostes per capita consumption figures of 12 kg in rural Java and 5 kg in urban Java. However, these figures result in aggregate consumption 1evels of 935 thousand tona of starch for Java alone. These data would suggest that household production of cassava starch is undereatimated, which in turo would account for a large part of the discrepancy between cassavs production and consumption estimates on Java. A large and relatively diverse cassava starch industry already exists in Indonesia, moreover, the limited evidence on demand suggests that this market wi1l continue to grow tor a significant period into the future. Most of this growth comes from the use of starch as a food souree, with eonsumption in this case being skewed toward the higher incomestrata. Dixon (1984) estimates income elasticities for krupuk of 1.56 in rural areas and 1.35 in urban are as • Significantly, consumption patterns for cassava starch, skewed as they are toward che rich. are the mirror image of those for gaplek, which are highly skewed toward the poor. Product differentiation and market segmentarion a1low5 cassava in this case to serve two very distinet roles, as a basie secondary staple for the poor and as something of a luxury food for higher income groups. A unique featurÍ! ',of the cassava stareh industry in Indonesia. comparad to that of the other countries in Asia, is that there is no effeetive competition from maize stareh, even though maize is a major crop in Indonesia. The situation is further eonfounded by the fact that maize is, at leas e intermittently, exported at world prices, while gaplek, while also exported, competes at the higher price levels set in the European Community. Maize should thus be more competitive as a raw material source for starch produetion than cassava. However. in the particular case of Indonesia, starch substitution i5 limited by quality factors, and, in particular. course, sun-dried starch ls necessary in preparing krupuk, the domlnant market. The fine, flashdried starch cannot be used in krupuk unIess mixed with the eoarser starch. Thus, maize starch was constrained to competing in the much smaller, industrial market with eassava starch produced in the larger factories, and, given the scale economies in wet 16 milling, maize could not establish a large enough market to justify a factory. Nevertheless, the competition between maize and cassava becomes a factor in the recent interest in the production of high fructose sweetners. Indonesia has over the past decade consistently increased its imports of sugar to the point that imports now total between 500 to 700 thousand tons ayear. Not only are imports increasing but Indonesia maintains high internal sugar prices to support producers, on the one hand, and to limit consumption, on the other hand. A policy directed at self-sufficiency in sugar is limited by the availability of land suitable for sugar cane and the competition between rice and cane for this land. Therefore, producing high fructose sweetners from either maize or cassava in upland areas holds sorne attraction. However, the substitution of liquid high fructose sweetners for sugar occurs over only a limited range of end uses of sugar. The largest market, direct human consumption, has limited possibilities for substitution at this stage of market development. Development of the HFS market depends on exploiting industrial uses, ,especially food processing and bottled beverages. Estimates on the size of this market are based on scanty data; two sources put the potential consumption at between 220 and 500 thousand tons per year (Argento and Wardrip, 1983; Tate and Lyle, 1981). Moreover, this market is expected to grow at a estimated rate of 5% through the rest of the century (Pearson, 1984). Indonesia has already committed itself to producing high fructose sweetners. A cassava-based factory with an annual capacity of - tons is already in operation in Malang on Java. Licenses for the construction of 4 more factories have been issued to bring total production capacity to 110 thousand tons of HFS. Nevertheless, two basic factors will largely determine the future of this industry. First, the economic viability of high fructose sweetner production will necessarily rest on the maintenance of the high domestic price level for sugar. Domestic wholesale prices for sugar in 1984 were $.57 per kg, compared to a world market price of $.15 per kg and the medium term prognosis for world price levels to rise only to about $.26 per kg. Second, licensing procedures and subsidies on capital investments will be critical in determining whether sweetner production is based on cassava or maize. The economic advantage of one crop over the other is difficult to project with any degree of certainty but the most complete cost analysis to date is that of Pearson (1984). The Pearson concluded that maize would be a lower cost alternative than cassava in HFS production due to three principal tenets. First, there are significant economies of scale in the maize wet milling process, while in cassava these are minimal. Second, the price distortions in the world market for cassava relative to maize are assumed to persist and will in turn influence domestic profitability. Third, domestic production of maize is projected to increase significantly on the basis of an improved hybrid technology; should this ,technology not produce increased yields, imports will have ~o -i,ticrease markedly to meet increases' in demand for maize for both feed concentrates and HFS production. 17 Nevertheless, planning of the HFS industry has been based on cassava for several practical reasons. First, HFS production based on cassava i8 profitable under present domestic sugar prices as set by BULOG. Second, expansion of cassava production does not depend on yield increases but can be based on further area expansion in the off islands, especially those with good infrastructure as in south Smru.tra. A supply response is much more assured in the cassava case. Third, capital requireroents for RFS production are significautly less as a HFS production liue can be added to _ existing cassava starcb factories, as was done in tbe Malang case. Tbis al10ws a more evolutionary and 1ess risky approach to roarket development since production can initially be based on rela.tively sroall sca1e plants that have alternative product linea and not on major capital investments in large-scale, maize wet milling plants. Basing HFS production on cassava al10ws significantly more flexibility in market development than does maize. For cassava-based RFS, factories can be located in cassava production areas and based on starch slurries from the direct root processing or alternatively can located next to major market areas and use processed ~tarch as a raw material. Relative transport costs and control over raw material costs will determine the choice. Maize, wet m111ing plants w11l probably be located near to consumption points and wi11 depend on steady supplies of maize froro major Btorage facilities or imports. In this regard maize- based RFS will be competing with the animal feed industry for raw material supplies, most of which is currently supplied to the concentrate industry from BULOG stocks which are often imports (Table ). Cassava's potential role in this industry will thus be based on BULOG's sugar price policy and on the relatively immediate demonstration of impact from the improved maize technology. Gap1ek in Feed Markets Gaplek forms an integral part of cassava production and market systems in Indonesia. When properly dried, gaplek ls a stable commodity and provides the farmer the aption of harvesting and storing his cassava especially wben there 15 a time premium on harvesting the cassava to plant the next crop. Moreover, gaplek, since it can be stored and transported, provides a means of integrating cassava markets. Finally, gaplek has roultiple uses; it can be used directly for human consumption, can be ground into flour for noodle production, or can be a raw material source for feed concentrate production or even for manufacture of low quality starch and its derivatives such as glucose or fructose sweetners. Gaplek is currently used principally for human food, especially by the lower incorne consumers in rural areas. Indonesia is also a consistent, although highly variable, exporter of gaplek to the European Community. This export market serves the very important function of setting a price floor under domestic prices for gaplek and in turn cassava in general (Unnevehr, 1982). Tbe export market is effective in setting this price floor, even though this market rarely accounts for more than 10% oí cassava production. Only twice since 1970 have gaplek exports exceeded 400 tbousand tons (Table ) and export levels more gene rally osci11ate between 150 and 350 thousand tons. 18 Internal gaplek prices have in general followed the general rising trend in world prices (Figure ), with exports being particularly responsive to the devaluation of the rupiah in 1978. A similar devaluation in 1983 has yet to produce such a response. This apparent tightening of domestic markets is especially evident in Lampung, where the gaplek export market was the engine of growth for the cassava industry in the first half of the 1970's. Gaplek exports from Lampurg stagnated after 1975 and have declined markedly since 1981. The gaplek industry has had difficulty competing with the expanding starch industry on Lumpung, even when world prices were recently relatively high. This declining trend was exacerbated by the poor crop years in 1982 and 1983. The tightening of export supplies of gaplek have made the voluntary quotas, formalized with the EC in 1982, rather superfluous. The quota was set at 500 thousand tons in 1982, rising to 825 thousand tons by 1986 when the agreement ends. Compared to the Thai quota, which declined over the period, the Indonesian agreement was very much largesse, even though of a very gratuitous kind. There is very little potential for meeting the quota volumes, even with the 1983 devaluation. The advantages of the lattet were negated by abad crop year and the 1984 fall in the world price, brought on by the effect of the quota on the Thai cassava industry. The current level of the gaplek export market undervalues its importance. An export price floor set in the EC not only earns Indonesia a significant economic rent but also serves to maintain price incentives should future production growth increase. New cassava production tech­ nology or further transport infrastructure development on Sumatra could bring about such growth and the export market could serve to buffer farmer prices were production growth significant. The short term problem with current strong domestic markets for cassava is to maintain sufficient pelleting and export capacity to insure the world price linkage. The medium term problem is to insure that a sufficiently large quota in the EC market is maintained to allow the cassava industry to expand without significant price instability. Certainly, in renegotiating the quota agreement, the short-term problem should not militate against the longer term gains from maintenance of export flexibility. The maintenance of the world price export floor for gaplek, while earning significant rents for Indonesia, nevertheless is one factor inhibiting the development of gaplek as a raw material source in mixed feed. Since gaplek prices are set in the EC and maize prices are to a degree linked to the world coarse grain market, gaplek prices are often out of line with maize. Nevertheless, as table demonstrates, calorie prices of gaplek are often as not competitive with maize. Two additional factors milita te against gaplek use in balanced feed rations. First, there is a preference for maize because of its carotene content, which gives the eggs and poultry meat a yellower color. Second, BULOG can be relied on for supplies when these are DOt available on the local market, especially since the major milIs are located near to major urban areas (Table ). Third, BULOG has recently brought soybean meal imports under its control, principally as a means of regulating foreign exchange. Although BULOG continues to change prices in line with world market prices, in 1983 it decided to cut imports by a half to save foreign exchange. This resulted in milIs importing rapeseed and sunflowerseed meals, which are not 19 currently controlled. Any limitation on protein supplies would give a relative advantage to maize over cassava. The balaneed feed/cornmercial livestock sector is not as well developed as similar industries in such countries. as Thailand or the Philippines. This 18 principally due to a relatively late start, as the first feed factories vere only established in 1972. Hovever, the other structural features of this industry are very similar. Growth in mixed feed production has been spectacular, rising from essentially no industry in 1972 to an estimated 400 thousand tons in 1982 (Alfred C. Toepfer Company, private communication). About 85 to 90% of production is poultry rations and the commercial poultry industry has grown in close aS8ociation with the feed sector (Table ). This growth in the poultry/mixed feed industry has been motivated by increasing demand for meat and eggs, preeipitated by rising per eapita incomes during the 1970·s. In sum a viable poultry/mixed feed industry has been established in Indonesia with prospeets for continued future growth as reflectad in the high income elssticities for meat and eggs. A factor that may be a constraint on growth in the poultry industry, and by implication for the mixed faed industry, is the presidential decree limiting the size of layer units to 5000 birds and of broiler operations to 750 head per week. The objective of the decree is the maintainence of a labor intensive poultry industry and a more equitable distribution of income opportunities. For the feed industry, per se, the decree in effect expands their market, since the large poultry operatíons m1x their own feed. The principal effect wi11 be on costs of eggs and poultry meat, sinee the larger producers are able to achieve higher feed conversion rates and fewer losses. Mink (1984) estimates the result of such a shift to small producers wi1l be an annual reduction of 35,000 tons in demand for carbohydrate sources. Between 450 (World Bank, 1984) and 700 thousand (Mink, 1984) tons of maize are estimated to be used as animal feed currently in Indonesia. This represents no more than - per cent of the total maize crop. No eassava i8 currently used in the animal faed industry snd there i5 little potential of entering this market as long as gaplek prieee are set in the EEC. Moreover, improved hybrid varieties, the first released by Cargill in 1983, are thought to have significant