IDENTIFYING SERIOUS FARMERS in the FORMER CISKEI: Income, Although Lying Third Overall, Was Very Much the May 1993

Agre/con, Vol 34, No 2 (June 1995) Helm and Van Zyl Agre/con, Vol 34, No 2 (June 1995) Eckert and Williams capita income, in South Africa (1993) and in some of GENERAL AGREEMENT ON TARlFFS AND the OECD countries (1992). Agricultural support in TRADE. (1993). Trade Policy Review Mechanism, The South Africa expressed as a percentage of per capita Republic of South Africa, Report by the Government, 3 IDENTIFYING SERIOUS FARMERS IN THE FORMER CISKEI: income, although lying third overall, was very much the May 1993. IMPLICATIONS FOR SMALL-SCALE FARM RESEARCH AND LAND same as was the case with both Canada and the United States. HARWOOD, J.L & BAILEY, KW. (1990). The World REFORM Wheat Market - Government Intervention and Taking the uneven distribution of income in South Multilateral Policy Reform. Economic Research Jerry B. Eckert Africa into account, the Blacks are worst off, with a Service, United States Department of Agriculture. Staff Professor, Department ofAgricultural Economics, University ofStellenbosch. percentage of per capita income of 3.39 percent, Report No. AGES 9007, January 1990. Coloureds 1.63 percent, Asians 1.11 percent and Whites William Williams only 0.45 percent HATIIAWAY, DE. (1987). Agriculture and the GATT: Doctoral candidate and former research officer, Small Farm Systems Research Project, Agricultural and Rural Rewriting the Rules. Institute for Economics, Development Research Institute, Fort Hare University 5. Conclusion Washington, D.C., September 1987. Rural ~ousehold_ survey da~ from Mgwalana (Eastern Cape) are used to assess the intensity of involvement in agricultural The comparison of support data between countries, HELM, W. (1994). Agricultural Support in South ent~n~. This area typifies the expected results of Alan Low's theory regarding disincentives to farming in southern bearing in mind the recognised limits of these Africa. Unpublished MSc(Agric) dissertation, Africa, 1e. m~st households do no~ acti!ely ~ their !and resour~. Statistical stratification methods are developed and indicators, gave a clear indication of the relative extent Department of Agricultural Economics, Extension and ~st~ to proVI~ ~e means for"q~ckly identifying senously coIIIIIlltted farm households. While the percent of households of agricultural support in South Africa. With the Rural Development, University of Pretoria, Pretoria. 1dentifi~ as senous farmers 1s small, they acc:ount for a disproportionately large share of the region's agricultural exception of Australia and New Zealand, South Africa production ~d farm resource ~- -piey thus constitute an attractive recommendation domain for farming systems research had a relatively low degree of support compared to the HELM, W & VAN ZYL, J. (1994). Domestic and extension programs. Implications from Mgwalana are drawn for land reform and support programs for emerging other developed countries. agricultural support in South Africa from 1988/89 to farmers. 1993/94: A calculation. Agre/con, Vol 33, No 4 : 213- Bearing in mind the low per capita income of the 219. IDENTIFISERJNG VAN ERNST/GE BOERE IN DIE CISKEI: IMPLIKASIES VIR KLEINBOERNAVORSING EN GRONDHERVORMING. majority of South Africans, the question still remains to Opnamedata van landelike huishoudings in Mgwalana (Oos-Kaap) word gebT7'ik om die mate van betrokkmheid in what extent South Africans can afford even the current OECD. (1992). Agricultural Policies, Markets and Trade - 1991, OECD, Paris. landbou-ondememings te bepaaL Hierdie gebied illustreer die verwagte resultate van Alan Low se ontmoedigingsteorie relatively low levels of support within the agricultural ten ~p3!gte van ~oerd_ery in_ suidelike Afri~. d. w.s. dat die meeste huishoudings nie hu1 grondhulpbronne aktief benut nie. sector. The issue clearly is not only how South African OECD. (1993). Agricultural Policies, Markets and Statisties~ stratifiktu1etegn1e~ ~rd-onr_w;kkel en. getoets ten einde n metode daar te stel waarvolgens plaashuishoudings agricultural support compares to competitors, but also wat emstig,!ot landb_ou verbind 1s v,nmg ~e'i'fen!ifiseer kan word. H~el _die persentasie huishoudings wat as "ermtige one of affordability and specifically who benefits and Trade - 1992, OECD, Paris. landbouers gerdentifiseer kan word, kle,n 1s, 1s hulle verantwoordel1k v1r 'n buite verhouding groat geckelte van die who pays for it gebied ~e landb_ouP_roduksie. en benutting van plaashulpbronne. Bulle verteenwoordig dus n aantreklike RUNGE, F. (1988). The Assault on Agricultural aanbevelmg_sterrem vir nay,orsmg oor boer:Ierystelsels en voorligtingsprogramme. Vanuit die studie van Mgwalana word Notes Protectionism in the Multilateral Trade Negotiations. gevolgtrekkings gemaak v1r grondhervormmg en ondersteuningsprogramme vir opkomende ldeinboere. University of Minnesota. Staff Paper P88-10, April 1. This article is based on a MSc(Agric) dissertation by 1988. William Helm at the University ofPretoria. 