Socio-economic characteristics and allocative efficiency of upland rice farmers in ,

D. O. AWOTIDE* & A. S. BAMIRE (D. O. A.: Department of Agricultural Economics, Olabisi Onabanjo University, Campus, Ayetoro, Ogun State, Nigeria; A. S. B.: Department of Agricultural Economics, Obafemi Awolowo University, Ile-Ife, Nigeria) *Corresponding author

ABSTRACT The paper analysed the relationship between allocative efficiency and a set of socio-economic variables. A multi-stage sampling procedure was used to select 165 rice farmers from six local government areas (LGAs) from the 10 rice-producing LGAs in Ogun State. The paper considered a Cobb-Douglas stochastic frontier cost function applied to farm-level data of upland rice farmers in Ogun State to empirically determine the level of allocative inefficiency, using a single-stage model. The theoretical model predicts a positive relationship among cost of production, capital and output of paddy rice. The study showed a positive and significant relationship between allocative efficiency and farmers’ age, access to technical assistance and extension services. Policy measures aimed at increasing rice farmers’ access to technical assistance and extension services will go a long way toward boosting rice production in the study area.

Original scientific paper. Received 12 Nov 07; revised 03 Feb 10.

Introduction today is determining whether agricultural Rice is the most important staple food for about production under existing technology could be half of the human race (Hawksworth, 1985; Oteng increased without using high capital investment & Sant’Anna, 1999). It ranks third after wheat or developing new technology (Leibenstein, 1978; and maize for indigenous production. Wudiri & Bravo-Ureta & Rieger, 1991; Bravo-Ureta & Fatoba (1992) and Ladebo (1998) estimated that Evenson, 1994; Xu & Jeffery, 1998). It is no rice contributed about 12 to 14 per cent of the surprise, therefore, that considerable effort has food requirement of the Nigerian populace. They been devoted to analysing farm-level efficiency further opined that production capacity of in developing countries. An underlying factor Nigeria’s peasants is well below the national behind these works is that if farmers are not requirement. For over two decades, the nation making efficient use of existing technology, then has relied on importation to meet local demand. efforts designed to improve efficiency would be Rice is important in Nigeria for several reasons, more cost-effective than introducing new and it is a major contributor to internal and technologies to increase agricultural output subregional trade (Longtau, 2003). (Shapiro, 1983). If farmers are efficient in A major issue that agricultural economists, allocating inputs, it leads to minimization of cost other researchers, and policy makers are facing for a given level of output. As a result, they

Ghana Jnl agric. Sci. 43, 17-23 18 D. O. Awotide & A. S. Bamire (2010) Ghana Jnl agric. Sci. 43, 17-23