potential for increasing production above domestic requirements. This yield impsct however, remains to be demonstrated on a widespresd scale at the farro level. Nevertheless, only with a marked changa in relative priess wi1l cassava be used in animal feed, and this appears unlikely as long as the export price floor remains effective. Given the growth potential of other markets and the social profits derived from exports, such a situation continues to be advantageous for cassava. Pricing and Market Efficiency The Indonesian cassava economy represents in many ways the ideal development of the crop; that is, cassava i6 deployed within diverse and complex cropping systems across a range of agroclimatic conditions and i8 fully utilized in a broad spectrum of end uses. Such full exploitation of the production and utilizatíon potential of the cassava crop relies fundamentally on well functioning markets and in partícular on integrated LV markets in which prices serve to allocate cassava between the range of end uses. That is, farmers are receiving a price for their cassava roots that reflects its best end use in the country. Such a situation requires that cassava prices be linked spatially across the country and linked vertically across different forms. The development of such linkages for a highly perishable, bulky commodity i5 difficult and i5 dependent on the existence either of a highly developed transport, refrigerated storage and marketing system (eg. vegetables in the U. S.) or processing of the roots to a stable, storable commodity. Since the first does not exist in Indonesia, the role of gaplek can be singled out as crucial to well integrated cassava markets in the country. Unnevehr (1984a), (l984b) has analyzed market integration and price transmission on Java and what follows is drawn directly from that research. The key to her analysis i5 the concept that "cassava prices within Java are set by domestic supplies of staple foodstuffs and demand for cassava products, subject to a lower bound set by export parity the local demand curve for cassava has two portions - a downward sloping domes tic curve and a pe1:fectly elastic export froor." (Unnevehr, 1984a). -A demand curve was estimated to test for this "kink". When East Java prices were at export parity the correlation with world market prices was 0.95. Gaplek prices at the East Java port, Surabaya, in the 1971-79 period were at export parity 79% of the time. This demonstrates the effective operation of the price floor and the fact that the export market was a principal determinant of domestic prices throughout this periodo This is seen in Figure ,charting world and Indonesian gaplek price5. Effective price transmission and adequately linked markets implies relatively. competitive price formation throughout the country. This, however, does nat imply that all farmers face the same price since transport and marketing costs will differ radically depending on location relative to markets and the level of development of transport infra­ struture. In fact marketing and transport costs make up a very significant portion of the wholesale or retail price for both fresh roots and gaplek. Assembly costs of fresh roots for starch plants and gaplek for pelleting plants are relatively high, compared to the eventual farm level price (Table ). On Lampung assembly costs alone consume half of the factory price paid for roots and 40% of the price paid for gaplek. This significantly reduces price incentives for farmers, since the complete marketing margin (farmer to retail) for alternative crops is only around 30 to 40% (Table ). The effective operation of the export price floor under domestic cassava prices throughout Indonesia, Moreover, depends critically on spatial integratian of the various cassava markets. Such integration relies on two components, first, integration between fresh root and gaplek prices and, second, between gaplek prices in different markets throughout the country. In terms of the linkage between fresh root and gaplek prices, variation in fresh root prices explained over 90 percent of the variati~n in gaplek prices in 7 of 19 markets on Java and over 80 percent of the variation in 18 of the 19 markets (Unnevehr, 1982). 21 Not only were gaplek and fresh root priees strongly 11nked but there was also a strong linkage of gaplek priees between markets across Java, and this linkage was prineipally due to the operation of the export priea floor. Thus, when domestic prices were at export par1ty the eorrelation coefficient of gaplek prices in the 19 different markets was greater than or equal to 0.90 for 106 of 171 potential paira. On the other hand, when domestic prices were above export parity, only prices in 27 pairs of markets were eorrelated at the level of 0.90 (Table ) . When domestie prices were at export parity, domestic priea variation of gapIek was due aImost eompletely to variation in the export priee (Unnevehr, 1982), and sinee there was a generalized priee linkage between markets and between roots and gaplek, the operation of an effective priee floor was demonstrated for Java as a whole. When domestie priees rose above export parity, price variation was much more influenced by regional supply and demand eonditions for eassava. Moreover, internal transportation costs tended to lower the export floor for more remote markets, inereasing the influence of local supply and demand conditions. Thus, the number of months the priees at 19 internal markets were at export parity varied frem 32 to 70% of the time, all less than the 78% at Surabaya. Nevertheless, what ls remarkable is how often domestie priees have been at the priee floor. In the period 1971 to 1979, monthly prieea in major markets were at export parity between a third to four-fifths oí the time. Produetion in this periad grew at an annual rate of approximately 2.8%, at ~ time whan population growth was 2.0% and income growth was 5.3%. Normal grawth in food demand for cassava (assuming a combinad income elasticity of 0.1) snd the rapid growth in starch produetlon, should have put soma upward pressure on cassava pricas. Moreover, never mOre than 15% of domestic production was exported and the figure "as usually les s than 10%. Surpluses, at export priees, thus, were never that large. Part of the reason ls that there was a general upward trend in export prices. However, the other major factor affeeting cassava prices is the domestic price of rice and over this period the real price of rice fell substantially (Figure ) due to the impact of lmproved rice technology and import poliey. Timmer (1980) finda a cross-prica "elastieity batween cassava and rice of 0.77, indieating slgnifiesnt decreases in cassava consumption for a decline in rice prieas. During the period of rapid expsnsion in rice supplies the eassava export market served s critical funetion of providing an effactive price floor and thus msintaining ineomes of cassava farmers. As Indonesia exploits most of the yiald gain possibla frem the rice teehnology, domes tic rice prieea and rice imports are again likely to beeome important policy issues. Cassava, because of this priee linkage to rice, a110ws additional flexibility in meeting rice price policy objectives. In the future, improving cassava production may be a far less expensive means of maintaining rice priees than ri"ce imports. Any cost reductions in transport or scaIe economies in assembly wil1 tend to favor cassava over other eropa. On the other hand, to assembly costs must be added processing costs. Both the gapIek and stareh processing industry has been found to be socially efficient (Nelson, 1982). Only about a quarter of the export par1ty price for both starch and pelleta 22 1a consumed by processing costs (Table ). The cassava processing industry 1s relatively dynamic snd as well permits a significant degree of diversity. Labor intensive, household starch production co-exists with capital intensive, large scale factories. All are profitable, although government tax and capital credit policies tend to favor the large-scale plants, when the household unfts are social1y more efficient and employ significantly more labor (Nelson, 1982). Cassava marketing systems in Indonesia have evolved in response to transport infrastructure development and changes in market demando There has been almost no intervention by government agencies apart from the tax credits for large scale processing plants snd the import tax on starch. As the evidence suggests, cassava markets function very efficiently in Indonesia, given the constraints improved by infrastructure. There ls not only little need for government involvement in cassava msrkets, but unlike rice, any su eh 1ntervention 1n a commod1ty w1th multiple markets would be counter productive without a comprehensive policy and this would be difficult to attain. Unlike many other countries in Asia, lndonesian cassava markets reflec~nat10nal supply snd dernand conditions with a buffer~ provided by the export market. Further development of cassava in Indonesia will be relatively easy given such a well functioning market1ng system. Conclusions Growth in the Indonesia economy has been impressive over the decade of the 1970's, cont1nuing through to 1982. GDP growth averaged 7.6% per annum in the 1970's and was aboye that mark in 1980 and.1981. These growth rates were we11 aboye the average for either industrial or developing countries. Only in 1982 did the economy start to be affected by the international economic recession and GDP growth fell to 2.3% rebourd1ng to around 4% the following year. The decline in 011 pr1cesand demand for agricultural exports led to a significant decline in the foreign exchange reserve position, culminating in a devaluat10n of the rupiah in 1983 and 1986 and tighter controls on imports. Future growth in the Indonesian economy i5 highly dependent on what happends in the petroleum export market; nevertheless, the economy 1s projected to grow by 5% per year through the rest of the decade (World Bank, 1984). Such significant growth in incomes have a market impact on food demand. Extimated annual per capita consumption of rice increased from 107 kg in 1970 to 145 kg in 1983. Fortunately. rapid demand growth corresponded with the rapid adoption of ahart stature rice technology and rice production almost doubled in this period, even with very minor change in the land area planted to rice. Nevertheless, Indonesia remained a major net importer of rice, importing as much as 2 million tons in 1980. Growth in production of rice i5 expected to slow somewhat through the end of the decade, as the growth rate in yields declines. However. Indonesia is expected to remain at or near self-sufficiency in rice while continuing to maintain some capacity to import when production deviates from trend (World Bank, 1984). Indonesia has been relatively successful in attaining self-sufflciency in the production of basie foodstuffs and in maintaining relatively atable consumer pricea, especially for rice. While the government has been 23 successful in meeting two of its food policy objectives, impact on raising farmers incornas, the third principal food policy objective, has been less widaspread. This 18 because the income generation from the new rice technologies was directed almost exclusively toward the irrigated sector. The benefits from the new rice technologyhave been inequitably distributed between regions and since the bulk of the population countinues to depend on agriculture for their income, coninued neglect of the upland areas viII further increase these disparities. Two principal conceros should govern policy toward the upland sector. The first is the relative priority between development of the upland areas on Java and those on the outer islands. Java accounts for 41% of Indonesia's GDP, 62% of the population and only 7% of the land area. The soils on Java are relatively fertile, transport infrastructure i8 relatively vell developed, and very labor intensive production systems have evolved to suit the extremely small average farm size. On the outer islands, on the other hand, the soils tend to be infertile and highly acidic and infrastructure is not as highly developed. Land ls relatively plentiful. The population distribution between Java and the outer islands creates, a situation where both land and labor reseurees are underutilized and the transmigration projects were established to remedy this imbalance. Between 1971 and 1980 approximately 2.1 milIion migrants resettled in the Outer Inslanda, of which one million ",ere resettled through the transmigration program. This program had a significant impact on agricultural employment. Of the 1.8 million increase in agricultural employment in this period, 1.4 million was off Java (VIorld Bank, 1982). Certainly any incraase in area planted to crops will have to come en the Outer Islands and the government 18 currently attempting through agricultural research estate development, and the transmigratíon projects to establísh a base for future growth on the Outar Islanda. The second íssue is the choice of crops, where technology can be expactad to. raise productívity and markets are suffcíently expansive to absorb the increases in production, thereby leadíng to increases in farmer income. Certainly cassava must be considered as a principal choice for both Java and the Outer Islands. Maize is an al ternative choice on Java and tree crops are an alternative on ths Outer Islands. However, cassava could have the widest potential impact of these crops, given a higher committment of resources to support research on ths crop. As a crop development of the upland areas, cassava has several advantages. Most importantly the cassava marketing systern in Indonesia 1a prebably the best developed in Asia, with the possible e"ception of the larger but more spec1alized system in Thailsnd. Prices efficiently allocate cassava between regiona, across different and uses, snd ovar time. Moreover, and effective price floor i5 provided by the gaplek export market. Effícient markets toge~her with the multiple and uses for cassava, particularly the high consumption oí gaplek and fresh cassava by the poor, allovs the introduction of improved production technology to achieve the dual policy objective of increasing farmers' incornes and improving calorie intake of the rural peor. Moreover, the rapidly growing starch market, with potential under current policíes for the development of high fructose sweetners, provides scope for the absorption of signiflcant increases in production, with any surpluses up to the EEC's 825 thousand ton quota restrict10n being exported. Nevertheless, the very uncertain situation in the EC market for cassava pellets will continue to affect the Indonesian cassava economy, if not in lower import