Determinants of Farmer Demand for Fee-for-Service Extension in Zimbabwe: The Case of Mashonaland Central province

Transcription

1 Determinants of Farmer Demand for Fee-for-Service Extension in Zimbabwe: The Case of Mashonaland Central province Richard Foti, Lecturer Innocent Nyakudya, Lecturer Mack Moyo, Lecturer John Chikuvire, Lecturer Nyararai Mlambo, Lecturer Bindura University of Science Education P.O. Box 1020 Bindura, Zimbabwe Abstract The increasing complexity of African agriculture has put greater pressure on agricultural education and extension. The important role played by agricultural extension has led many African governments to devote a lot of resources to agricultural extension. This however is at odds with the increasing fiscal deficits and the rampant poor governance of public programs in these countries. As a result attention has been redirected towards making extension less burdensome to the governments and relevant to farmer needs. In Zimbabwe, while several studies have concentrated on describing the operation and effectiveness of the current government dominated extension system, insignificant work has gone into finding out the potential for the establishment of a private and fee for service extension system. The main objective of this paper was to find the factors determining the establishment of a private feefor-service extension system in Zimbabwe s smallholder agriculture and to give recommendations on the possible prime movers to a private, commercial agricultural extension system for smallholder farmers in Zimbabwe. A logistic regression model of binary choice was used as the major analytical tool. The study found out that the degree of commercialisation of farm enterprises, farmer income, farmer location (whether urban, rural or commercial), farm size, and risk attitude of the farmer significantly affect the demand for private fee-for-service extension and it was concluded that these variables should be considered when targeting farmers for provision of commercial extension services. Keywords: Commercialisation, Excludability, Fee-For-Service Extension, Logit Model, Risk, Subtractability Spring

2 Introduction The increased dependency on science-based agriculture bolstered by rapid agricultural technological advances in recent years has placed greater importance on the rapid and efficient transfer of these advanced technologies to farmers (Ruttan, 1987). The focus of all agricultural extension endeavours is to transfer information to farmers so as to increase their productive capacity. Srivastava & Jaffee (1992) noted that Extension serves as the link between farmers to transfer best practices of one farmer to another and to introduce or even enforce agricultural policies (p. 16). Agricultural extension economic impact studies have shown a positive effect of extension on technology adoption, farm productivity and farm profits (Birkhaeuser, Evenson, & Feder, 1991; Judd, Boyce & Evarson, 1987). The important role played by agricultural extension has led many African governments to devote a lot of resources to this sector. This is however at odds with the increasing fiscal deficits and the rampant poor governance of public programs. As a result attention has been redirected towards making extension less burdensome to the governments and relevant to farmer needs. Umali & Schwartz (1994) noted A central objective in a private fee for service extension system is in getting the right message to the right individual or group through the creation of a demand driven extension service system that is cost effective, efficient and of high quality. (p. 1) Charging for the service will ensure that the service is reaching those groups that are interested in the information and would put it into practice. Mitei (1998) noted that when farmers pay for the service, the attendance and implementation rates are greater than 70%. According to Schwartz (1992), the commercialisation of traditionally publicly provided agricultural extension services, however, raises several related issues. Will fee for service systems, necessarily lead towards greater efficiency and equity? What are the social and income distributional implications of commercialisation, in terms of access to the services by small farmers and the rural poor? Will farmers be willing to pay for the extension services? In Zimbabwe, while several studies have concentrated on describing the operation and effectiveness of the current government dominated extension system, insignificant work has gone into exploring the potential for the establishment of a fee for service extension system, which may relieve the government of most of its agricultural extension burden and may save billions of dollars by redirecting its efforts to other needy sectors of the economy. At the same time, it would give room for the identification of demand-driven knowledge gaps. This paper uses an econometric approach to answer the following research questions. 1. To what extent are farmers currently paying for agricultural extension services in Zimbabwe? 2. What are the determinants of the establishment of a fee-for- service agricultural extension system in Zimbabwe s agricultural sector? Conceptual Framework Economic Classification of Agricultural Information Welfare economics provides the analytical framework for examining the public and private good characteristics of agricultural information and in determining the efficiency of market forces. The framework is based on the principles of excludability and subtractability, which determine whether a good or service is more inclined towards being private or public (Musgrave & Musgrave, 1989). Excludability applies when access is denied to those who have not paid for the good, while subtractability (or rivalry) applies when one person s use or consumption of a 96 Journal of International Agricultural and Extension Education