maximize profit and are encouraged to produce selecting the sample. Six local government areas more; thus, leading to food security, import (LGAs) ( North, , , , substitution and competitiveness in rice Obafemi-Owode and ) were selected production. from the 10 rice-producing LGAs based on the In the theoretical literature, the efficiency of a intensity of rice production, varying from low, firm is defined with two separate concepts, medium to high levels. One LGA (Ikenne) was technical efficiency and allocative efficiency selected from low rice-producing LGAs, while two (Farrell, 1957). Following the classical definition LGAs ( Yewa North and Ifo) and three LGAs of Farrell (1957), a firm is considered to be (, Ewekoro and ) technically efficient if it obtains the maximum were selected from the medium and high rice- attainable output, given the amount of inputs and producing LGAs, respectively. The LGAs the technology used; while allocative efficiency constituted the first stage of sampling. reflects the extent to which firms use the inputs From the list of the rice-growing villages in in optimal proportion, for a given set of input each LGA identified by Ogun State Agricultural prices and a given technology. The two measures Development Programme (OGADEP), of efficiency can be combined into a measure of proportionate random sampling was used to total economic efficiency. select 30 villages. It was necessary because the Measuring efficiency is important because it number of rice-producing villages in each LGA is the first step in a production process that may varied. Cost and time informed the use of 30 lead to substantial resource savings. The resource villages. It constituted the second stage of savings have important implications for policy sampling. formulation and firm management. For individual In each village, rice farmers were identified with farms, gains in efficiency are particularly important the assistance of OGADEP village extension in periods of financial stress. Efficient farms are agents (VEAs). From these, six farmers were more likely to generate higher incomes and, thus, randomly selected in each village as a third stage stand a better chance of surviving and prospering of sampling. The proportionality factor used is (Bravo-Ureta & Rieger, 1991). stated asV = n / N *30 ; where, V is the Although resource use efficiency has been number of villages to be sampled from each LGA, studied in recent times in Nigeria (Amaza & n is the number of rice-producing villages in the Olayemi, 1999; Oredipe & Akinwunmi, 2000; LGA, N is the summation of rice-producing Adewuyi & Okunmadewa, 2001; Amaza, 2001; villages in the six LGAs, and 30 is the desired Ajibefun, Battese & Daramola, 2002; Awotide, number of villages for the survey. In all, 180 2004), research focused explicitly on the farmers were sampled. However, only 165 relationship between allocative efficiency and questionnaires were used for the analysis. Fifteen some socio-economic variables, using single- were rejected for inconsistency and inadequate stage estimation procedure, has been wanting. information. This paper attempts to bridge the gap in knowledge by analyzing the relationship between Empirical model socio-economic characteristics and allocative Stochastic frontier cost function efficiency among upland rice farmers in Nigerian Following Aigner, Lovell & Schmidt (1977) and agriculture with particular reference to Ogun State. Meeusen & Van den Broeck's (1977) method of estimating a stochastic frontier production Materials and methods function in which the disturbance term ( ) is Sampling procedure and sample size composed of two parts, a systematic term (V) and A multi-stage sampling procedure was used in Socio-economic characteristics and allocative efficiency of rice farmers 19

one-sided (U) component, a Cobb-Douglas and ownership characteristics) in an attempt to production function of the following form was identify some reasons for differences in predicted specified: efficiencies between firms in an industry. This has been recognized as a useful exercise, but the Q = g (X ; ß) + 1 a two-stage estimation procedure has also been

where Q is the quantity of agricultural output, Xa recognised as one, which is inconsistent in its is vector of input quantities, and ß is a vector of assumptions regarding the independence of the parameters; inefficiency effects in the two estimation stages (Coelli, 1996; Kyi & von Oppen, 1999; Ajibefun et ∈ = (error term) is defined as: j al., 2002). The two-stage estimation procedure is unlikely to provide estimates that are as efficient ∈ = v + u j = 1, 2,…… n farms 2 j j j as those that could be obtained using a single- If the functional form of the production frontier stage estimation procedure (Coelli, 1996). is self-dual, for example Cobb-Douglas, then the For this study, a Cobb-Douglas cost frontier corresponding cost frontier can be derived specification was estimated: vi is a random noise analytically and written in general form as: σ2 term assumed to be distributed as N(0, v); ui is farm-specific inefficiency effect assumed to follow C = h (K, Q; γ ) + 3 a truncated (at zero) normal distribution with where C is the minimum cost associated with the µ σ2 µ σ2 µ mean and variance {N( i , u)} where i = f production of Q, K is the capital input, and r is a β (Zi, i) and Zi is a vector of farmer-specific factors vector of parameters; β and a constant; is a vector of parameters to be ∈ estimated; and ƒ () is a suitable functional form, j is as defined above. ∈ usually assumed to be linear. The specific j Allocative efficiency of farm j (AE ) is given j equations for the stochastic cost frontier and the by: inefficiency model are presented in Equations 5 AE = exp (+u ) 4 j j and 6. For this study, the specific Cobb-Douglas The efficiency estimates from the cost cost frontier estimated is: function, exp (+U) must be > 1 because U > 0, by In (C /w ) =δ +δ In Q + δ In (K /w ) + (v + u ) 5 construction. AE obtained in Equation 4 were i i 0 1 i 2 i i i I j It is assumed that the inefficiency effects are inverted; that is, AEj = 1/exp (+Uj) so that 0 < AEj independently distributed and Uij, arises by < 1 (Personal communication with Prof. Tim Coelli, truncation (at zero) of the normal distribution with 2004). In this cost function, the non-negative mean uij and variance σ2 , where u is defined by random variable u , which is assumed to account ij j equation: for the cost of inefficiency, defines how the farm µ = β + β Z + β Z + β Z + β Z + β Z + β operates above the cost frontier. If allocative i 0 1 1 2 2 3 3 4 4 5 5 6 Z6 6 efficiency is assumed (Coelli, 1996), the non- where the subscript i indicates the ith farmers in negative random variable u is closely related to j the sample the cost of technical inefficiency. C = cost incurred in rice production (N) Several empirical studies (Pitt & Lee, 1981; Q = output of rice in kilogrammes Awotide, 2004) have used the two-stage K = capital (value of implements used in procedure to estimate stochastic frontiers and rice production in Naira) predict firm-level efficiencies using specified W = labour price (wage rate per day in production or cost functions or both, and then Naira) regressed the predicted efficiencies on firm- δ = parameters to be estimated (i = 0, 1, 2) specific variables (such as managerial experience v = two-sided, normally distributed 20 D. O. Awotide & A. S. Bamire (2010) Ghana Jnl agric. Sci. 43, 17-23