quotes when these are renegotected in 1986 then in the impact on world prices and the impact that lower world prices will have on Indonesia farmers. There is some opinion (World Bank, 1984) that Indonesia will be in a surplus position in both maize and cassava by the end of the decade, with little hope of absorbing these production increases in domestic markets. For cassava the report overlooked the large and dynamic starch market, but certainly any major productivity increases will probably result in internal prices remaining effectively tied to the export price with the accompanying need to maintain some flexibility in the export market. Certainly there are trade-off in maintaining this price linkage to the EC market. The gains are in the social profits reaped by the high export prices; the costs are that cassava cannot compete with maize in certain domestic markets, especially the animal feed market. It remains to be seen whether these domestic surpluses of secondary carbohydrate sources develop and- to a large degree the -advent of such surpluses "ill depend on what happens in the rice sector. All in all there is no need to intervene in casesava markets until major breakthroughs are made on the technology front. At that point the maize and rice situation together with cassava production coste will dictate whether the cassava price should bealigned with the domestic and presumably world maize price. Until that time there are losses in the social profit for cassava if forced to compete in domestic maize markets. Providing resources for cassava research is a medium to "long term investment and more than anything else a dynamic cassava sector provides flexibility in Indonesia's food and agricultural policy. When rice yields start to plateau out at the end of the decade, cassava can add flexibility to price and import policy for rice. Moreover, the starch, high fructose sweetner, and, when necessary, the export markets can be a basis for expanding cassava on the outer islands, agricultural areas where a well adapted cash crop for smallholders has been difficult to identify. This type of flexibility will be key for balanced agricultural and industrial development in Indonesia's future. MALAYSIA Tbe agricul tural eeonomy of Malaysia, like that of Thailand, has traditionally been export-oriented. Export growth has relied on the faet that Malaysia has always been a land surplus economy, and at several points in ita history even had to rely on immigration of both Chinese and Indiana to meet rising labor demand in agriculture and mining. Export orientation within a land surplus economy put a premium on the development of an effective land policy. In this aspect, Malaysia differed from Tbailand in that the foeua of land poliey was on promoting large-scale, plantation agriculture, although land availability did not preclude the development of smallholder agriculture, both for the production of rice and export crops. A focua on plantation agriculture has remained a primary component of agricultural policy to the presento Cassava was the first of the series of export crops that have spraad across Malaysian agricultura. The establishment of the first tapioca factory in Malacca in the early 1850's coincided with the rapidly expanding use of commereial steamships. The evolution in sea transport together with the ope1!ing of the Suez Canal in 1869 opened European markets to other agricultural eommodities than just high valued spices. Ibe tapioca industry expanded rapidly and relied on eassava's particular advantages as a frontier crop. The forest was cleared to feed the steam engines of the plant, while cassava was planted in a shifting cultivation sytem characteristic of a land-surplus, labor-searce economy. This produetion system. which ostensibly took place wíthin s plantatíon-type land eoncession but where the land was abandoned to lalang when soíl fertility declined to unprofitable levels, gave cassava the image of a soil-depletíng crop, especially compared to the rapidly increasing tree crops. Altbough soil depletion was due more to the shifting cultivatíon system than to the crop itself, this image has remsined upto the present, resulting in controla on cassava expansion through restrictions on land coneessions snd laases. The oscillations in the export market for tapioca snd stareh, land policy, and competition witb export-oriented, tree crops have remained the key factors influencing the Malaysian cassaVa industry to the presento Production Trends Cassava productivn in Malaysia hss never repested the boom period of 1860-1890. In Malacca cassava area c1imbed from virtually nothing to around a peak of 30 tbousand hectares in 1882. In the 1870'1 s eassava area had also began to expand into neighboring Negri Sembilan, reaching its peak areas in the 1890's (Jackson, 1968). Area planted to cassava in this early periad probab1y did not exceed 45 ~housand hectares. The cassava industry fluctuated yith the prices on the wor1d market through to the turn of the century but then got caught in a squeeze between the rapidly expanding rubber industry in Ma1acca and the deve10pment of an export oriented cassava industry en Java. These trends were remarkably rapid. In 1906 thare was 15 thousand heetares plantad to rubber in the Straits Settlement Provinces (Malacca and Provinee Wellesley and Penang) versus 43 theusand hectares planted to cassava. In the same year Java exported a 1ittle over 6 thousand tans of cassava products. By 1913 rubber area had expanded to 64 thousand hectares in the Straíts Settlements and Javanese exports had inereased to over 90 thousand tons. Cassava area in the Straits 2 Settlements declined to only 6 thousand hectares (Greenstreet and Lambaurne, 1933). After this majar structural shift, cassava area oscillated between 10 and 20 thousand hectares over the next 70 years ti11 the present (Table 1). The other major e1ement in this stagnation of the cassave industry was the restrictions on land concessiones and actual planting of cassava by many of the states. Thus, Negri Sembilan prohibited planting of cassava in 1912, Perak restricted plantings in 1909, and Selangor did the same in 1925. In Kedah in 1905 cassava was allowed only as a catch crop for tree crop establishment (Greenstreet and Lambourne, 1933). Thus, in the period between the two world wars, the cassava industry shifted to Johore, where there were no restrictions on cassava, and Kedah, where it was grown as a cateh erop. The shifting nature of the cassava industry continued, since following the Seeond World War, and especially after the 1958 Emergency, cassava rapidly shifted to Perak, which is the locus of the industry today. Nevertheless, land - policy continued to play a domina te role in the organization of production. In particular, Aw-Yong and Moo:!. (1973) estimated that in the mid-1960' s appraximately 75% of the eassava :I.n Perak was planted illegally on unalienated state land or torest, railway, or m~n~ng reserves. As a result, shifting cultivation remainad the dominant production system for cassava. Sh:!.fting cultivatíon systems and che uncertainty of access to land for cassava are possibly reflected in recant trands in production (Table ). In cassava are a thara is sígnificant variation around a relatively stable trend of 16 thousand hectares. Yields a1so are high1y variable, rang:l.ng from 11 to 22 t/ha .• >lith no necessary tendency for variaríon in area to compensate variatian in yield. Productian, as a result, 1s high1y variable. However, this year-to-year variability is not reflected in the output of cassava products. Converting starch and chip production to fresh root equivalent, shows a consistent rige in root utilization through the early seventies and a decline from the 1976 peak over the latter part of the de cad e (Table ). A comparison of the two series suggests much more stability in the utilization series and a consistent underestimation of utilization when using the productíon series. Given the large percentage of illegal plantings, the production series probably does not capture all the actual area planted to cassava. On balance there 18 probably muchmore stability underlying the Malaysian cassava industry than is reflected in production statistics; on the other hand, over the last half of the decade there has been a persistent, decli.ning trend in eassava production. Cassava Production Systems Cassava's principal comparative advantage vis-a-vis other eraps is its adaptation to relatively marginal agro-climatic conditions and therefare its exploitation of land with a low opportunity cest. Because tharc i5 no climatic constraints on crop production in Malaysia and tree crops are well adapted to a ",ide spectrum of tropical soils, cassava has no particular niche to expleit in the agricultural economy and must compete with tree CTOpS tor land. Thus, of the 25% of Malaysian land under cultivatíon, we11 over 80% i8 planted to the three principal tree crops, rubber, oi1 palm and coconut. Paddy land accounts for another 10%, leaving under 10%, for al1 3 other crops. Tree crops are by far the most profitable agricultura1 activities, and in fact, cassava i8 prirnarily grown in those areas where farmers do not have the option of planting oi1 palm or rubber. Land tenure primarily influences where and the type of production system that cassava is grown under in Malaysia. The more minar area where cassava is cultivated i5 as a catch crop in the establishment of oil palm or rubber. This i5 done principally by smallholders, although some plantíng of cassava as a catch cro]> by tree crop estates has also been reported (Lulofs, 1970). The caS5ava 1s p1anted for 2 or 3 seaaOns as a souTce of 1ncome .unt11 the tree crop is e8tablished. However. this i8 not a widespread practice and i8 1imited to those areas whieh have aecess to cassava processing plants. The major portion of the cassava 1s grown in monoeulture. This 1s in part due to the fact that a large portion of the erop is planted on land where the grower has no usufruct rights. Aw-Yong and Mooi (1973) in a study of cassava production in Perak in the .. id 1960's found that over 70% of cassava area was planted illegal~y. Illegal planting of cassava is done on a much more extensive basia than legal cultivation (Table ). Area planted i8 often done on a large-scale, sometimas exceeding 50 hectares. Wbera virgin jungle i8 clearad, all work i5 done by hand. However, with the rising costs of labor, areas covered with lalang which have the possibility oi mechanized land preparation are now probably eultivated more generally than virgin foresto This early study reports that most illegal cultivatíon is done within a systam of shifting agriculture, where the land is planted two or three time to cassava without application of fertilizar and then a naw area is opened up and brought under production. Wbether the rising labor costs of opening new land has caused even illegal planting to shitt to a more permanent, cultivation system is only open to hypothesis, but certainly the incentives are increasingly to shift to more continuous cropping, even within an insecure tenure situatíon. Legal production, on the other hand, ís concentrated in the hands of smallholders. Area planted in casssava averages less than 2 hectares and cassava is usually only one of several crops cultivated. Even in this situatíon cassava is often grown on rented land or on state land with temporary occupational licences. That is, there is sufficient uncertainly in tenure nat to plant tree crops. Also, cassava i6 often a component in the initial cropping system in those areas where farmers have recently been settled but have not yat invested in tree eropa. Thus, €ven for the legal planting, eassava i8 only planted in that land where investment in tree crops 18 risky. Nevertheless, production systems are much more stable. Rotational systems with other annual erops are often practiced along with applicat10n of fertilizer or manures. Over the last couple decades fert11ization has apparently shifted from farmyard manure and woodash (Aw-Young and Mooi, 1973) to reliance on chemical fertilizers (Tunku Mahmud, 1979). Moreover, with the rising COS1: of labor farmers have as we11 moved to the app1ication of herbicidas in order to control weeds. Rising labor costs and the competition with tree crops for land have put a premium on achieving low costs of produetion per ton. More intensive production methoda are now mOre eeonom1c than extensive produetion methods, as the emphasis has 4 shifted to lower labor costs and hígher yíelds. In effect, shifting production systems have become íncreasing1y uneconomic in Malaysía, making cassava' s reputatíon for so11 impoverishment mOre of an historical red herring rather ever than a point of fact. The other majar production system for cassava 15 plantations. In the early stages of the cassava industry these systems had their impetus in the forro of land concessions allocated by the stata governments. However, root production eperated en a basis of shifting agricultura and it was not tíll the advent of rubber at the turn of the century, that plantations based on permanent production systems were established. At this stage production of cassava on a larga seale declined. However, in the post-war period more permanent cassava plantations have been established, usually under government sponsorship. The motivation for plantations is usually to assure regular supplies to relatively large-scale faetories. However, the operations of large-seale, caSsava plantations have not met with much success. Of four plantations that have been operating in the 1ast decade, only one 1s st111 operating. High labor and overhead costs make plantation production much more costly than smallholder production within an industry that 15 highly competitive, both from other domes tic factories and international competition from Thailand. YieIds Cassava is grown purely as a commercial crop in MalaY5ia and moreover must compete with tree crops for both laud and labor. Yie1ds are therefore, a primary determinant of eassava' s economie viability in the country's agrieultural eeonomy. Not surprisingly, average yields in Malaysia are high by world standards or even by comparison to other Asian eountries. National production stat1stics suggest an average yield in the range of 11 to 22 t/ha. As has been suggested, the re11ability of these estimates are open to question. Nevertheless, the few surveys of cassava producers that have been carried out do support the higher end of this range of yield estimates. Tunku Mahmud (1979) found an average yieId of 28 t/ha in the Manong area oí Perak. Rahman Binti Adam (1974) found an average yield of 18 t/ha in a survey of farmers in Pahang. The point where these survey areas reside within the overall