3 good reduces its availability to others (Feldman, 1980). A pure private good is characterized by high subtractability and high excludability. The high subtractability and excludability characteristics of a particular commodity enable private firms to capture reasonable returns on their investments, and given competitive markets, to supply goods at optimal level (Umali & Schwartz, 1994). A pure public good, on the other hand, has low subtractability and low excludability if it is available to one person it is available to all and its use by one person reduces quantities available for others. Private firms will find it unprofitable to supply public goods, because it is difficult to restrict use Volume 14, Number 1 only to people who pay for them (the free rider problem). Between these two extremes we find toll goods and pool goods. Toll goods are characterized by high excludability but low subtractability. There is therefore an incentive for private provision of the service since people who do not pay for the service do not consume it. Common pool goods are subtractable but have low feasibility of exclusion. Private firms can only supply them if property rights can be established through government regulation (Kessides, 1992). Figure 1 shows the general representation of how agricultural information can be classified using the excludability-subtractability framework. Subtractability Excludability Public Goods Toll Goods General agricultural information, General agricultural information, Farm prices, pest infestation warnings management, marketing, processing Common Pool Goods Private Goods Modern technologies, self pollinated Modern technologies, machinery, chemicals, seeds biotechnology products, hybrid seeds Figure 1. Economic Classification of Agricultural Information. Determinants of Willingness to Pay The analytical model is based on the neoclassical utility theory of demand, in which consumers seek to maximize satisfaction subject to limitations imposed by their incomes and the prices if the goods and services they consume. That is: Maximise: U (X 1, X 2, X n ) Such that: P i X i = Y (1) Where: U = Total utility that a consumer gets after consuming goods and services; X i = Goods or services consumed; P i = Prices of goods or services; and Y = Total consumer income Farmers would be more inclined to demand fee-for-service extension if the expected satisfaction or utility from the fee for service extension is greater than that of public extension services (Beach, Syten & Rebeck, 1994). Thus the i th farmer will be more likely to pay for extension services if the utility derived from fee for service extension, U i1 is greater than the utility currently being derived from the current government provided extension services, U i0. Because there are errors in optimisation and perception, the utility function is assumed to be stochastic or random and thus: U ij = V ij + e ij (2) Where: J = indicates whether farmer is willing to pay for extension and assumes the value 1 if farmer is willing to pay for extension services and 0 otherwise; Spring

4 V ij = a function of gains or profits derived from extension for the i th farmer for the j th choice; and e ij = a random disturbance term to account for unobserved variations in preferences and errors in perception and optimisation. The probability of willingness to pay for extension services is then: P i1 = P (U i1 >U i0 ) = P(V i1 V i0 >e i0 e i1 ) (3) Where: P = Probability; and P i1 = Probability of the i th farmer s willingness to pay for extension services. We can assume that the stochastic components of equation (3) are independent and have a similar distribution. Then their difference follows a logistic distribution (Pindyk & Rubinfeld, 1991). Thus the factors influencing willingness to pay for extension services may be analysed using the logit model. Methodology Site Selection The study was conducted in Mashonaland Central Province, a province characterized by a wide variety of land tenure typologies, namely: communal areas, newly resettled small scale (A1), newly resettled large scale (A2), small scale commercial, large scale commercial and old resettlement areas). The province is also made up of areas of varying agricultural potential ranching from agro-ecological zone 2 receiving as much as over 1000mm of rainfall annually to agro-ecological zone 5 which receives as little as below 450 mm of rainfall per year. The dominant extension system is government provided and managed. However for some larger commercial farmers, some element of private fee for service extension can also be witnessed. Small-scale farmers also sometimes seek fee for service extension especially from livestock specialists. Research Process A reconnaissance visit was the starting point of the data collection process. The purposes of the reconnaissance visit were: general familiarization with the research area and the key players in the extension system, to introduce the research to local administrators and research participants, to interview key participants, to draw the sampling frames on which sampling was to be based, and to carry out sampling. The first sampling procedure was purposive and led to the choice of 4 districts based on their proximity to Bindura town (to cut on transport costs) and also on their having a wide range of land tenure typologies and agricultural potential. The sampling frame here was a total of 12 districts in Mashonaland Central province. The districts selected are Bindura, Mount Darwin, Mazoe and Rushinga. From these four districts, stratified random sampling with probability proportional to district size was then carried out. A total of 125 farmers was the targeted sample size basing mainly on the availability of resources including time and capacity of enumerators (each enumerator had to handle roughly 30 farmers. Of the four districts that were purposively selected, Bindura, Mazoe and Rushinga are approximately of the same size and thus an equal number of farmers (30) were drawn from each of these districts. For Mt. Darwin, which is larger than the other three, 35 farmers were selected to reach the targeted sample size of 125 farmers. However, due to inaccessibility of some sample elements and the fact that some questionnaires were invalid, 120 questionnaires were successfully entered in the analysis. A structured questionnaire was used to collect the data about a wide range of issues to be used in the analytical model. These issues ranged from general household 98 Journal of International Agricultural and Extension Education