1. Introduction ploughed, unplanted (Rose and Williams, 1988). Those UNISA (1994). Socio-economic profile of the nine that were planted often suffered from lack of related provinces in South Africa, 1994. Bureau of Market 2. This n:search was conducted while William Helm was In 1988, the authors examined socio-economic data from inputs, primarily weeding and fertilizer. To the layman, Research, Research report No. 207, University of South employed by the Directorate Mmketing, Department of a sample of African farmers in the former Ciskei. The the overall picture is one of an abandoned or only Agriculture. He is presently with ABSA Bank. Africa (UNISA), Pretoria. intent at the time was to develop a method of stratifying passively used land resomce. In the former Ciskei, the farm households to identify those with higher research problem is not so much identifying a group of USDA (1990). Estimates of Producer and Consumer probabilities of responding to and benefitting from a References target farmers with common characteristics from among Subsidy Equivalents, Government Intervention in farming systems research/extension project which was a diverse farm population but rather one of identifying Agriculture, 1982 - 1987. Economic Research Service, then in progress (Williams, et al., 1988). Current CHINA POST. (1994). US-Japan trade dispute heats serious farmers, households for whom farming is a Statistical Bulletin No. 803, Washington, D.C, April movement toward a significant land reform provides up. Taipei, 17 February 1994. significant enterprise if not a life style, from among the 1990. additional contexts for this analysis and an urgency for its dis:iemmation. The original issue, identifying high general rural population. A similar problem faces DEPARTMENT OF AGRICULTURE. (1994). Abstract potential, or "serious" farmers and quantifying their designers of the forthcoming land reform, identifying of Agricultural Statistics, Directorate Agricultural characteristics, remains a valid research topic. recipient households that will likely utilize newly Economic Trends, Pretoria. However, in addition to targeting farm support acquired farm lands at or near their agricultural programs, such analysis should also now be useful in potential. Survey data and field observations indicate guiding land redistribution toward those households that the "serious farmer" criterion might limit the most ~ely to use agricultural land productively. selected group to only a small portion of the rural Further, m the context or emerging farmers, quantifying population. Yet there is sound developmental logic for the c~teristic~ and c~nstraints of high potential farming systems research projects and other support farmers 1s essential to gwde research on appropriate technologies for these conditions. Finally, in the policy programs to assist these emerging farmers to develop arena, there is much current optimism about land reform their farm enterprises and to better utilize the country's and its possible contributions to various politically limited agricultural potential. This paper develops two endorsed, rural reconstruction objectives. Data reported methods of identifying serious farmers using survey data here offer sobering insights into the existing incentive from the former Ciskei, quantifies the farm and family structures faced by small scale agriculture and their resources at their command, and discusses implications possible impact on farm management and output The of these findings for a possible land reform. analysis below suggests that at least some of the current optimism may be misplaced. 3. Theoretical and Practical Background 2. The Research Problem in the Former 1 Alan Low's household economics model (1986a), Ciskei developed with southern African data, provides strong logic to explain the relative lack of serious attention to The former Ciskei is largely rural. Approximately 44 farming in areas such as studied here. He reminds us percent of rural residents are land holders with arable that economic rationality will allocate household labour holdings averaging some hectares. However, in 1988, to its highest paying opportunity. He then notes that in 42 percent of these fields lay unploughed, or if southern Africa this is frequently off-farm in the