yield distribution for the country cannot be specified. Aw-Young and Mooi (1973) suggest in Perak a very broad yieId variation of between 7 and over 40 t/ha base on differences in soil and production system, where the production system as well reflects principally variation in soil fertility (Table ). Nevertheless, it was not possible to associate productíon weights with the different strata so that average yields could not be calculated. The fact that cassava is nat grown in continuous production sytems, as in othar parts of Asia, contributed to the high yields obtainable in Malaysia. Other factors are the favorable rainfall and growing season, the existence of relatively h1gh yielding varietíes, and the apparently wide use of fertilizer on cassava. However, defining the gap between average yields and the potential productivity of the crops remains uncertain due to Iack of reliable data on cassava. Costs oi Production and Labor Utilization Cassava Is a highly commerciali,zed crop in Malaysia. The erop i5 fully marketed, usually for industrial processing. Moreover, cash costs 5 form a high percentage of total costs, because most labor i8 hired, land preparation i8 mechanized, and input use is relatively high. Cassava farmars are thus reponsive to changas in input or output prices and likely to adopt technical innovations. Production costs and root prices ara therefore principal indicators of economic incentives that cassava producers face. Technology development and the evolution of costs hava reflectad the relative scarcity of labor in the agricultural economy. Where possible land preparation is mechanized, and tractor servicas are provided by farmers cooperativas. Moreover, herbicidas have assumed increased importance in cassava cultivation in order to reduce labor costs. Weeding and harvesting ara usually done on a contract basie. With this tendancy to reduca labor usa as much as possibla, labor input ie ralativaly low. A survay in Perak (Tunku Mahmud, 1979) found an average labor use of 62 mandays/hectare (Table ). Any further reductions will require the mechanization of the harvest. Labor costs"make up just lass than half of total production costs for cassava. Melaysia providas a counter example to the normal tendency for labor to make up the major portion of total production costs in cassava. Moreover. waeding is one of the more minar costs items. again running contray to norroal patterns. Land preparation, fertilizer costs, and harvesting a11 are usually larger cost !tems (Table ) • !be tandency toward labor substitution i8 clear in the cost structure; however. the scarcity of land forced both by government land policy and by high opportunity costs· has also put a premium on yield per hectare. as is reflected in the high costs for fertilizer. High yields, low labor input, and moderate input use, which is often subsisdized by the farmer cooperatives, result in a very low variable cost of production per ton of roots, comparable to that of Thailand. However, farro-level prices oí roots are normally higher in Malaysia than in !bailando !bis is principally due to the high opportunity cost of land. The annual net incorne for rubber was M$3651' (at a rubber prica of M$2.40!kg) and for oil palm was M$5030 (at an 011 price of M$1200!ton) (Tunku Manour and Sto Clair-George, 1979). !bis compares to an average net incoma for cassava in Perak of M$979 (at a root price of M$74/tons) (Tunku Mehmud, 1979). High supply prices for cassava in Melaysia reflect the profitability of alternative crops, which has provided sorne ímpetus to the search for higher yields and lower production costs but i5 primarily reflectad in the utilization of land with a relatively low opportunity cost. Technology Development Research of a rather aporadic nature has been earried out on cassava sinee at least the 1920 t s. The focus of this research was principally oriented to evaluation and characterization of imported clones and to appropriate fertilization of the crop. In the 1970's a cassava research program was established within the Malaysian Agriultural Research and Development Institute (MARDI). Cassava research broadened in acope at MARDI but continued to maintain traditional lines of amphasis. Germplasm evaluation waa expanded to include a major erossing snd selection programo The principal breeding objectives were high yield and snd high starch content of roots, reflecting che demands made by the starch aud chip market8. Agronomic research continued the long tradition of focusing on plant nutrition and maintenance of soil fertility. tong-term fertility trials and evaluation of nutritional requirements of cassava grown on peat soils became principal lines of investigation. !he few diseases of any potential significance were incorporated into the program as secondary screening objectives. Little direct impact of this research is yet visible on cassava yields. Fertilizer and herbicide use by farmers has signfiicantly increased but this i8 due as much to subsidies on these inputs as to the research that has been carried out. Breeding, on the other hand, is a longer term investment, and wh:!.le some lines have been identified which give superior yields to the dominant variety, Black Twig, none of these as yet has been released as a new variety. Emphasis on increasing yields i8 a .. e11 justified strategy under Malaysian conditions, given the need to achieve higher returns to land. A complementary strategy, on which there has been some research, is to direct technology to low opportunity cost land areas. Peat soi1s have be en one area where there has been some research. !he other area 1s as a catch erop in the establishment of tree crops. Little research _exists on competitiva interactions between these ~t ..o crops in association and the means to minimize them. Certainly shade toleranee will be a principal l.ssue in sueh research. Markets and Demand A synthesis of production and utilization Collection of accurate production statistics for cassava in Malaysia is hampered by the 1llegal natura of a significant percentage of the area planted to the crop. In·consequence a suspected downward bias exista in estimates of are a and product1on. However, since basically all the erop is sold for processing and data are collected on production of cassava starch and chips , an alternative production series can he constructed (Table ). !he utilization series in fact is consistently higher than the root production series. Since the downward bias in the production series can be identified, there is sufficient reason to suggest that the utilization series gives a much more accurate picture of cassava production trends in Malaysia. The two series offer quita contrasting views of trends in cassava production. The series developed by the extension department sho. .s little trend and ver y substantial variability. On the other hand, the utilization series displays a steady increase in the first half of the 1970's to a peak of almost 450 thousand tons of roots in 1976. Production then declined to about 300 thousand tons in 1980, ..h ere it has remained through 1983. !he latter series. as ..i ll be shown in the following sections, explains very well trends in exports and prices. The utilizatíon series will therefore be used as the best esimate of cassava production in Malaysia. The Domestic and Export Market for Starch Starch has always dominated the cassava economy of Malaysia, especially since cassava has never been a food souree in the country, except among some of the tribial groups. Moreover, starch production has traditionally been overted toward export, in line with moat of the rest of the agricultural economy. Finally, the history of the starch industry in 7 Malaysia has been one of constant movement in search of areas where cassava roots could be produced most cheaply, Le. where competition with tree crops was least or where illegal land use was not rigidly enforced. In the post-war period the starch industry settled in Perak and the following analysis will focus on starch production in that state. Only two starch factoríes existed ín Perak prior to 1945. By 1968, 19 plants were operating ín the state, wíth most of the growth comíng in the 1950' s, when 10 factoríes were set up (Table ) . At this poínt starch productíon depended primaríly on the sedimentatíon metbod, as only two plants were using centrifuges. Production from these latter plants was higher than for the sedimentatíon plants (Table ) , even though the centrífugal plants were only operating at 30% capacity. Also, the centrifugal plants ohtained an extraction rate of between 20 to 23% whíle the sedimentation plants averaged between 13 to 18% (Onn and Yet, 1971). With continuing problems with root supply and increasing competition from Thailand, it is not surprizing that a shake-out of tbe industry would occur in so competitive an environment. Thus, by 1982 only eight starch factories were operating in Perak (Table- ) • lfuat i8 clear, however, ía that thla shake-out did nat occur until the late 1970's. Prior to that and contrary to the root production statistícs -- the starch industry showéd steady growth in the post-war period. Starch exports íncreased steadily through the 1950! S and 1960' s and peaked in 1976 (Table ) • The sborter series on starch production complements these export trends and suggests that total starch production also peaked in 1976 at 68 thosand tons. Production declined from that level and bas been stable at about 50 thousand tons through the 1980' S. Exporta, however, declined much more dramatically and Malaysia became a net importer of starch in 1981. Two factors were responsible for this reversal: rapidly increasing doemstic conaumption and increased price competition from Thailand. Domestic starch consumption in Malaysia increased very rapldly during the 1970's, rislng from 1ess than 20 thousand tons in 1971 -- Onn and Yet (1971) estimate domestic consumption at 16.3 thousand tons in 1967 to about 50 thousand tons by the end of the decade. Majar users of cassava starch are monosodium glutamate and glucose producers and the textile industry. As industrialization proceeds in Malaysia starch demand is certain to continué to inerease. Particularly, any future developments in either the plywood or paper industry should lead to signficant increases in consump t ion. A market with signficant potential is the sweetner market. This market has expanded rapidly in Japan and Taiwan, while Indonesia 15 currently starting a sweetner industry. Malaysia imports about 85% of its consumpt10n requirements of sugar, even though domestic sugar prices are maintained at levels well abové world market prices in arder to cover Malaysia costs Di production. Sugar imports of tons in 1984 and a protected domestic sugar market offer scope for the development of a high fructuose sweetner industry basad on cassava starch. Moreover, development of th1s industry requ1res relatively moderate investment, since present starch processing factor1es can form the basis for au integrated starch-sweetner operatíon. However, domestic starch production 15 the limiting factor in the development of this industry. The other factor influencing recent production and export trends i8 increa8ing price competition from Thailand. This price competition i5 amply portrayed in Figure Before 1976 wholesale starch priees in Ipoh, Perak were well below Thai wholesale prices. This coincided with the period of expanding starch production in Malaysia. From 1976 to 1981, Malaysia starch prices in Perak were more or less on a par with Bangkok wholesale prices. During this period, Malaysia lost export markets even though prices in general were rising. In 1981 Malaysian starch become more expensive than Thai starch and Malaysia bacoma a net importer of starch. The situation was compounded by a falling price level. Thus, after two decades of growth, the Malaysia starch industry stagnated, caught between the high supply price for roots and the prices of imported Thai starch. For Malaysia to rema in competitive in starch would require further cost reductions in the production of cassava roots. PHltIPPINES Like Indonesia, the Phi1ippines i5 a multi-i51and economy; yet, un1ike Indonesia, the Phi1ippines has major population concentrations on a1l the major ialands, a1though Luzon stll1 figures as the economic center. !be agricultural economy i8 dominated by two grains, rice and maize, and two principal export crop8, coconut and sugarcane. Grain and food production in general are concentrated in the small farm sector while the export crops tend to be dominated by plantation systems, although smallholder production of copra is also important. The Philippines has an apparent comparative advantage in the production of copra and i8 by far the dominant exporter of this product. !bis agricultural structure has ereated something of a dual approaeh to po1icy. !be export erops have to a large extent been 1eft to the plantation companies in the private sector. There has not been, unti1 very recent1y, much government invo1vement in either research, exports or prieing in these crops. In the food sector, on the other hand, the situation has been just the reverse. !bree themes run through agrieul tural policy for grains: 11 eommitment to self-sufficieney in grain production apart from wheat, very heavy intervention in setting domestic prices, and an apparent connnitment to increasing productivity in the smallholder sector. !be achievement of self-suffielency ls seen as being dependent on priee po1iey and small farm programs. Control over domestie prices ls in the hands of the Nationa1 Food Authority (NFA), which has authority to control imports snd exports, to buy in the domestic market, snd to set both support prices and eeiling prices. Trade in foodgrains and domestie prices are to a large extent. administratively determined. Policy toward the small farro sector has included land reform, investment in irrigstion infrastructure, and specialized eredit snd extension sehernes. The stsge was thus approprlately set for the advent of the high yielding rice varieties. Under the Masagana 99 Program the Philippines "'ent from a consistent net importer to a net exporter of rice in the mid-1970's. This BUccesa has led to the recent development of the Maisan 99 Program, which hopes to achieve self-suffic1eney in maize in three years. Concern also runs to the large and growing wheat imports and 1dentifying means of either controlling such imports or substituting for wheat flour. Cassava fits well into this poliey contexto !he crop is essentially grown by smallholders, although sorne plantation produetion does existo Moreover, cassava csn be a domestically-produced substitute tor imported grains. !bis concern for self-sufflciency has even extended to the