5 characteristics, cropping and livestock activities and the degree of commercialisation, access to and source of extension services, farmer attitudes about different ways of extension delivery, and general audit and ownership arrangements of resources on the farms. The questionnaires were administered by a team of four trained enumerators who were then posted to districts to collect the data with the assistance of district extension officers. Data Transformation and Analysis Analytical Model Binary logit regression model was used in this study. The dependant variable willingness to pay assumes the value 1 if farmer is willing to pay for extension services and 0 otherwise. The model in its empirical form is based on the assumption that the probability of willingness to pay, P i relies on a vector of known variables (X ij ) and a vector of unknowns, β. Thus: P i = F(Z i ) = F(α+βX i ) = 1/[1+exp(-Z i )] (4) Volume 14, Number 1 Where: F(Z i ) = The standard normal density function for the possible values of the index Z i. ; P i = the probability of willingness to pay for extension services; β = regression parameters to be estimated; X i = set of explanatory variables; α = regression intercept; and βx i = a combination of explanatory variables, such that: Z i =log[p i /(1-P i )]=β 0 +β 1 X 1 + +β n X n +ε (5) Where: i = 1, 2 n are observations; Z i = the natural logarithm of choice for the i th observation; X n = the n th explanatory observation; and ε = the error or disturbance term. The variable, Z i in equation 5 is the logarithm of the probability that a particular choice (for example willing to pay) would be made. Table 1 is the empirical presentation of the logit regression model that was used in this study together with the hypothetical signs of the coefficients of variables that were included. Spring

6 Table 1 Empirical Representation of the Logit Model and Hypothesized Relationships Variable Description Hypothesized Sign Willpay Dummy variable: farmer will pay for extension service; 1 if farmer has consulted fee for service extension or indicates that he is Dependant variable willing to pay for extension services and 0 otherwise. Enters the model as the logarithm of the probability of willingness to pay. Gender Gender status of household head; 1 if male and 0 otherwise. Positive Labour Total amount of labour available per household per year (people). Positive Age Age of household head in years. Negative Educ Whether household head reached at least secondary school Positive education. Yes = 1, 0 = No. Agtrain Dummy variable: household head has formal education in Positive agriculture; 1 if yes and 0 if no. Tenure Dummy variable: 1 if farmer owns any land and 0 otherwise. Positive Farmsize Size of the total area (Ha) available to the household for farming. Positive Income Total amount of household income per year. Positive Comcrop Degree of commercialisation of crops enterprises-percentage of Positive crops sold. Comlive Degree of commercialisation of livestock enterprises-percentage of livestock sold. Positive Agrozone Agro-ecological location of farm; 1 if region 2a and 0 otherwise. Positive Cattle Dummy variable: farmer owns cattle; 1 if yes and 0 otherwise. Positive Goats Dummy variable: farmer owns goats; 1 if yes and 0 otherwise. Positive Donkeys Dummy variable: farmer owns donkeys; 1 if yes and 0 otherwise. Positive Chicken Dummy variable: farmer owns chickens; 1 if yes and 0 otherwise. Positive Maize Dummy variable: farmer produces maize; 1 if yes and 0 if no. Negative Cotton Dummy variable: farmer produces cotton; 1 if yes and 0 if no. Positive Tobacco Dummy variable: farmer produces tobacco; 1 if yes and 0 if no. Positive Propinc Proportion of total farm income coming from farming. Positive Results and Discussion Results from the study are presented in Tables 2 and 3. Table 2 shows the general characteristics of the surveyed households and presents summery statistics of the variables used in the logit regression model. General Household Characteristics Table 2 shows a tabulation of most of the data that were used in the study and a summary of the explanatory variables entered in the logit regression analysis. The following is a discussion of information presented in Table 2. Beginning with the level of education the household head, the majority of the interviewed farmers had gone through secondary school level of education and very few (only 6.2%) had never been to school. When it comes to the status of household head, most households were male headed (67.5%) with the male having the responsibility of making most of the farming decisions in the household. Very few (11.2%) households were child headed implying that decision-making is not significantly constrained by age of household head. 100 Journal of International Agricultural and Extension Education