49 - Agre/con, Vol 34, No 2 (June 1995) Eckert and Williams Agrekon, Vol 34, No 2 (June 1995) Eckert and Williams

relatively well developed non-agricultural labour men left at home may grow some crops, they face A common mistake in earlier agricultural development efficient method of selecting farmers from among rural market. This off-farm market, Low suggests, dominates another rational economic threshold defined by the programs was to assume that "small farmers" were an households was needed. household work incentives and labour allocations. He household's internal food needs. Beyond that undifferentiated group which could be accurately also theorizes that, after migration to off-farm quantitative threshold, food production is valued at described with mean or median statistics. Technologies 4. The Mgwalana Study employment, labour remaining in the rural household market rates and these are set by the productive capacity were developed for average or median farmers in an will be allocated first to production for home of large, well capitalized, white owned farms. Per attempt to ensure widespread adoption and to pre-empt 4.1 Methods consumption which is valued at retail food prices plus kilogram gross margins, when applied to the yield levels a concentration of the benefits of publicly supported transportation costs to the kitchen, and only last to of traditional farming, simply do not justify allocating research among larger, more affiuent farmers. The The Agricultural and Rural Development Research production for sale which is valued at lower farm gate, very many scarce resources to small scale agriculture. frequent failure of these programs proved that much Institute (ARDRI) at the University of Fort Hare unprocessed commodity prices minus transport to greater variability existed among farmers than undertook several benchmark surveys from 1985-1987 market. In the study area, this situation is particularly previously assumed. It became obvious that even micro under a Small Farms Systems Research Project accentuated. Nearly half of the household heads are 60 variations in the physical, economic and institutional sponsored by the Development Bank of Southern Africa. This theoretical structure has much to say about the years old or older and thus retired, or nearly so, from environment for farming can significantly affect Prior to this time, there were few statistics on farm difficulties facing small scale farming in the region. wage employment. Furthermore, thirty to forty percent technology adoption decisions. The concept of "target populations in the region and descriptive information Low's model suggests that off-farm employment of households are headed by women. Of the younger farmers" emerged in the late 1960's as a means of was needed before designing technical improvements opportunities seriously deplete the available labour household heads, two-thirds are employed off-farm. An gaining further precision in identifying small farmer for traditional agriculture. The resulting data base supply of rural households for farming. Workers additional 21 percent of wives are also involved in the groups in need of particular program attention and to provided several village-level samples for analysis. For remaining on the farm are those with the lowest off-farm labour force (Rose and Williams, 1988). discover their particular needs and characteristics. the present study, a 1987 survey of the Mgwalana opportunity costs as defined by the external labour Household incomes in 1988 averaged R 2 315 per year Target groups were normally identified by some Tribal Authority area was chosen. The sample was market. The off-farm market favours men. Thus many of which 94.4 percent was non-agricultural. Of the combination of household characteristics, area large enough for statistical analysis of representative rural households are de facto headed by women for remaining income, an estimated two-thirds is the value cultivated, dominant enterprise and geographic location. patterns. And at the time, it was the most recent of the whom household and child rearing responsibilities pre of home consumption. Agricultural cash incomes, ARDRI surveys, thus incorporated all methodological empt extensive field labour in agriculture. Labour averaged across this rural population, amounted to only As the on-farm research orientation of the early 1970's refinements arising from previous studies. Mgwalana is intensive farm technologies are probably not appropriate R 42 per year per household (Williams, 1987). These became farming systems research (FSR) in the late a tribal area administered then by Chief Zibi in the in this setting even though they might be elsewhere in data showing