random error Jayasuriya, 1983; Bravo-Ureta & Evenson, 1994). u = one-sided efficiency component with Technical assistance and extension are variables a half-normal distribution that measure the number of times the farmers had µ is = are allocative inefficiency effect contact with research institutions and extension predicted by the model itself field staff respectively. Extension has been shown

Z1 = age of the rice farmer in years to have positive relationship with efficiency

Z2 = the number of years of schooling (Bravo-Ureta & Evenson, 1994; Amaza & Olayemi, completed by the rice farmers 1999).

Z3 = the number of years the farmer has been in rice production Results and discussion

Z4 = equal to 1 for farmers that received Stochastic cost function: results from half-normal credits and zero otherwise model

Z5 = equal to 1 for those farmers that Table 1 presents the maximum likelihood estimates received technical assistance from (MLE) (Greene, 1980) of the parameters of the sources other than the state ADP and stochastic frontier model using the programme zero otherwise FRONTIER 4.1 (Coelli, 1996), which can predict

Z6 = equal to 1 for farmers that reported the variance parameter in terms of sigma square having contacts with the extension and gamma. The estimate of the gamma parameter services from the state ADP and zero was high, showing the value of 0.72 and otherwise significant, which suggests that the inefficiency The βi coefficients are unknown parameters effects are highly significant in analyzing the to be estimated, by the method of maximum output in physical terms for the rice farmers. The likelihood, using the computer programme coefficient of output (0.496) was statistically FRONTIER Version 4.1 (Coelli, 1996). significant at 1 per cent level and had expected positive sign. The coefficient was highly A priori expectations of variables included in significant and had a positive correlation with the frontier model the cost of production. It suggested that farmers Age variable was measured in years. In this whose output were high might have increased study, it is hypothesized that effect of age on gross margins (from sales of output), which may allocative efficiency could be either negative or be ploughed back into production. Farmers with positive. Education measured by years of higher output have better capacity to use schooling of farmers is hypothesized to be improved farm inputs with the associated cost positively related to allocative efficiency as done that are usually higher. The coefficient of capital in previous studies (Bravo-Ureta & Evenson, (0.235) was significant at 1 per cent level and 1994; Xu & Jeffrey, 1998; Seyoun, Battese & positively related to the cost of production of Fleming, 1998; Ogundele & Okoruwa, 2004). rice paddy in the study area. This is in line with a Experience represents the farmer’s experience priori expectation. As the use of improved farm measured by the number of years he or she has implements is desirable for improved output, the been engaged in maize production. Studies have implements usually have associated costs, which shown that farming experience is positively might increase the cost of production. related to efficiency (Parikh, Farman & Shah, 1995; Socio-economic, demographic, farm Seyoun et al., 1998). Credit is variable that environment, and non-physical factors are likely measures farmers’ access to credit facilities. to affect efficiency (Kumbhakar & Bhattacharya, Studies have shown that credit has positive 1992; Ali & Chaudhary, 1990). Using the impact on efficiency (Lingrand, Castillo & inefficiency model in the cost frontier, the study Socio-economic characteristics and allocative efficiency of rice farmers 21