development of a national alcohol program based on sugareane and cassava; however, with the recent fall in wor1d oil prices the program has been scrapped. Nevertheless, cassava is seen as a crop that can contribute to meeting the increasing demand for carbohydrate sources. Sinee cassava i5 only a very minor arop in the Philippines and sinee che crop has reeeived little government support, the question to be pursued 1s what difference government involvement can make in developing cassava as a coromereial crop in the Philippines. L Production Production trends and distribution: The official production series for cassava in the Philíppines ia presentad in Table l. The series shows relatively stable area, production and yields from 1960 to 1974, followed by very dramatic increases in both area and yields. Such increases led to more than a tripling in production in three years and to over a quadrupl1ng 1n Uve years. This remarkable growth immediately bege the questions of what WaS responsible for this sudden take-off. - An analysis of such rap1d growth in production first turns to the impact on utilization patterna and market prices. As i8 discussed in the section on markets and demand, there is no corroborating evidence on either consumption or price levela to suggest that sucb production increases took place. On the other hand, alternative estimates of area and yield are limited. The agricultural census of 1971 estimated cassava area at 47,061 hectares, yields of 5.75 tlba, and production of 270,714 tons. Even at tbis stage there were major discrepancies between tbe census estimate and the Bureau oí Agricultural Economics (BAE) estimate. The major difference between the two production estimates 16 due to the reported area figures; the y1eld estimates are similar at this date. This discrepancy with the census figure raises some doubt about the adequacy of the sampling and estimation techniques for cassava estimates. This i6 not surprising given that cassava is such a minor crop in the Philippines. The only data whicb correspond to the BAE's estimate of increasing yields from 1976 to 1979 i8 tbe Special Study Division' s survey of 901 cassava farmera in the period 1977-79. Average yields for tbis non-random sample were 4.3 t/ha; however, this average was biased downward somewhat because the major growing area of Central Mindinao was not included in the survey. However, even this would not raise yields to the BAE estimate of 11.7 t/ha. A regional breakdown of production and area providas insight 1nto the regional locus of this supposad growth in cassava production (Table 2). Cassava 16 produced throughout the Philippines but most i8 produced in the southern islands. There is little production on Luzon, apart from the Bicol region 1y1ng at tbe southern tip of the island. The major producing areas are tbe V1sayas reg10n and Mind1nao. The production data suggest that cassava production 1ncreased at an annual rate of 20.4% on the island of Mindinao in the period 1970-81, while increasing in the rest of Philippines at a 9.6% annual rate. Mindinao accounted for 78% of the increase in cassava production in the periodo The years 1975 and 1976 sre psrticularly striking. Production in 1975 was 134 thousand tons and in 1976, 656 thousand tons. Tbis incresse almost doubled national p.roduction. In a single yesr are a increased from 20 to 44 thousand hectares and yields fram 6.8 to 14.8 t/ha. In just tbe Central Mindinao region production increased from 14 thousand tons in 1975 to 1.1 milI ion tons in 1979. These data suggest either explosive structural change in cassava production on Mindinao or a major revision of the data. The starch industry, based on plantation systems, is 3 concentrated on Mindinao but the data on cassava starch product1on suggest no major changas in tba industry in 1975-1980. Thus, ft appears that this major increase in cassava production in the last half of tha 1970's was in major part artefact. Independent comparison of production data with tha utilization data is left till the discussion of markets and demando Cassava production systems: Cassava in the Philippines 1s grown in botb plantation and smallholder produetion systems. There are faw estimatas of tha percentage of cassava grown in tbesa two systema. However,' plantation systems are associated only with starch milIs, and at least three factorias on Mindinso snd one in Eastern Visayas operate estates. As much as 6,500 hectares may be grown in plantation systems. This would imply that the graatar portion of cassava is grown by smallholders. These systems will be considerad in most detail. Cassava, while it is grown throughout the Philippinas, has never achieved the status of a major commereial crop, even on a regional basis. Maize is the most prominent upland crop for smallbolders. The reason for this follows prineipally from the relatively favorable agro=climatie eenditions that exist throughout tbe Philippines and the relatively universal distribution of paddy landa across tbe different regions. A ahort maturtty crop whieh produces relatively consistent yields under upland conditions fits better than a long maturity crop in smallholder systems, where rice production requires substantial resources during critical periods of the year. In general shortage of rainfall is not a limiting factor in cassava produCtion nor for the production of other upland erops. Because of eassava's better adaptation to poorer soils, caS8aVa i8 often found on the more infertile billside areas. Cassava i8 plantad tbrougbout the year and the only eonstraint on plant1ng time .is confliet with rice production activities. Such constraints are accentuated because very 1ittle hired labor is used in cassava production. In tbe Speeial Studies Division (SSD) surveyabout 75% of labor use in eassava comes from family labor (Table 3). Cassava producers, aeeording to tbe SSD survey, operate farros of a 11ttle over 3 bectares. of which only .6 of a hectare 18 devoted to cassava. Rarely are plots of over 2 hectares planted and of tbe 916 farmers in this survey, only about 40% aetually owned the1r land. Yet even on cassava producing farros, only abou! 11% of total cash income was derived from cassava. Otber crop sales accounted for far more incorne than cassava, even though over 80% of the cassava that was produced was soldo Cassava was thus grown as a minor cash crop by essentially small-scale producers on land not typically suited for other crops. Land is typically prepared by animal traction, although some small plots may be prepared by bando Because of the relatively high rainfall the land is either furrowed prior to planting or ridging 1s done at tbe time of the first weeding, usually by interrow animal cultivation. Ridging is apparently necessary to control root rot as the erop matures. This type of weeding limits any type of intercropp1ng. and eassava i6 usually found planted in monoculture. 4 Although a substantial range of varieties are found in the Philippines --the SSD survey found 22 different varietiea--, about half the farmera in the survey grew a variety named "white", while two-thirds of farmers grew either "white" or "yellow" (Table 4). These varieties are apparently selected for their good eating quality. The one peculiar feature of cassava production systems in the Philippines 1a the very low labor input devoted to weeding (Table 5). This partly reflects the use of animal cultivation but animals can be used at most twice for weeding and are often ineffective at controlling weeds within the rows. Moreover, weed control would be expected to be a problem under such relatively high rainfall conditions. Low labor input for weeding thus reflects other factora, including the reliance on family labor, competition with other crops for labor resources, and the relatively low commercial status of cassava. This same phenomenon applies to other input use. In the survey only 18 of 916 farmers or 2 percent used fertilizer on their cassava plots. For those farmers who did apply fertilizer the average application rate wae­ about 125 kg/ha of chamical fertilizers. For smallholder cassava production cash expenses were kept to very low levels, which may reflect the risky nature of marketing the crop. The riskines9 is as well reflected in harvesting patterns. Cassavs in general in the Philippines can be harvested anytime after sil' or seven monthe. Farmera in general harvest in small lots, partly for home consumption but principally as a mean s of insuring disposal at a remunerative price in the market. Substantial labor i8 as well expended on trimming, cleaning and packing the roots for sale. At least one study has shown that there i8 no los s in yield when harvesting in small lots between 6 and 9 months as compared to a single harvest at nine months (Villamajor, 1980). Cassava plantation systems in the Philippines are normally in the range of one to 1.5 thousand hectares in size. Planting and harvest are staggered to provide a continuous supply of cassava to the starch f aetories. This production is as well supplemented by purchases froro smallholders. However, in such large estates it has be en difficult to achieve any significant economies of scale in cassava production. The only significant changas are that land preparation i5 done by tractor rather than by animal traction and that herbicides are used in weed control. The rest of the operetions are performed by hand labor, usually on e piece rate by farmers contracted in the area. A 1978 survey of starch plants suggested that the higher overhead costs resulted in substantially higher own production costs as compared to purchased prices from local farmers - 249 pesos/t versus 174 pesos/t (Villanueva and Laguna, 1979). Yields: Compared to standards elsewhere in Asia, cassava yields in the Philippines are 10"', even though agro-cllmatic condítions are in general more favorable. The 1977-79 survey of 916 smallholder found an average yield of 4.02 tlha (Table ), a figure comparable lOO the pre-1975 BAE estima tes of around 5 t/ha. There was some variat10n in yields bet"een 5 regions but in general yields were uniformly 10107 throughout the Philipp1nes. The immed1ste question 1s why, especially if agro-clima tic constraints (except for soils) are not sn iasue. S1nce the Philippines has had no cassava research program until just recently, s potential cause of 10107 yielda may be the lack of well adaptad, high yield1ng varieties. The principal evidence that may be brought to bear On this hypothesis is that the first varietal releases by the Institute of Plant Breeding (Lakan 1 and Data 1) were selections that went by the more common names of golden yellow and Hawai1 5. These varieties were already being grown by farmers (Table 4), and yet the yield trials prior to release of these varieties gave an average yield of 42 t/ha for Datu 1 and 32 t/ha for Lakan l. Lack of adequate cultural practices thus appears to be the principal constraint on yields. Two principal factors appear to be involved: lack of appropriate soil fertility management and inaufficient weed control. As in other parts of Asia (except India) diseases snd pests do not appear to be a major problem in cassava, apart frcm the occasional incidence of cassava bacterial b11ght. One other possible limiting factor is lodging, given the frequency of high winda in the Philippines. Of these factors the very limited labor input 1n weed control is probably the major constraint on higher yields. Overcoming this constraint requires a closer study of labor utilizarion on the farro and the value of the production gain from further labor inputs in weeding of cassava. Yields on plantations are considered to be substantially higher, alrhough there are practically no published reports of yield levels on estates. One estate on Mindinao reports average yields of 18 t/ha (field notes, 1982). There i8 continuOU8 planting of cassava on this estate and apparently there has been problems in maintaining yield levels. Yields on newly opened land without fertilizer averaged about 30 t/ha. Yields have declined from this level and stebilized around the 18 t/ha average, while at the same time fertilizer application increased from zero to 400 kg and finally to 600 kg/ha. On another estate in Eastern Visayas the maximum yield obtainad in large fields was 29 t/ha on former rice land without fertilizer application (field notes, 1982). On this same estate as a whole average yields are in the neighborhood of 20 t/ha, w1th the flat, former sugarcane land sveraging 25 t/ha snd the hilly areas averaging 10-15 t/ha. Cost of production and labor utilization If cultural practices are a principal constraint on yields, this should be reflected in low rates of labor utilization. Labor input, in fact, is very low (Table 5), even by Thai standards where land preparation is performed by tractor. At an average of 53 mandays/ha the cassava plots can only be quite extensively managed, unless purchased inputs that substitute for labor are used. and this 1s not the case. The extensiva nature of cassava cultivarian i6 particularly reflected . in labor expenditure for weeding. In more usual labor profiles for cassava, weeding usually forms the largest single sctivity. In the Philippines most of the labor 1s utilized in land preparation snd planting and secondly in harvesting snd marketing. Little labor 1s expended on maintenance of the cassava crop. The impression is that resourees with a low opportunity cost are principally employed in eassava, family labor and animal power in the slaek seaSons and either marginal land or "excess" land which cannot be planted to more labor intensive cropa given the stock of family labor. Searce resources such as capital are used only when absolutely neeessary. Cassava is able to yield under such extensive conditions, a1though not at high levela. If this is so, then the costs of production derived by the SSD may be overestimated since family labor snd land were costed at average market prices. Just less than 80% of variable production costs is made up by labor charges (Table 6); of the wage bill 70% in imputed to fami1y labor. The rest of variable costs are principally delivery and transport charges and, for the 19% of farmers who were share tenants, the payment in kind to landlords. The other principal cost is the interest charged against fixed assets devoted to cassava. In the SSD study land was not costed at its rental valua but rather as an interest payment (12%) on its value. !bis interest charge to land forms the other major cost component. For per hectare production costs there i5 a certain stability in total cost across the different regions. - What i6 substantially more variable between regiona is yield levela, and this results in a substantial variabi11ty in pelO ton production costa from 