the relative unimportance of farm incomes l 970's and farming systems research and extension Middledrift district. Access to the region and Africa. Second, while the women, children and older are captured in Figure 1. (FSR/E) in the mid-l 980's, the concept of permission for the study were obtained following "recommendation domains" supplanted "target groups". extensive discussion with Chief Zibi and his advisors. The idea of recommendation domains recognizes that The former Ciskei Department of Rural Development the determinants of farmer decision making include not assisted in financing the study and provided only household characteristics, arable land holdings and enumerators. Cattle 41.0% enterprise mix, but also the totality of the physical, Sheep 36.5% biological, social, economic and institutional setting in Mgwalana contained five residential locations ( extended Goats 11.4% which the farmer operates. By definition, the term villages) with a total population of 620 households. A Pigs 11.1% focuses on the technology diffusion component of the 20 percent sample (n = 125) was randomly selected research/extension continuum. Low (1986b:82) defines from household lists provided by headmen. Xhosa recommendation domain as "an homogeneous group of speaking enumerators were trained and field supervision fanners who share the same problems and possess was detailed, with daily verification of questionnaires by Livestock 80% Crops similar resources for solving these problems. This ARDRI supervisors. Data were analyzed at the 20% group offanners is expected to adopt (or not adopt) the University ofFort Hare computer centre. same recommendation, given equal access to NON-AGRICULTIJRAL information abaut it." 4.2 Simple Correlations INCOME Subsistence Production 94.4% 3.8% Several authors have suggested recommendation The search for patterns that would cluster into domains for southern Africa. Low (1986b:87) suggests recommendation domains used Pearson correlation a stratification based on availability of household labour coefficients as a starting point. Relationships between CASH CASH and draft power but does not provide quantitative selected variables which are significant at the 99 percent HOUSEHOLD INCOME parameters for the four groupings described. Eckert, et level or greater are shown in Table 1. The first ten 100% al., ( 1982) disaggregated rural landholders in Lesotho items in the table are simple measures of agricultural HOUSEHOLD INCOME HOUSEHOLD INCOME on the basis of control over farm equipment, animal activity. Items 11-16 are social and household draft and availability of migrant remittances. The intent characteristics. Items 17-I 9 are a subset of the was to establish for agricultural planners, who had household parameters; those dealing with characteristics previously touted the egalitarian nature of Lesotho's of the households' labour complement. Table l provides Subsistence Production agriculture (Lesotho, 1976), that important differentials the basic analytical point of entry into understanding 3.8% existed which warranted separate development agriculture in Mgwalana. strategies and policies. Tschabalala and Holland (1986) extended Eckert's work to formally specify the 4.2.1 Socio-Economic Relationships parameters of recommendation domains for farming systems research in Lesotho. Their graphical analysis Socio-economic relationships in Mgwalana generally Livestock 36.2% Crops 63.8% confirmed that differences do exist based on resource follow expected patterns among themselves. Y 0W1ger control and access to migrant remittances. Relating household heads tend to have more schooling. Y0W1ger these stratifications to cropping parameters, their individuals are more likely to be involved in off-farm evidence was suggestive but mixed. Statistical tests of employment which in these data appears as "residential Cattle 14.3% deterministic significance are, unfortunately, not status" with higher numbers reflecting longer periods Sheep 33.1% provided in these earlier studies. fu the context of this away from home at work. Residential status and age of Goats 37.2% study, and recognizing the impediments to traditional the household head correlate as expected with total off Pigs 15.4% agriculture suggested by Low's observations, we sought farm income. Schooling is apparently a determinant of a recommendation domain defined as those households access to off-farm work. As expected, off-farm income Source: Williams, 1987 seriously interested in farming in order to focus relates positively to total expenditures and savings. appropriate farmer support interventions. In short, an Figure 1: Sources of income for rural households in Ciskei, Southern Africa