7 Table 2 Volume 14, Number 1 Tabulation of Explanatory Variables Variable f (%) M SD Age (Years) Formal Education: Never been to school 6.2 Primary Level 23.7 Secondary 47.1 Tertiary 23.0 Annual Household Income (ZWD million) Status of Household: Male headed 67.5 Female headed 22.3 Child headed 11.2 Agro-ecological zone: Region IIa 53.4 Region III 17.5 Region V 29.1 Total labour availability (people) Degree of commercialisation crops (% sold) Degree of commercialisation livestock (% sold) Total farm size (acres) Pay for livestock extension: Yes 28.7 No 71.3 Pay for crop extension: Yes 4.6 No 95.4 Number of extension visits per month Results on household income show that the average annual income per household was ZWD million (about US ) and most of this (about 61%) was derived from farming with cotton and livestock sales emerging as the major contributors to household income. Income variability across households is very high with a range of ZWD 9.87 million (US ). Other sources of income included gold panning and formal employment mainly in mines and in the civil service. The degree of commercialisation was measured as the percentage of either total crop or livestock value that was sold per household per year. The degree of commercialisation for crops was found to be higher in agro-ecological zones IIa and III, which have a higher crop production potential than for the drier agro-ecological zone V. In region V, livestock commercialisation was found to be much higher than crop commercialisation and it was also higher than in regions IIa and III. The degree of crop commercialisation was mainly accounted for by cotton while goats and chickens dominated that for livestock commercialisation (larger stock such as cattle and donkeys are not normally sold in the short run and also they provide draft power). Only less than 10% of the interviewed farmers had ever asked for any form of fee-for-service extension. The main Spring

8 reason given was the abundance of free extension services from the government, non-governmental organisations and private companies (input suppliers and output buyers). Farmers therefore find fee-for service extension generally unnecessary especially for crops. For livestock however, the percentage of farmers who had looked for fee-for service extension was relatively higher (28.7%). This was because farmers generally value their livestock more than their crops. The interviewed farmers also mentioned that livestock related problems are normally more complicated than those for crop enterprises and they therefore need specialised knowledge, which the local communities including local, free extension systems might not have. A very small proportion of farmers were involved in irrigation agriculture. Therefore, whether the farmer irrigates was not considered an important determinant of the demand for fee-for-service extension and was left out in the final computation of the logit regression model parameters. Determinants of Willingness to Pay Results of the Logit Model Willingness to pay for fee-forservice extension was measured in terms of whether the farmer would be willing to pay for commercialised extension. The model was run with both crops and livestock combined. It is important to note that the coefficients against each of the explanatory (independent) variables represent the change in the odds ratio of willingness to pay per unit change in the explanatory variable in question. Results (Table 3) show that maleheaded households are associated with a higher willingness to pay for fee for service extension than other household typologies (female and child headed households). Labour availability has a statistically insignificant effect on the odds of willingness to pay for extension services while age of household head has a significant negative effect. The negative effect for age could be due to tradition. Older people are not used to the practice of paying for information and are therefore less inclined to pay for fee-for-service extension. Farmers who have undergone training in agriculture (whether formally or informally) had negative odds of willingness to pay for agricultural extension. They mentioned that they already have farming knowledge and find paying for extension a waste of resources. Total household income and the contribution of agriculture to total household income were both found to have positive effect on the odds and therefore probability of farmer willingness to pay for extension services. This is because higher incomes are associated with a greater ability to buy. As for the contribution of agriculture to total household income, demand for agricultural extension services depends upon the expected net benefits from investment in new information (Umali & Schwartz, 1994). This implies that farmers are more willing to pay for extension if they derive greater benefits from the commercial extension services. This is why the degree of commercialisation for both crop and livestock enterprises is associated with higher odds of willingness to pay for agricultural extension. Farmers with large farm sizes and who are in agro-ecological zones IIa and III which have high agricultural potential have a greater probability to demand fee-forservice extension than those with small land areas or who are in areas of low agricultural potential (region V). Livestock ownership and the cultivation of commercial crops such as cotton and tobacco were found to be positively associated with the odds to pay for agricultural extension. For livestock, cattle and chicken had a statistically significant positive effect on willingness to pay for extension services. This can be used as a rough indication of the value attached to these livestock by farmers. 102 Journal of International Agricultural and Extension Education