160 pesos/t in Western Mindinao to' 338 pesos/t in Rico!. In fact, four of the nine region were producing cassava at a higher production cost pelO ton than farmers ~ere receiving as a market price (Table 6). Ho~ever, in all cases except region VIII cash income was greater than cash expenses. Costing indigenous farm resources at their oppertunity cost could make caasava profitable in these other regiona as wel!. However, ~hat i6 striking 16 that farm-level prices to a substantial degree reflected production costs and that profit er 10ss depended critically on yie1d level. A yield less than 3.5 t/ha ~as just not remunerative, at least when costed at market prices. Teehnology development: Designing appropriate technelogy fer cassava in the Philippines w1l1 be no eaay task, since the process is dependent on answers to several unknowns. Tha basie question i6 why eassava 16 grown 1n such extensive production systems whan the average farm size of cassava producers in just over 3 hectares. If cultural practices are the principal constraint on yields, mod1fying cultural pract1ces is going to require either providing farmers with further incentives to gro~ cassava (e1ther higher prices or more assured markets) and/or relieving what may be significant resource constraints within the farro. Answer to these questions can only come from a more extensive study of cassava within the complete farm system. Moreover, although cassava 15 clearly a commercial crop in these systems, what is not clear is the type of market toward ~hich increased production can be directed. The two issues of farming systems and markets together define the appropriate design parameters for the development of improved technology. There had been little research on cassava in the Philippines until the formatíon in 1977 of the Philippines Root Crop Research and Training Center 7 (PRCRTC). The center is located on the campus of the V1sayas State College of Agriculture and besides a staff of 15 researchers, the centar draws on the staff of the College to assist on research projects. Besides cassava the center doee research on sweet potatoes, yam, and taro. There i6 no cassaVa program as such, since the different disciplines divide their time between the different root crops, except for a breeder whose sale responsibility 15 cassava breeding. Research on cassava extends from breeding through crop protection and management to post-harvest utilization. The center in its few years of operation has principally been involved in defining research strategy and research priorities between root crops. Research by each discipline i5 defined on a project ba5is, which can be influenced by outside funding, especially the funding from the Philippine Council for Agriculture and Resources Research (PCARR). Policy development can have a marlted influence on research direction, such as was the case with the abortiva alcohol programo The center 5t111 is in the process of eompleting the development of a fully structured breeding, selection and varietal testing programo A germplasm banlt has been assembled and evaluated and at least three selections have been suggested as recommended varieties for release (Radix, 1980). A crossing and selection program has been started. The breeding focua ia on higher yield, with starch content being a secondary objective. This program i5 complemented by some cassava breeding which is done at the Institute of Plant Breeding at the Universlty of the Phil1ppines at tos Baños. A varietal testing system is in the process of being structured w1th the input of PRCRTC, IPB-UPl.B, and the Bureau of Plant lndustry. Trials will be carried out on six different experimental stations. Definition of the potential yield gap that may be exploited remains as yet relatively undefined. The yield data on the first three selections released by PRCRTC (two are already grown by farmera) show the almost traditional yield of promising varieties under experimental conditlons of over 40 t/ha (Table 7). Defining what potential yield leve la are at the farm level is more difficult, as well as the even more critical quastion of how to incraaae farm-level yields within farmer resource availabilities. What probably can be aaid i5 that a target of 15 t/ha i8 realistic, which for the Philippines amounts to a tripling in average yields. Marltets and Demand A synthesis of_production and utilization: The BAR cassava production series raises several questions about the accuracy of the estimates, particuIarly when they are compared to alternative production or yield estimates. The other test of the production series i8 a compari8on with data on utilization of cassava. Two st:udíes have attempted to reconcile production and consumption data for cassava. M. E. Constantino (1979) compiled known estimates of cassava consumption and found that between 1971 to 1976 these consumption estimates accounted for between only 50 to 80% of estimated supply (TabIe 8). The total consumption estimate of 252 thou8and tons in 1971 compares favorably with the agricultural census estimate of 271 thousand tons. She reconciled the two series by accepting the production series' and assuming human consumption as the residual. Per capita consumption thus increased dramatically. This, however, i8 not supported by SSD estimates for human consumption of cassava. The Palicy Analysis Staff in the Ministry of Agricultura adopted a different tactic. Area estimates were assumed reliable and yields were re-estimated based on long-term trends (Table 9). Per capita consumption figures were estimated on the basis of a cansumptian function. The production series, human consumption series, and starch series were then put together and feed use was estimated as a residual. Tha results shows rapidly rising feed use of cassava 1n the periad 1975-81. There are no ather corroborating data that feed1ng of cassava on-farm has increased dramatieally nor that majar increaaes in the use of dr1ed caSsaVa in concentrates has occurred. There is thus no eorroborating evidence for the BAE's rapid rise in production since 1975. Real farm level prices in the period 1975-80 were ver y stable, and they were only slightly lower than during the first half of the decade. All thingl! considered, it 1s probably best to base the product10n estimate on known consumption data. Ibis is attempted by region (Table 10). These regional eonsumption estimates asaume no inter-regional trade in fresh roots. Given th.. bulkiness and perishability of cassava roots, this is a reasonable assumpt1on. The SSD production and marketing survey, in fact, found very little inter-regional trade, except on Luzon where there was movement of caSSava from regions 1, 111 and IV to Manila. In the development oi the consumption estimates several assumptions were made concerning wastage, on-farro feeding of cassava, and production of chips. Waste WaS assumed to be a straight 15% of total consumption. Dn-farro animal feeding followed in part from the results of the SSD survey, which found that about 5% of production was used in on-farm feeding and that this occurred essentially off-Luzon. 11: was assumed that 10% of small-holder production in Mindinao and Visayas was fed to swine on farros. Production of dried chips was more difficult, sinee there i8 essentially no data on this consumption formo The SSD survey found production of cassava chips in only Central Visayas and Western and Northern Mindinao. These areaS were in general areSS without access to a starch plant snd with ready aecesa to either Cebu City or Cagayan de Oro, citíes where either flour or concentra te mills are located. Chip production in these three regions was assumed to be 25% of total amall-farm product10n. The regional utilization estimates more or less fo110w the regional distribution of production as presented in the 1975 BAE production statistícs, except for the Bieol region in southern Luzon. Up to 1976 the Bieol reg10n was always represented in the product10n statistics as the major producing region in the Philippines. Yet, on the consumption side there is no evidence to suggest what this production is utilized for, althougb there i5 occasional mention of chip production in Bieo1. This region remains something of a queation mark as far as csssava production and utilization are concerned. The utilizatíon estimate suggest that cassava i5 grown throughout the Philippines but that production i5 larger in the southern islands than on 9 '{,uzon. For most regions there is little alternativa to the fresh market for human consumption. Generally where production i8 largar there i5 access to alternative markets, and the analysis now turne to a closer look at cassava markets. Cassava for direct human consumption: Where cassava is consumed as a food source in tropical Asia, it is usually in areas where there is a "shortfall" in rice availabilities, either becauee of limited purchasing power and/or insufficient production levels. Cassava has not been incorporated as a major component in the Philippine diet beeause rice production i5 in general relatively evenly distributed throughout the islands, and in regions where riee supplies are short, grain supplies are supplemented by maize (Table 11). Moreover, consumption of wheat products has stesdiIy incressed in the post-wsr period snd has reaehed quite signifieant levels in urban areas. Root crops are generaIIy of minor importanee in the diet, and cassava, in faet, i8 less important than sweet potatoes as a food source. At an average annnal consumption of around 4 kg/eapita, cassava is consumed more as a vegetable crop then as a basre staple. Cassava can go through quite elaborate proeessing in the heme and oft.en the fresh root is milled fresh and used to produce a type of cake. In the larger eassava consuming areas in Visayas and Mindinao, there ia a certain seasona11ty to· eonsumption, quite direetly related to price changes (Table 12). On Mindinao cassava consumption 1s usually highest in the third quarter of the year and on Visayas in the first quarter. In botb areas priees reach their seasonal low in these periods. NevertheIess, low per eapita consumption leve la and casaava's role as a vegetabIe erop often implies a certain eIastieity in demando However, the limited available data suggest that per eapita consumption declines with income and that over timecassava consumption has also deelined (Table 13). In the best of circumstanees it is difficult to build a relatively expansive production base pureIy dependent on the fresh food market. Given the long history of cassava in the Pbilippines, it is highIy unIikely that cassava wilI evel: develop as a major stapIe. In part tbis was because agroelimatic conditions were not poor enough to favor eassava in any part of the Pbilippines; maize could always be grown as a secondary staple to rice. Developing eassava as a major commercial crop wilI thus depend on the development of other alternative markets for eassava. The starch market: The principal existing alternative market for cassava in the Philippines is for starch production. Cassava starch production through the last deeade bas been stagnant (Table 14). At the sama time net imports of cassava starch, while never large, have declined to relatively insignificant levels. Viewed in isolatlon these trends would appear to imply a relatively stagnant market for stareh; yet whi1e cassava starch produetion has been stationary, maize stareh production has been increasing at a reIatively rapid pace (Figure 1), indicating quite substantial growth in total starch demand. The issue then is why eassava starch has 10st a significant market share to maize starch. The major part of the cassava starch industry is located on Mindinao, together with part of the maize starch industry. The industry i8 by nature large-scale and in 1978 consiated of sevep plants with a combined annual capacity of 90.2 thousand tons of starch -l. In 1981 an additional plant with a capacity of 11.3 thousand tons carne into operat10n. What ls clear 18 that the industry 1s operat1ng well .below capacity and in large-scale processing plants this is bound to profoundly affect returns on cap1tal lnvestment. The cassava starch industry must operate w1th1n two majar constra1nts. First, the ·price of cassava starch 1s currently set by the pr1ce of maize starch and this price is lsrgely determined by the price of the raw mater1al and, to a lesser extent, the prices of the relatively high-valued by-products of maize wet milling. Second, the industry 18 constrained by the availability of cassava roots. As 1s not the case with maize, the cassava processing plants must rely on a continuous harvest of roots rather than on stored supplies or imports. At least for the starch industry there appears to be a distinct seasonality to cassava supplies. Table 15 shows the monthly production of five of the seven stareh mills operatlng 1n 1978. Only two of the five plants could opera te the year round snd for these two plants production in the first part of the year was about haH of the production in the latter parto This coincides to a large extend w1th the seasonality in the human consumption of fresh roots. The rationale of plantation production 1s to plan supplies in relation to proeessing needs. Ironical1y. the two plants which remained elosed for the longest period during the year were exactly those which relied principally on their own produetion from their estates. The otheT plants relied to a large extent on purchases of smallholder production (Table 16). Moreover, according to the companles' own estimates, it was cheaper to buy cassava from smallholders than to produce the roots in estates. Without further efforts at meehanizing cassava production, the evidence suggests that lt is very diff1cult to achieve economies of scale in cassava production, even with such a large yield margin between smallholder and estate production in the Phil1ppines. Another factor which may contrlbute to the seasonal undercapacity in operation of eassava starch plants is an apparent price squeeze due to seasonality in ma1ze prices. Cassava starch prices tsnd to be lowest in the first half of the year rather than in the peak processing period during the second half of the year. The milIs appear to be eaught in a squeeze between high root priees and low maíze, aud therefore maize starch, prices. The squeeze between input and output prices and the limíted root availability in the first half of the year put severe constraints on the ability of the industry to operate at fulI eapaeity. Even for large-scale plants the costs of produeing cassava stareh depends prineipallyon the cost of the root. Fuel is another large cost !