51 52 - Agre/con, Vol 34, No 2 (June 1995) Eckert and Williams Agrekon, Vol 34, No 2 (June 1995) Eckert and Williams

51 52 Agre/con, Vol 34, No 2 (June 1995 Eckert and Williams Agre/con, Vol 34, No 2 (June 1995) Eckert and Williams

Conventional practice in small fann research is to Involvement in household gardening is enigmatic in asswne that various socio-economic household these data. No clear pattern of correlation with either parameters link, in some deterministic way, with household or farming parameters. A limited number of ~ ~ farming enterprise. Table 1 suggests that this interviews conducted to supplement the main survey generalization cannot be uncritically applied to South gave impressionistic evidence that an interest in African traditional agriculture. Perhaps as important as vegetable gardening is passed from mother to daughter ~ ~ the data in this table are the blank spaces, marked as and that garden size may be a function of interest more boxes A and B. The relative absence of entries in these than need. Further, several of the largest gardens were boxes indicates that these specific correlations did not found in female headed households with children but no !::: N meet the chosen threshold level of statistical adult males present Perhaps a recommendation domain ~ "'l ~ significance. As shown by the empty Box A, several for garden technology must be separately identified. standard socio-economic household characteristics bear :2 0 little relationship to agricultural activities, e.g., One of the more interesting observations to emerge from Cl schooling, residential status and off-farm income. Age the Mgwalana data is the relative absence of - of the household head, often used as a proxy for the correlations between cattle numbers and other key 0 stage of the household in household life cycle analyses, variables. This contrasts with relationships evident with 0 .., 0 Yet in these data, the household head's age correlates sheep and goats. Ignoring significance levels of less -"' ...; "! <'! well with only a few agricultural activities. Schooling than 1 percent, cattle numbers are 112! correlated with and residential status, while they correlate with off-farm household labour units, the labour.land ratio, adult :!: 0"' employment and thus the bulk of income, have little males in the household, summer cropping, cropping ~ C'"'!~ i:Q relationship to farming. intensity or sales of animals and animal products, - whereas in each case small stock numbers are. Cattle Those household characteristics that do correlate well numbers are correlated with garden si7.e whereas sheep :! 0.., ~ ~ "! <'! with agricultural activity form a distinct subset; the and goat numbers are not Cattle numbers are positively - various measures of household labour supply. Total associated with the level of involvement in other N family size also correlates well but was omitted from livestock enterprises, with off..fann incomes and savings !::! ~ <'! Table 1 because it did not improve on relationships for and with the number if implements present household labour force expressed in adult equivalents. 0 ..,0 .., Labour.land ratios emerged as particularly strong, being The implication of these contrasts is that cattle - Cl .. ~.. ~.. <'! the only labour measure highly com:Iated with the farm ownership and management appear to be governed by - - enterprises in which women dominate: eg. pigs, poultry somewhat different criteria in Mgwalana than is 2 N Cl\ and gardens. In something of a surprise, these labour ownership and management of other stock species. 5~ "! <'! ~ measures did not correlate well with other household While the data do not specifically define the role of characteristics (Box B). cattle, they can be read to suggest that other stock .., species are managed more as agricultural enterprises °' 5 -<'! "! Of interest in the African context is the pattern observed than are cattle. This finding would support a "store of

for savings. Monetary value of savings correlates well wealth" concept for cattle although cultural, social and 00 8.., r- o- .., with holdings of cattle, sheep and goats but not with political attributes could be explanatory as well. ....: "1 <'! "! <'! <'! pigs and poultry. This is consistent with the widely accepted hypothesis that livestock, especially cattle, are 4.3 Compantive Analysis I"- .., N ~ "! <'! °'<'! used as a store of wealth under traditional farming - conditions (Krige, 1936~ Tapson and Rose, 1984). In The above correlations suggest several interrelated ID 00 ID Table 1, the higher are off-farm income, savings or total factors which might be utilized to stratify rural ~ -"! "! °'"! <'! expenditure, the higher also are savings and ownership households and identify active farmers. Table 2 - < of ruminant animals. Savings arc not closely associated compares selected single dimension descriptors with 80 r-..., 0 with any measure of cropping activity. others combining two or more characteristics. In the "' ...: C'"'! ~ ~ <'! <'! <'! <'!- latter category are three bi-variate combinations of land,