TABLE 1 Stochastic Cost Frontier for Rice Based on a Sample of Ogun State Rice Farmers in 2002

Variable Coefficient T-value

Stochastic frontier Constant 0.332 1.117 Output (Q) 0.496 10.57*** Capital (K) 0.235 3.352*** Inefficiency model Constant 1.111 3.042*** Age -2.010 -2.171*** Level of education -3.135 -1.403 Experience in rice farming 1.871 2.056*** Access to credit 0.445 2.257*** Access to technical assistance -0.477 -1.901*** Access to extension services -0.585 -2.985*** Variance parameters 2 2 2 Sigma square s s = s + s v 0.111 4.506*** 2 2 2 Gamma {¡ = s /(s + s v)} 0.72 2.626*** Log likelihood 33.48

Results from data analysis (2002) ***Significant at 1% captured the determinants of the farmers’ studies have shown that older farmers have lower allocative efficiency. The estimated coefficients probability of adopting new production are of interest and have important implications. technologies, and this might lower their efficiency The negative coefficients for the age variable of production vis-à-vis their productivity. Similar implied that older farmers were allocatively result was recorded by Ajibefun et al. (2002). In efficient than younger ones. This could be this study, experience in rice farming was found explained with the adoption of modern to be significant in determining allocative technology. Older farmers tend to be more efficiency of rice farmers in the study area. conservative and less receptive to modern and Access to credit was used to capture the effect newly introduced agricultural technology; of credit on the efficiency of farmers. The because newly introduced technology comes with availability of credit is expected to loosen the additional cost. Coefficient for age variable was constraints of production, facilitate timely access significant. The variable of education showed to inputs, and increase the efficiency of farmers. negative relation with allocative efficiency though Contrary to expectation, the variable had a not significant. The negative coefficient for positive sign and is statistically significant; education shows that high level of education suggesting that availability of credit hinders results in increased allocative efficiency of rice attainment of higher level of allocative efficiency. farmers. One possible explanation is that access to credit The positive coefficient for rice farming facilities may prompt farmers to overuse experience implies that farmers with more years productive resources. of experience tend to be less efficient. This does There was a negative relation between access not conform to a priori expectation. Adoption to extension and inefficiency effect. This implies 22 D. O. Awotide & A. S. Bamire (2010) Ghana Jnl agric. Sci. 43, 17-23

that access to extension services tends to increase stochastic frontier cost function in application to the allocative efficiency of rice farmers. This is in farm-level data. The theoretical model predicts a line with the general belief that farmers learn from positive relationship among cost of production, the extension services; and if farmers decide to capital, and output of paddy rice. The study follow the advice of extension agents, then it can showed a positive and significant relationship certainly enhance the level of efficiency of farmers. between allocative efficiency and farmers’ age, Similar result was reported by Kyi & von Oppen access to technical assistance and extension (1999). The negative coefficient for access to services. Policy measures aimed at increasing the technical assistance implies that access to it tends managerial ability of the rice farmers will go a long to increase the allocative efficiency of rice farmers. way toward boosting rice production in the study The coefficients for access to technical assistance area. and extension are significant at 1 per cent. The distribution of allocative efficiency of the Acknowledgement rice farmers in Table 2 showed that none of the The authors are grateful to Prof. Tim Coelli of the farmers had an allocative efficiency of less than Center for Efficiency and Productivity Analysis 70 per cent, while 78 per cent of the rice farmers (CEPA), University of Queensland, Australia, for had an allocative efficiency greater than 90 per downloading the FRONTIER 4.1 software, and cent. The results suggested that the sampled for assisting in solving some problems during farmers were fairly allocatively efficient. The mean stochastic frontier analysis. allocative efficiency of 92 per cent suggests that there is the scope for increasing rice production REFERENCES Adewuyi, S. A. & Okunmadewa, F. Y. TABLE 2 (2001) Economic efficiency of crop farmers in Kwara State, Nigeria. Frequency Distribution of Allocative Efficiency (AE) Estimates Nigerian Agricultural Development for Rice Based on a Sample of Ogun State Rice Farmers in 2002 Studies 2, 45-57. Aigner, D. J., Lovell, C. A. K. & Schmidt, AE level No. of farmers Percentage P. J. (1977) Formulation and estimation 90-100 129 78.2 of stochastic frontier production 81-90 27 16.3 function models. J. Econometrics 6, 21- 70-80 9 5.5 37. Total 165 100 Ajibefun, I. A., Battese, G. E. & Daramola, A. G. (2002) Determinants Mean (%) 91.61 of technical efficiency in smallholder Minimum (%) 70 food crop farming: Application of Maximum (%) 98 stochastic frontier production function. Skewness -1.592 Quarterly Journal of International Computed from results from data analysis (2002) Agriculture 41 (3), 225-240. Ali, M. & Chaudhary, M. A. (1990) Inter- in the study area by 8 per cent if they would regional farm efficiency in Pakistains operate at the frontier. Punjab: A frontier production function study. J. Agric. Econs. 41, 62-74. Conclusion Amaza, P. (2001) Resource use efficiency The paper presented the empirical relationship of food crop farmers in Gombe State, Nigeria (PhD Thesis). Dept. of between allocative efficiency and some socio- Agricultural Economics, University of economic variables of upland rice farmers in Ogun Ibadan, Nigeria. State, Southwest Nigeria, using a Cobb-Douglas Amaza, P. & Olayemi, J. K. (1999) An Socio-economic characteristics and allocative efficiency of rice farmers 23