9 Table 3 Volume 14, Number 1 Results of the Logit Regression Model Variable Coefficient Adjusted R 2 P Intercept -0.56* Gender 0.75* Labour 0.21 Age -0.70* Educ 0.43 Agtrain -0.31* Tenure 0.94* Farmsize 1.88* Income 2.99* Comcrop 0.21* Comlive 1.46* Agrozone 0.65* Propinc 0.74* Cattle 0.74* Goats 0.77 Donkeys Chicken 0.40* Maize 0.11 Cotton 0.85* Tobacco 0.75 Note. The dependant variable is Log(probability willpay). *p < Conclusions and Recommendations The current level of farmer demand for fee-for-service extension is very low. This is more so for crops than for livestock. Crowding out of private commercial extension services by free extension from the government, non-governmental organisations and private companies were found to be the chief causes for this low demand. The oversupply of free but inappropriate extension by most governments in developing countries results in farmers not feeling the need to pay for commercial extension services. There is high demand for commercial extension for livestock enterprises than for crop enterprises due mainly to the value that farmers attach to their livestock as a result of their role as measures of wealth and their use for draft power (for cattle). The other reason is also because livestock problems such as diseases require specialised skills which farmers or local free extension might not have. Among crop enterprises, crops that yield greater financial returns or that had a high commercial value were found to increase demand for commercial extension. Commercial extension should therefore target high value crops and livestock and farmers who are large and located in high farming potential areas and therefore are capable of spreading the costs across a wider income base. These farmers would also have a higher willingness to pay since they derive higher benefits from agriculture and therefore have a lower psychological marginal cost for commercial extension. By targeting farmers with high farm sizes, good quality land, high input output prices differences (for returns to farming), Spring

10 easy access to credit (for additional income), and more permanent land tenure arrangements, the commercialisation of agricultural extension would take advantage of the high incomes of these farmers and hence their greater abilities to pay for extension. Most importantly, the government, NGOs and other private companies that provide free extension services to farmers should not compete directly with private, commercial extension providers for areas where the demand for fee-for-service extension is high. The provision of free extension services should only be to those farmers who are in need. This study however, does not provide a complete exploration of all the factors that affect the demand for fee-forservice extension in the country let alone in less developed countries. It however provides a methodology for carrying out a wider and more encompassing study that would give recommendations that are applicable across countries and farm typologies. References Beach, M., Syten, J., & Rebeck, S. (1987). New developments in agricultural extension, privatisation of extension services. Journal of Extension Systems, 3(2), Birkhaeuser, D., Evenson, R., & Feder, G. (1991). The economic impact of agricultural extension: A review. Economic Development and Cultural Change, 34(2), Feldman, A. M. (1980). Welfare Economics and Social Choice Theory. Boston: Kluwer Nijhoff Publishing. Judd, M. A., Boyce, J. K., & Everson, R. (1986). Investing in agricultural supply: The determinants of agricultural research and extension investment. Economic Development and Cultural Dhange, 35(1), Kessides, M. (1992). Externalities and the returns to agricultural research: Discussion. American Journal of Agricultural Economics, 71(2), Mitei, R. (2001). Improving extension systems in tropical agriculture. Washington, D.C.: The World Bank. Musgrave, M., & Musgrave. T. (1989). Agricultural research in the private sector in Africa. The case of Kenya. The Hague: ISNAR. Pindyk, S., & Rubinfeld, K. (1991). Econometric models and economic forecasts. Konark Publishers, USA. Ruttan, V. (1987). Some concerns about agricultural research policy in the United States. Quarterly Journal of International Agriculture, 26(4), Schwartz, L Private technology transfer in Sub-Saharan Africa. Washington, D.C. Srivastava, J., & Jaffee, S. (1992). Seed systems development, the appropriate roles of public and private sectors. World Bank discussion paper number167. Washington, D.C.: The World Bank. Umali, D. L., & Schwartz, L. (1994). Public and private agricultural extension: Beyond traditional frontiers. Washington, D.C.: The World Bank. 104 Journal of International Agricultural and Extension Education