4.2.2 Intn-Agricultunl RelatioMhips labour and animal ownership. The Mgwalana data also 00 .., .., 'D - 0"' N N 0 permitted explicit comparisons with the three-way ~ <'! "! <'! <'! ~ "! "! "! <'! -<'! Many significant correlations were measured between stratification used by Tschabalala and Holland. Several - various agricultural enterprises. Most of these alternative class boundaries were explored for each .., ... ~ 0 I"- r-. 0 ... relationships support the general picture that active stratifier and those with the highest discriminatory "!- <'! <'! <'! ~ <'! <'! --"! "'l <'! farmers tend to be involved in several crop and livestock power were utilized. Figures in the body of Table 2 are - .., .., enterprises simultaneously. For example, farmers with estimates of the probability of a larger value than the N ~ ... r-. 'D ... ID the larger areas of swnmer crops also have larger measured Chi-square. °'"'! ~ °'<'! "'! <'! ~ r-: -<'! "! "'"! "! <'! <'! numbers of most stock species, higher cropping - intensities and more farm implements. Not shown in Perhaps the most striking observation is the robustness r-. .., r-. 00 'D 'D 'D "! "'~ -"! "'! ~ "'"! <'! <'! "'! <'! "! <'! ~ <'! -"! Table 1 but also significant, they also have larger areas of the simplest stratification in the da~ the number of - ~ °' °' of winter crops and sell more crops, livestock and livestock species. Five species were enumerate~ cattle, livestock products such as milk, eggs and wool. This sheep, goats, pigs and poultry. Table 1 shows the ·a ,.._ ,.._ ,.._ ,.._ ,.._ ,.._ tightly woven set of interrelationships establishes the interrelatedness of numbers of each species owned. :::i g g g fact that serious farmers do exist in Mgwalana and that Simply tabulating the number of stock species proved ! ! ! ! ! "'s l ££ £ ~ I! they tend to be multiple enterprise farmers. This highly effective as an indicator of the intensity of other contrasts with the relatively few correlations between farming activities as well. § socio-economic characteristics and agricultural ..:i ~~ activities. Methodologically, this suggests that the most At the bottom of Table 2, various stratification methods ] ] ~i .5 " ,.._l ·a::, l ·i effective indicator of overall agricultural involvement is are compared based on the strength of their relationship :i:: ~ "' .I OIi ] "' likely to be one or several agricultural measures with five key indicators: ruminant stock units, chickens ! i -~ t :i:: ] 1 ; I &:I gi, 1 ~ 0 J themselves. This insight is tested more fully in a and pigs, hectares of summer crops, garden size and 1 :i:: i:i:: -~ :i:: subsequent section. implements possessed. ] j -I j :i:: :i:: ) lCl.I t ~ c'J t J:i:: r'1 I l J I .,; I0 00 0\ ,.; .,; ... ::!; - N -.r ID r-: 00 °' ---~ !ti :2 !:: ... - .. .0 U oi:, 0 Agre/con, Vol 34, No 2 (June 1995 Eckert and Williams Agre/con, Vol 34, No 2 (June 1995) Eckert and Williams

4.2.2 Intn-Agricultunl RelatioMhips labour and animal ownership. The Mgwalana data also 00 .., .., 'D - 0"' N N 0 permitted explicit comparisons with the three-way ~ <'! "! <'! <'! ~ "! "! "! <'! -<'! Many significant correlations were measured between stratification used by Tschabalala and Holland. Several - various agricultural enterprises. Most of these alternative class boundaries were explored for each .., ... ~ 0 I"- r-. 0 ... relationships support the general picture that active stratifier and those with the highest discriminatory "!- <'! <'! <'! ~ <'! <'! --"! "'l <'! farmers tend to be involved in several crop and livestock power were utilized. Figures in the body of Table 2 are - .., .., enterprises simultaneously. For example, farmers with estimates of the probability of a larger value than the N ~ ... r-. 'D ... ID the larger areas of swnmer crops also have larger measured Chi-square. °'"'! ~ °'<'! "'! <'! ~ r-: -<'! "! "'"! "! <'! <'! numbers of most stock species, higher cropping - intensities and more farm implements. Not shown in Perhaps the most striking observation is the robustness r-. .., r-. 00 'D 'D 'D "! "'~ -"! "'! ~ "'"! <'! <'! "'! <'! "! <'! ~ <'! -"! Table 1 but also significant, they also have larger areas of the simplest stratification in the da~ the number of - ~ °' °' of winter crops and sell more crops, livestock and livestock species. Five species were enumerate~ cattle, livestock products such as milk, eggs and wool. This sheep, goats, pigs and poultry. Table 1 shows the ·a ,.._ ,.._ ,.._ ,.._ ,.._ ,.._ tightly woven set of interrelationships establishes the interrelatedness of numbers of each species owned. :::i g g g fact that serious farmers do exist in Mgwalana and that Simply tabulating the number of stock species proved ! ! ! ! ! "'s l ££ £ ~ I! they tend to be multiple enterprise farmers. This highly effective as an indicator of the intensity of other contrasts with the relatively few correlations between farming activities as well. § socio-economic characteristics and agricultural ..:i ~~ activities. Methodologically, this suggests that the most At the bottom of Table 2, various stratification methods ] ] ~i .5 " ,.._l ·a::, l ·i effective indicator of overall agricultural involvement is are compared based on the strength of their relationship :i:: ~ "' .I OIi ] "' likely to be one or several agricultural measures with five key indicators: ruminant stock units, chickens ! i -~ t :i:: ] 1 ; I &:I gi, 1 ~ 0 J themselves. This insight is tested more fully in a and pigs, hectares of summer crops, garden size and 1 :i:: i:i:: -~ :i:: subsequent section. implements possessed. ] j -I j :i:: :i:: ) lCl.I t ~ c'J t J:i:: r'1 I l J I .,; I0 00 0\ ,.; .,; ... ::!; - N -.r ID r-: 00 °' ---~ !ti :2 !:: ... - .. .0 U oi:, 0 Agre/con, Vol 34, No 2 (.lune 1995) Eckert and Williams Agre/con, Vol 34, No 1 (.lune 1995) Eckert and Williams