¡ There is reported cases of household productíon of cassava starch. There are no data to suggest how large such production is but it is assumed to be minor. component in large-scale plants. As can be seen in Table 17, the costs of production are not substantially different from the selling price. Small changes in the root purchase price would thus substantially affeet the profitability of cassava starch production. As in most countries, the market for starch is not understood in any detail. One survey of 64 industrial users showed a relatively broad use in both food and industrial uses (Table 18). If the total cassava starch production figures are correct, this sample would appear to aecount for about one-third of total consumption. The use of cassava starch in monosodium glutamate production used to be a substantial part of end demando About 1972 m.s. producers invested in new equipment which utilized the cheaper molas ses as the raw material, eliminating most of this demand for cassava starch. Constantino (1979) also estimates that about 30 to 35% of cassava starch goes into the manufacture of tapioca pearl. The potential growth in the starch market has uot been studied. The consensus in the cassava starch industry is that demand is currently not a major constraint. This is not reflected in imports, but _low import levels can mostly be attributed to a- 70% ad valorem duty. Three additional cassava plants witb a total annual capacity of 90,000 tons of starch are either under construction or in the advanced planning stage. This would appear to indicate an expected continued growth in demand for cassava starch. Yet, Buch investments seem somewhat superfluous in an industry that ls only operating at 30 to 40% capacity. Data available on the starch industry would thus seem to raiee more questions than they answer, and, moreover, they produce a quandary as to planning the future directian of cassava develapment. That is, the first constraint on the expaneion of the caseava starch industry is the limited capacity to produce sufficient Cassavs roots at a competitive price. Indications are that smallholder production is botb a more economical as well as socially preferable means of increasing cassava production. Yet tbe nagging question remains that if smallholder productivity snd production are increased, 15 starch demand sufficient to absorb major increments in production? Clearly, the export market will not be an option for surplus starch production. The starch processlng capacity that is now in place represents about double current national production of cassava roots. Since cassava plants will now be distributed through mast regions in the Philippines, the starch indust~y could provide the basis for major expansion in cassava production, given an increment in farm productivity. The starch induttry thus provides an inltial base on which to develop cassava production -l. However, this ~I Planning ia critical to these large-scale plants. The farmers in the Bohol region were contracted ta supply a new, 60,000 tan plant on that island. For such a larga plant production was increased by a major increment over previaus levela. The plant did not open as projected and farmare bad to chip their production and se11 at prices which were less than half of the previous year's level. The plant's ab1lity to contract for the next few year'e production bas now been badly compramised. market does not provide the certainty for major expansion in cassava production, nor, since large-scale plants are the rule, does every farmer have access to this market. Analysis of other market alternatives would thus appear warranted. The dried chip market: Gaplek-type, dried chips are produced in the Philippines but production has never been large enough or sufficiently continuous to allow the development of a broad-based market. Chip production i9 based in the Visayas and Mindinao areas and principally serves as a means of venting fresh root surpluses where there are constraints on access to fresh markets. Prices tend to be cheaper than their fresh root equivalent and chips are absorbed as cheap substitutes in industries such as feed concentrates, starch (for making glucose),' and flour (for noddles and non-Ieavened bakery products). In general, pricea are too lowat eurrent yields to provide incentives for increases in chip production. Currently, chips are the market of last resort for roots that need to be harvested or once harvested, have no ready market. Producing roots just for the chip market, however, does not cover total costs of root production. However, the question is what would be the potential market for cassaVa chips if root yields were increased? Development of a broader based chip market would relieve the uncertainty about the starch market. Like a host of other tropical, wheat-importing countries, the Philippines has for a long time had a law which required that wheat floure be substituted with domestically produced flour up to a minimum of 10%. Cassava flour was assumed to be the alternative flour with the most promise. The law prompted the establishment of at Ieast one cassava flour milI on Luzon. !he milI never operated at capscity - and it was never possible for the wheat flour industry to meet the requirements of the law, sinea sufficient cassava flour at a remunerative price was never available. As with similar lsws in other countries, the market was potentially larga (Tabla 19) but cassava flour could not be produced st a competitive price. !he composite flour market off era potential if cassava chip prices can be reducad but experience has shown that basing a cassava chip industry on mixed feeds presents far fewer organizational constraints (as well as quality problems) than developing cassava chips for a composite flour industry. In the Iast decade there has been a structural change in the poultry industry, as production has shifted from small-scale units to large, vertically integrated commercial operations. Meat production from these operations has tripled in the last decade (Table 20). Such structural change has spawned rapid growth in the feed concentrare indUstry and the production of mixed feeds has increased at an annual rate of 12.2% over the last decada (Tabla 21). Of total production of the mixed feed industry, 70% goes to poultry whiIe the other 30% is swina feed (Table 22). A principal feature of the industry, however, 1s it locus on Luzon, where 90% of mixed feeds are produced. Since the locus of cassava chip production is in the South, inter-island transport costs will be a major cost component affecting the farro-Ievel chip price. Growth in industrial demand for maize has causad a fundamental change in the strueture of the maize market (Table 23). Although maize production 13 has increased at the very respectable rate of 4.3% per annum over the last decade, increased use of maize for feed and for starch have resulted in a reduction of supplies going to human consumption and a continuing, if not rising, level of imports. Moreover, maize production has stagnated over the past three to four years, raieing concerns that imports will have to increase even further. The Philippines i5 currently pursuing a self-5ufficiency program in maize, along the lines of their successful rice programo Maize yields at less than one ton per hectare are low and the heárt of the Maisan 99 program i5 a tropical maize technology, in particular a hybrid maize resistant to downy mildew. There are two scenarious that follow from the success or ineffectiveness of the new maize technology. If the technology should succeed, planners in the Ministry of Agriculture hope to move the Ph11ippines into a net export position in maize. Expans10n in cassava chip product10n is designed to be used domest1cal1y and to re1ease further maize supplies for export. If such exporte are to be hand1ed by the private sector and not the Nat10na1 Food Authority (NFA) , then domestic price levela wi11 have to be brought in line wirh wor1d prices from their presenr position above world pricea. In turn, cassava chip prices wou1d have to be brought into line with world maize pricea •. However, Thailanrl has found the social profit to be h1gher by exporting dried cassava to Europe and using domestica1ly produced maize in its concentrate industry. Were. the Philippines to deve10p a competitive cassava chip industry and assuming that the EEC doea not renegotiate the tariff binding on cassava, the Philippines would gain more by exporting cassava than maize. The orher Rcenario ls that the tropical ma1ze technology proves ineffective in the faee of continued increaaea in demand. Without an a1ternative carbohydrate souree, polícy makers have to decide between increased maíze imports or higher pricea (or a price squeeze, since price cei1ings on food commodities are maintained) for poultry products. Deve10pment of a cassava chip industry which services the feed concentrate industry would thus provida a sort of insurance against continuad stagnation in maize yields with no risks, since the chips could always be exported. However, development of the cassava chip market wi11 pot be easy and raising farm level yields w111 probably be the easiest component in the expansion of the chip market. A cheap drying techno10gy wi11 be a critical constraint. lt is not clear how and whether this can be solved under the ganerally high rainfall and humidity conditions prevalent in the Philippines. Possibly, the locua of cassava production could be shiftad to the drier areas on Luzon or coconut snd rice drying units cou1d be adapted to cassava. Second. internal transport costs will playa critical role in determining cassavs's ability to compete. lnter-island trsnsport ia relatively expensive for a bulky commodity like cassava chips, snd with most of the cassava production area in the south and the feed industry on Luzon, -transport costs will capture a not unsubstantial portion of the output price. This, however, may be counterbalanced by a recent trend to Iocate new feed milI capacity in Visayas and Mindinao. Finally, given the Philippines' policy focus on improving the we1fsre of the rural poor, development of the cassava crop wi1l take place within the smsllholder sector rather than within a plantation system. Su eh a foeus would require substantial institutional aupport to develop produetion and proceasing systems and market linkages. A national Caasava produetion program has been formulated by the Ministry óf Agriculture. The plan focuses on raising cassava yields in all regiona in the Philippines. Where starch plants are already in operation, increased production will be directed at servicing the plant. For those caasava production regiona that lie outaide ehe effective transport radius of a stareh plant, increased produetion will be chipped and dried. Production credit and loans for financing of chipping and drying capacity will be extended through farmers associations. The credit will also be extended only on the basis of a marketing contraet between the aesociation snd sn accredited buyer, either a starch or feed milI or the National Food Authority. The program, as currently conceptualized, focuses on both production and marketing and foresees the principal market to be for use in feed concentrates. Priclng and market efficiency: Apart from the supply areas of the starch plants, prices for caSSava are principally determined by demand in the fresh foad market. Cassava ls a vegetable and not a staple foad in the Philippines. Retail prices are high and do not follow staple grain prices (Table 24). The ratio of retail, milled maize price8 to retail cassava prices over the period 1970-79 varied from 1.4 tú 2.4 and varied dramatically from year to year. For pricas af fresh cassava and milled maíze to be equal on a ealoric basis the ratio should be around 3.5. Calories derived from cassava are just too expensive to be considered a staple. However, this high retail price for cassava is not translated into high farm-level prices. Fa1'1ll priees make up as little as 30% of the eventual retail prica (Table 25). These marketing margins are somewhat typical for cassava consumad in urban areas, whera transport frem farm to urbau ceuter i8 relatively expensive. However, the SSD surveyed 222 cassava midd1emen throughout the Philippines and found the gross margins between farmer and who1esaler as well as between wholesaler aud retailer to be much smaller than that reflected in the average price data (Table 26). Moreover, actual marketing costs (without accounting far los8es) were low. There is thus sorne doubt as to the extent to which the gross margina, . as reflectad in the BAE price data, can be generalized to cassava market channels. Nevertheless, margins for fresh cassava remain high. To evaluate whether caasava is going to compete with grains in alternativa markets, the relavant price is the farm, aud not the retail, price. The price ratio between maize and cassava at this level is much rnore favorable (Table 24). Accepting a minimum price equivalent ratio of 3.1 -3/, farro-Ievel prices were very nearly competitive with maize between 1972 to 1978. This would be expected if cassava starch or chips were to be competitiva with maize-derived products. Cassava root prices have remained 11 The ratio assumes a convers!on of roats ta chips of 2.5:1 and that dried cassava i8 competitive at 80% of the maize price. 