These five were chosen to provide a relatively balanced lends itself to rapid and unambiguous identification of distribution of indicators among areas of agricultural the more involved farmers. endeavour. Significance levels for each indicator can be ranked by their maximum value or by the median value Table 3 illustrates mean differences in several among the five. Two stratifications emerge as notably parameters for subgroups defined by these two better than others: number of stock species and a classification systems. Originally the cattle x area 1 combination of cattle owned and arable area available. classification was a 2 x 2 matrix, however, : distinguishing between smaller and larger areas within 'c Tschabalala and Holland combined cattle, household cattle owning fiums added little to explanatory power. .e labour and implements to derive their recommendation Because the number of sampled households with cattle domains. That their measure does not rank well in was limited, these two subgroups were pooled to obtain Mgwalana reflects, in part, differences between Lesotho an adequate sample size for the combined cell. Within and Ciskei in the importance of cattle as draft animals. each classification system, all between-group differences Nearly all ploughing in Ciskei is done by government or are significant at the one percent level except for goat privately owned tractors. When Tschabalala and ownership. I"a Holland collected their field data in 1980, cattle were still a significant draft resource in Lesotho. Their Comparing the recommended target group with the I classification system also suffers in this comparison middle group in each case leads to the following through its weak association with garden size ( a variable generalizations. Households seriously involved in i not measured in their study) and pigs and poultiy, farming are larger by some three persons, have at least enterprises particularly suited to female-headed one additional adult male residing at home, and their households. Both of these associations should have family labour force is two full adult equivalents greater. I been particularly important in Lesotho where female Household heads average 7-8 years older than those of headed households are a major factor in rural society. other subgroups. These farms comprise some 17 i percent of rural households, but, due to their larger size, - 5. Conclusions and Recommendations account for approximately one-quarter of the rural population. I 5.1 Recommendation Domains for Mgwalana ~ Active farmers in Mgwalana are characterized by 1-1.5 l Recommendation domains are used to disaggregate farm hectares (40-80 percent) more summer cropping, 200- -- populations into sub-groups which might warrant group 300 percent larger gardens, and about 400 percent specific development strategies. It is common to isolate greater holdings of stock. Although less than one-fifth progressive farmers who receive advanced technology, of rural households, they grow one-third of the summer j "median" fanners for less sophisticated and less crops, cultivate two-fifths of the total garden area, and expensive packages and perhaps other special groups. raise over half the sheep and nearly all the cattle. Thus .!l Most fanning systems research programs then agricultural assistance programs can PQtentially have a concentrate on the median or smaller, limited resource major impact on agricultural resource use and farms, motivated by a desire to impact larger numbers of commodity output by concentrating on a limited number f the rural poor and to avoid appropriation of research of carefully selected households. i results by more affluent farmers. Present evidence suggests that the farming households j The task differs conceptually in Mgwalana. Rural identified by either of these two measures would be residents and landholders are, for the most part, only suitable recommendation domains for an on-fann - passively interested in agriculture as a result of the systems research program in the Mgwalana Tribal incentive situation discussed above. The need is to Authority area. Following preparation of an early draft !... identify a group that was active enough in farming to of this paper, the research was extended across the warrant an applied research program. This requires former Ciskei using data from previous socio-economic searching for the upper tail on a frequency distribution surveys conducted by ARDRI in Gaga, Sheshegu and of agricultural involvement, rather than looking for Khambashe (Williams, et al., 1988). Most of the J median or perhaps second quartile farmers. In relationships reported here for Mgwalana and the Mgwalana, "progressive" farmers may well be the only principal conclusions were sustained from analysis of small holder able to respond to new technologies. these other samples.