15 distinctly uncompetitive since 1979, at least on average. Moreover, as would be expected, root prices are much lower in the southern regions as compared to Luzon, by as much aS half (Table 27). Cassava root prices are only just marginally competitive with grain prices in the Philippines snd at present yleld levels these prices are not sufficiently high enough to draw forth the supplies that are needed to serviee alternative markets. The fresh market can operate at higher price levels and i8 thereby the principal demand factor in the market. However, there ia very limited capacity to absorb addítionsl supplies. With yield íncreasing technology priee determínstíon in the cassava root market will have to be linked to the coarse grain markets. The fresh root market ls small enough that making this transition, that ia driving prices downward in the fresh market, should be easiIy accomplíshed. As a broader based, chip market becomes estabIished, market efficiency snd better market integrstion between regions should be vastly improved. Conclusions: The Philippines wss the first country in Asia to receive cassava from the New World. Casssva was brought by the Spsnish from Mexlco in the 17th century. Cassava never estsblished itself as an alternative carbohydrate staple to rice. Given the generally favorable rainfall and soil conditions, this role was captured by maize. Moreover. maize, while at first being grown as s cheap foodgrain alternative to rice, provided the raw material base for the development of both a starch and feed concentrate industry. A large and growing domestic market far carbohydrate saurces for industrial uses now exista. The issue is whether maize or cassava has a better 'competitiva advantage in servicing .the continued internal growth in demand for carbohydrate sources. To complicate the issue this competitiva advantage will be defined by technologies not yet in place -l. Current farm-level yields in the Phillppines are unreasonably low. The patentisl yield gap that can be exploited is therefore much larger than in other Asian countries. Moreover, a sort of vicious circle ls seemingly operating, in whicb farmers do not intensify cultural practices because marketing i9 so risky and alternative markets do not develop becauss cassava is not competitive at current yield levels. A closer study of cassava within current farm syatem ia needed to identify the types of technology required to raise cassava yields. Increasing productivity, however, must be simultaneously linked ta market development. In this regard the national csssava· production program has formulated the requisite links between technology extension, credit, and marketing contracts. Nevertheless, if a broad based caS9ava market 19 to develop, it will depend on the ability ta produce cassava chips. Drying technology i8 potentially the major constraint on future development of cassava. Various ~/ A tbird aource of production growth is continued expansion in area planted to maize on Mindinao rather than yield increases on current production area. However, differential changes in yields between maize and cassava would as well influence tbe potential for are a expansion in maize. alternativas viII have to be tested under various climatic conditions and costs vill need to be assessed. Given drying constraints and relatively high inter-island transport costs, consideration of pelleting in southern production areas should be considered at an early stage. The future of caS5ava in the Philippines i5 thus partly dependent on the success of the Meisan 99 program but will principally rest on a systematic assessment oí the potential oí new production and processing technology. Cassava in the Philippines thus has the difficult task of proving its potential. Table Philippines: Production and lrade of Cassava Starch, 1968-80. . Trade Year Production Exports Imports (t) (t) (t) 1968 22,044 1,201 1969 18,204 . 350 1970 22,771 193 10 1971 29,277 404 1972 27,867 3,722 1973 15,616 2,211 1974 18,375 4,229 1975 17,425 4,220 1976 17,391 1 2,004 1977 16.,576 3 5 1978 17,024 3 3 1979 17,371 1 5 1980 N.A. 14 4 Source: National Census and Statistics Office. Table Philippines: Cassava Varieties Reportedly Grown on 916 Farms, 1976-1979. Varie1L- Golden ~ava Region White Yellow Red Native Yellow Hawaiian Brown Other1 llocos 105 3 Centra 1 Luzon 36 36 1 5 Southern Tagalog 29 14 29 13 Bieol 13 86 9 6 27 Western Vi sayas 27 8 57 46 Central Visayas 35 45 8 la Eastern Visayas 61 41 7 Northern Mindinao 48 42 5 Western Mindinao 72 37 7 3 7 Tota 1 Farms 426 172 86 79 75 13 11 116 % Farms 44 18 9 8 8 1 1 11 1 Ineludes 15 other varieties Source: E.B. Mejia, et. al., "Cassava Soeio-Economie and Marketing Study, Philippines," Special Studies Division, MinistrY of Agriculture, No. 79-26, Oetober 1979. Table Philippines: Nominal Prices Received by Farmers, Wholesale and Retail Prices, by Region, 1979 and 1980. 1979 1980 REGION Prices pri ces Received Wholesale Retail Received Wholesale Retail (Pesos/k9l (Pesos/k9) (Pesos/kg) (Pesos/kg)(Pesos/kg) (Pesos/kg) PHILIPPINES 0.37 0.74 1.19 0.44 0.85 1.28 Ilocos 0.60 1.20 1.26 0.75 1.29 Cagayan Va 11 ey 0.50 0.60 1.54 0.56 1.34 Central Luzon 0.56 0.65 1.02 0.48 0.69 1.ll Southern Tagalog 0.44 0.93 1.00 0.49 0.91 1.01 Bi col 0.38 0.64 1.09 0.42 0.69 1.07 Western Visayas 0.38 0.62 1.15 0.47 0.87 1. 53 Central Visayas 0.30 0.52 0.91 0.36 0.53 1.15 Eas tern Vi sayas 0.40 0.88 0.48 0.95 Western Mindanao 0.29 0.76 0:96 0.44 0.99 1.18 Northern Mindanao 0.34 0.61 0.86 0.43 0.80 1.05 Southern Mindanao 0.37 0.63 1.09 0.38 0.79 1.30 Central Mindanao 0.39 0.78 0.95 0.50 0.84 1.00 11 Sower: Bureau of Agricultural Economics. Table Philippines: Supply and Utilization of Cassava as Estimated by M.E. Constantino, 1971-77. 1971 1972 1973 1974 1975 19/ó -1917 (OOOt) (OOOt) (OOOt) (OOOt) (OOOt) (OOOt) (O OOt) Sumili Production 424.7 450.4 444.7 480.0 684.5 794.4 1011.1 Imports 2.0 18.6 13.8 21.3 21.0 10.0 Total 426.8 468.9 458.5 501.3 705.5 804.4 1011.1 Oemand Starch 148.4 157.9 91.9 113.1 108.2 97.0 103.6 Animal Feed 18.3 19.4 19.1 20.6 '29.4 34.1 42.5 Available for Human Consumption1 260.1 291. 7 347.5 367.5 567.8 673.3 865.1 Human Consumptian2 86.2 125.3 ·195.2 282.0 237.2 253.0 231.0 Tata 1 1 426.8 468.9 458.5 501. 3 705.5 804.4 1011.1 Total 2 252.8 302.6 306.3 415.8 374.9 384.1 377 .1 1 Calculated as a residual. 2 Calculated from SSD foad consumptian surveys. Source: M.F.. Constantino, "Cassava Market Study and a General Strategy of Implementa­ tion for the Cassava Program, unpublished M.B.A. thesis, Asian Institute of Management, 1979. Table Philipp1nes: cassava Varieties Selected for lelease by _t he Philipp1ne 1bot Crop ~search and Training Center MonthS to Dry Variety harvest Yield matter (t/ha) (%) PR-C13 10-12 42 34 PR-C24 8-10 43 39 PR-C62 10-12 46 33 Source: '!he Radix, VoltmE 2 (1), Jan-June 1980 Tab1e . Philippines: Ama, Pro::luction and Yie1d Cassava, 1960-19B1 Crup Year Area ProdúCtion Yie1d (ha) (tons) (t/ha) 1960 79,460 442,413 5.57 1961 100,310 546,611 5.45 1962 92,980 494,805 5.32 1963 BO,280 457,769 5.70 1964 93,540 596,156 6.37 1965 93,280 645,720 6.92 1966 89,700 614,386 6.B5 1967 86,520 528,727 6.11 1968 B3,880 481,928 5.74 1969 85,690 482,327 5.69 1970 82,620 442,223 5.35 1971 81,820 427,055 5.22 1972 82,680 439,697 5.32 1973 87,420 444,710 5.09 1974 96,710 480,015 4.96 1975 119,310 684,507 5.74 1976 144,650 1,153,958 7.98 1977 179,270 1,710,767 9.54 1978 181,770 1,781,961 9.80 1979 192,360 2,253,824 11.72 1980 204,190 2,277,338 11.15 1981 211,370 2,255,115 10.66 Source: Bureau of Agricult ural Econcxnics, published in National Econanic and Developrent Authority, Philippine Statistical Yearbook, Manila, 1981. Table Philippines: Per Capita Consumption1 of Cassava and Prices2 by Quarter and Region, 1973-76. Jan=Marcn A)2ril-June Jt.i1i-Se)2t. Oct=Oec. Region Consumption Price Consumption Price Consumption Price Consumption Price (kg/capita) ()2esos/kg) (kg/capita) (pesos/kg) (kg/capita) ()2esos/kg) (kg/capita) (pesos/kg) 1 1.4 0.53 1.5 0.53 1.8 0.62 1.4 0.51 II 1.9 0.53 1.0 0.60 1.7 0.50 1.8 0.55 III 1.9 0.52 1.5 0.61 2.1 0.53 2.4 0.53 IV 2.3 0.41 1.9 0.45 2.3 0.54 2.2 0.54 V 3.9 0.43 2.8 0.44 4.1 0.48 3.2 0.54 VI 2.6 0.47 3.2 0.70 2.1 0.49 2.9 0.48 VII 8.1 0.31 5.2 0.47 3.5 0.41 4.6 0.53 VIII 5.9 0.34 4.8 0.64 5.4 0.38 2.8 0.81 IX 6.1 0.31 4.5 0.66 10.,9 0.29 4.7 0.42 X 4.8 0.40 4.4 0.77 5.1 0.37 4.7 0.46 XI 5.4 0.38 5.1 0.33 4.0 0.36 4.2 0.40 XII 5.5 0.43 5.8 0.41 11.5 0.35 3.9 0.42 1 Per capita consumption expressed on an annual basis. 2 Constant 1972 prices. Source: Calculated from unpublished consumer food consumption surveys carried out by tne Special Studies Division, Ministry of.Agriculture. Table Ph il i pp i nes : Per hectare Production Costs, Yields, and Costs per Ton, 1977-79. _._ ............ - Region Cost Item 1 !II IV V VI VII VIII IX ·-X-- Average ------------------------¡Qesos¿ha)---------------------------- Variable Costs Labour Hired 29.1 26.6 103.5 124.8 28.0 181.6 167.0 113.3 75.1 98.8 Food 10.4 1.0 2.1 10.3 10.1 56.9 51.8 9.2 15.6 Family 288.2 322.6 280.2 363.4 165.9 179.2 267.9 368.8 266.2 282.8 Land Preparation Tractor 15.6 32.0 I 7.0 Animal 1.5 0.5 0.9 5.6 2.7 23.5 3.4 4.2 Planting Material 0.6 0.1 Fertil i zer 0.1 3.4 0.2 0.9 0.1 Landlord In ki nd 28.5 8.7 16.8 17.2 14.9 31.3 33.2 13.1 52.8 23.3 Cash 232.2 12.3 4.6 30.7 Transport 41.9 73.2 3.6 19.6 .2 18.9 35.9 21.1 Interest (Working Capital)lI 40.9 18.8 14.1 16.8 7.9 19.4 28.3 27.7 22.4 21.7 Sub-total 688.2 444.2 414.6 524.8 232.2 479.6 556.1 629.4 469.7 . 505.5 Fi xed Costs Depreci ati on 19.2 28.2 24.2 20.4 12.5 30.2 , 15.5 11.D 8.2 18.9 Repair 5.7 21.3 13.9 2.·9 16.5 3.4 3.6 6.1 21.1 9.1 Interest y 322.1 470.9 447.5 293.5 344.6 386.1 227.3 217.7 271.7 325.2 Sub-total 347.0 520.4 485.6 316.8 373.7 419.7 246.3 234.8 301.0 353.1 Total Costs 1035.1 964.6 900.1 841.5 605.8 899.3 802.4 864.2 770.7 858.6 Yield (t/ha) 6.19 5.84 3.36 2.49 2.21 5.46 2.16 5.39 4.03 4.02 Cost per ton 167.2 165.2 267.9 338.0 274.1 164.7 317.5 160.3 191.2 213.6 Farm Price 250 260 190 230 250 190 300 240 220 230 1/ Interest on cash expenses with interest rate of 12% Y Land costs fur land owners included as interest on land value, i.e. implicit land rent is 12% of land value. , Source: LB. Mejía, at.al., "Cassava Socio-economic and Marketing Study, Philippines," Special Studies Division, Ministry of Agriculture, No. 79-26, Oct. 1979. Table Philippines: Labor Use, Farm Size and Average Cassava Area in Cassava Production $ystems, 1977-79. Region I III IV V VI VII VIII IX X Average Labor Utilizatíon (man days/ha) Land Preparation 11.6 20.0 21.9 27.0 10.8 10.8 22.4 16.9 16.3 17.6 Furrowing 2.8 2.2 1.1 3.9 0.2 2.6 3.4 2.6 1.5 2.2 Planting 10.4 6.1 10.5 7.3 5.0 8.5 10.2 8.8 6.8 8.1 Weeding 3.6 5.2 11.1 14.9 2.9 5.9 14.0 19.2 6.3 9.5 Harvesting 5.9 6.3 15.7 7.8 5.3 27.8 8.7 9.2 7.5 9.8 Packing and Transport 6.7 4.2 4.6 1.9 2.0 1.8 3.9 5.7 10.0 4.4 Peeling and Drying 8.3 4.2 1.0 1.3 Total 41.0 44.0 64.9 62.8 26.2 65.1 62.6 66.6 49.4 52.9 Farro Size (ha) 2.25 2.25 2.93 3.72 4.29 2.82 2.38 3.15 2.50 3.03 Cassava Area (ha) 0.65 0.54 0.60 '0.79 0.49 0.85 0.47 0.58 0.52 0.61 Source: LB. Mejia, et.al.. "Cassava Socio-econoroic and Marketing Study, Philippines" Special Studies Devision, Ministry of Agriculture, No. 79-26, Oct. 1979. able Philippines: Annual Costs of Production of Cassava Starch for a Factory with a Capacity of 20 t/day of Starch, 1978. Total Per ton of starch Cos t Item (000 Pesos) (Pesos) ariable Costs Cassava Roots 6300 1050 labor 108 18 Fuel 1692 282 Gunny Bags 420 70 Interest on Working Capital 96 - 16 Transport (delivered ex-factory) 960 160 Total Variable Costs 9576 1596 ixed Costs Depreciation 1002 167 Interest on Fixed Capital 1200 200. Total Fixed Capital 2202 367 otal Costs 11,778 1963 e 11 ing Price 2100-2400 ource: M.E. Constantino, "Cassava Market Study and a General Strategy of Implementation for the Cassava Program," unpubl ished M.B.A. Thesis, Asian Institute of Management, 1979. Table Phi li ppi nes: Average Monthly Consumption of Cassava Starch by Type of Final Product for a Sample of Firms, 1978. Number of Quantity Percent Final Product Firms (t) (Xl Kropeck 22 97 19 Noodle 23 41 8 Glucose 2 175- 34 Adhesive 3 4 1 Cardboard 12 46 9 Monosodium Glutamate 1 113 22 Detergent 1 38 7 Total 64 512 100 Source: C.D. Vi11anueva and R.S. Laguna, "An Intensive and Critical Survey of Existing Industrial Processing of Root Crops and Projecti on for the Next Decade," PRCRTC Annua 1 Report. 1979. Table . Philippines: Type of Labor Used in Cassava Produetion by Region (man days/ha). Hired l eaf(J1n Resion Cash Kind O~erator Fami lX Exehange Total Ilocos 3.7 24.4 11.6 0.2 39.9 Central Luzon 4.5 28.0 11.5 15.0 59.0 Southern Tagalog 15.0 24.9 25.9 65.8 Bieol 14.2 24.0 25.0 0.3 63.5 Western Visayas 3.5 0.3 14.1 8.0 0.3 26.2 Centra ¡Vi sayas 12.2 21.8 17.5 13.7 65.2 Eastern Visayas 22.8 26.6 10.3 3.2 62.9 Eastern Mindinao 14.9 39.0 16.8 1.3 72.0 Northern Mindi nao 8.5 29.9 10.2 0.8 49.4 Average 11.1 2.8 24.8 15.6 0.7 54.9 Source: E.B. Mejía, et. al., "Cassava Socio-economic and Marketing Study, Philippines," Special Studies Division, Ministry of Agriculture, No. 79-26, Oct. 1979. Table Phil ippines: Estimates of Supply and Distribution of Cassava by Region, 1975. Per Capffa - - Tota 1 Human Dried Animal Region Consumption consum)tion Starch Chips Feed Waste Total (kg/caQita) {t (t) (t) ( t) (t) (t) Iloeos 1.5 4,904 10,370 2,695 17,969 Cagayan Va 11 ey 1.9 3,673 648 4,321 Centra 1 Luzon 1.6 6,736 1,189 7,925 Southern Tagalog 2.3 11,992 2,116 14,108 Bieol 7.6 24,274 4,284 28,558 Western Visayas 5.5 22,803 18,000 4,420 7,981 53,204 Centra 1 Vi sayas 7.5 25,402 12,701 5,080 7,621 50,804 Eastern Visayas 13.7 35",620 4,749 7,124 47,493 Western Mindinao 10.0 20,480 10,240 4,096 6,144 40,960 Northern Mindinao 8.2 18,975 15,000 13,8001 5,520 9,405 62,700 Southern Mindinao 4.9 13,304 1,774 2,661 17 , 739 Central Mindinao 11.0 22,770 47,340 6,665 13,549 90,324 Manila 2.5 12,425 12,425 Phi Ji ppi nes 5.4 223,358 91,710 36,741 32,304 65,417 449,530 Source: CIAT estimates. Table Philippines: Sources of raw material and unit costs of cassava roots purchased by five starch factories, 1978. Own Plantation Farmer MTOa1eman Firm Percent Unit Cost Percent Unít Cost Percent Unit Cost ~ .. _. _ (Pesos/kg) (%l (Pesos/kg) (!) ~ .. _(Pesos/J5.g) 1 60.0 0.23 40.0 0.23 2 90.9 0.28. 9.1 0.18 3 15.0 0.18 85.0 0.18 4 10.0 0.24 90.0 0.16 5 88.6 0.37 1.2 0.15 10.2 O.SOY Average 18.2 0.25 78.3 0.17 3.5 0.28 11 Gaplek Source: C,O. Villanueva and R.S. Laguna, nAn Intensive and Critical Survey of Existing Industrial Processing of Root Crops and Projection for the Next Oecad'e," PRCRTC Annual Report. 1979. Table Philippines: Monthly Production of Starch by Five Starch Factories, 1978. Firm Month 1 2 3 4 5 Total (t) ( t) ,( t) (t) (t) (t) January 203.2 1,098.8 656.9 1,954 February 741.0 283.9 1,025 March 42.8 576.4 399.9 1,019 April 123.3 437.7 350.9 912 May 173.3 678.5 258.9 1,111 June 180.8 753.2 242.5 69.1 - 1,246 July 166.1 707.6 412.7 239.8 1,526 August 195.7 1,028.5 689.1 113.6 2,027 September 171.1 1,091.8 644.6 118.9 2,026 October 166.3 81.1 1,110.6 683.7 159.5 2,201 November 161.7 161.3 1,272.0 671.5 165.9 2,432 , December 76.7 129.0 1,121.7 704.7 140.4 2,172 Total 1,458.0 574.7 10,612.9 5,999.2 1,007.1 19,652 Source: C.D. Villanueva and R.G. Laguna, "An Intensive and Critical Survey of Existing Industrial Processing of Root Crops and Projection for the Next Decade," PRCRTC Annual Report, 1979. Table Philippines: Nominal and Real Prices of Cassava at Farm, Wholesale and Reta;l Level, 1970-80. Farm Wholesale Retail Year (pesos/kg) (pesos/kg) ( pesos{J<.9.L Nominal 1970 .12 .19 .32 1971 .15 .24 .38 1972 .15 .29 .46 1973 .21 .32 .53 1974 - .29 .40 .70 1975 .29 .41 .71 - 1976 .28 .43 .71 1977 .30 .53 .80 1978 .32 .57 .74 1978 .37 .74 1.19 1980 .44 .85 1.28 Real (1975 pri ces) , 1970 .25 .40 .67 1971 .27 .43 .69 1972 .25 .48 .76 1973 .30 .46 .76 1974 .31 .43 .76 1975 .29 .41 .71 1976 .26 .40 .67 1977 .26 .46 .70 1978 .26 .46 .60 1979 .25 .51 .81 1980. .25 .49 .74 Source Bureau óf Agricultural Economics. Table Phil ippines: Marketing Margin for Fresh Cassava Root for Various Types of Middlemen, 1977-79. ,o,verageBuyfng -Average sen 1rig Gross Mil. rket iri-g----Net Middleman Price cprice Margi n Cost Return (PesosL~g ) -- _ (Pes9s/l