Either of the two most effective classification systems 5.2 Implications for Land Reform and Rural identified in Table 2 serve to identify the desired farm Development group. Active small holder farmers in Mgwalana can be quickly identified as those possessing four or more stock In Mgwalana, off-fann work dominates rural income species, or as those possessing cattle. These two groups sources and thus household time allocations. are not completely overlapping sets. In choosing Nevertheless, slightly less than 20 percent of rural between the two, we would recommend the species households are very active in farming. This subgroup count method for four reasons. First is the better fit differs significantly from the remaining rural population with intensity of involvement in predominately female in many household characteristics as well in as their activities of pigs, poultry and vegetable gardening. involvement in cropping and livestock husbandry. These Second, use of a classification system based on cattle households constitute a group which would probably and land might miss farmers who are intensively respond to technologies, infrastructure and institutions involved in small stock and other enterprises. Third, as developed for their specific needs. The quickest and discussed above, there is the question as to whether most effective means of identifying serious farmers in cattle are truly managed as an agricultural enterprise. this environment is a simple enumeration of the number Last, the sheer ease of counting stock species managed of livestock species managed by individual households.

55 56 Agre/con, Vol 34, No 2 (.lune 1995) Eckert and Williams Agre/con, Vol 34, No 1 (.lune 1995) Eckert and Williams

55 56 Agrekon, Vol 34, No 2 (June 1995) Eckert and Williams Agrekon, Vol 34, No 2 (June 1995) Eckert and Williams

TableJ Mean characteristics of recommendation domains identified by two classification system• for his contributed wisdom and tireless support to this MEYER, AT. (1994). Policy statement on land reform. paper. I Proceedings of the founding meeting, Farm Management Species count I Cattle x arable area Association of Southern Africa, Elsenburg. I Item I No cattle No cattle area Have cattle 0-1 2-3 4-5 I Area~ .5 ha

Conversely, more than 80 percent of rural households in the amount of uncultivated land in the former Ciskei has the sample area use their available land resource very increased from the 42 percent measured in 1987 to extensively if at all. This contrast suggests important perhaps over 60 percent at present. The apparent reason considerations for the design and implementation of a lies in the collapse of the availability of ploughing land reform program in South Africa. First, a random services provided by the former Ciskei government selection of current residents in historically black rural during the time of political transition (A.O. de Lange, areas to become recipients of redistributed land runs the personal communication, 1994 ). significant risk of cloning the passive resource use patterns found in Mgwalana and elsewhere in the Government 1s committed to coupling land former Ciskei. Purposive efforts to select serious and redistribution and the emergence of a small-scale capable farmers to receive land would seem well farming sector with provision of farmer support advised. "International experience clearly indicates services and a network of redirected agricultural that the characteristics of those who participate_i'! a institutions (Meyer, 1994). Alan Low's theory taken land redistribution program (are) a critical factor in together with the analysis here strongly suggests that detennining the success or failure of the program" such future public support must concentrate on (World Bank, 1993:37). This would seem particularly increasing the returns to labour applied to small-scale important if, as recently advocated (Cooper and van Zyl, farming. Finding small farm options that compete 1994 ), land reform is to make a significant contribution successfully with off-farm employment is not easy in to increased food security. If land redistributions are to South and southern Africa (Eckert, et al., 1980). be targeted, the analysis above provides a quick and However, it can be done, as nearly 20 percent of rural apparently accurate method of identifying potential households in Mgwalana have shown. Discovering the recipients. right enterprises, developing the appropriate technologies and sustaining them institutionally must Second, much of the neglect of agriculture in the former become guiding principles of agricultural policy under Ciskei can be traced to the fairly complete lack of public South Africa's emerging agricultural dispensation. investment in the supportive institutions, markets and appropriate technologies that might have made small Acknowledgements scale farming more economically viable. That institutional support is essential is emphasized by a The authors are indebted to the late Cecil J. Rose, recent observation in the region. Since 1988, and former chief research officer, Agricultural and Rural despite growing national and regional unemployment, Development Research Institute, University of Fort Hare

57 58 Agrekon, Vol 34, No 2 (June 1995) Eckert and Williams Agrekon, Vol 34, No 2 (June 1995) Eckert and Williams

57 58