investigation of production efficiency in food crop Luzon, the Philippines. J. Agric. Econ. 34, 163- enterprises in Gombe State, Nigeria. Journal of 173. Rural Economics and Development 13 (2), 111- Longtau, S. R. (2003) Multi-agency partnerships in 122. West African agriculture: A review and description Awotide, D. O. (2004) Resource use efficiency and of rice production systems in Nigeria. Ecosystems inputs substitution in upland rice production in Development Organisation (EDO) Report. Ogun State, Nigeria (unpublished PhD Thesis). Meeusen, W. & Van den Broeck, J. (1977) Department of Agricultural Economics, University Efficiency estimation from Cobb-Douglas of Ibadan, Nigeria. production functions with composed error. Int. Bravo-Ureta, B. E. & Rieger, L. (1991) Dairy farm Econ. Rev. 18, 435-444. efficiency measurement using statistical frontiers Ogundele, F. O. & Okoruwa, V. O. (2004) A and neoclassical duality. Am. J. Agric. Econ. 73, comparative analysis of technical efficiency 421-428. between traditional and improved rice variety Bravo-Ureta, B. E. & Evenson, R. E. (1994) farmers in Nigeria. African Journal of Economic Efficiency in agricultural production: The case of Policy 11(1), 91-108. peasant farmers in eastern Paraguay. Agric. Econ. Oredipe, A. A. & Akinwunmi, J. A. (2000) Resource 10, 27-37. productivity and efficiency among farmers adopting Coelli, T. (1996) A guide to FRONTIER version 4.1: improved maize technology in Ogun State. Nigerian A computer program for stochastic frontier Agricultural Development Studies 1(2), 27-38. production and cost function estimation. Centre for Oteng, J. W. & Sant’Anna, R. (1999) Rice production Efficiency and Productivity Analysis (CEPA) in Africa: Current situation and issues. International Working Paper 96/07. Rice Commission Newsletter FAO 1999 48. Farrell, M. J. (1957) The measurement of production Parikh, A., Farman, A. & Shah, K. (1995) efficiency. Journal of Royal Statistic Society Series Measurement of economic efficiency in Pakistan 120, 253-281. agriculture. American Journal of Agricultural Greene, W. H. (1980) Maximum likelihood estimation Economics 77, 675-685. of econometric frontier functions. Journal of Pitt, M. M. & Lee, L. F. (1981) Measurement and Econometrics 13, 27-56. sources of technical inefficiency in the Indonesian Hawksworth, D. L. (1985) Forward. In Rice diseases weaving industry. Journal of Development (ed. S. H. Ou). CMI Slough, UK, CAB. 380 pp. Economics 9, 43-64. Kumbhakar, S. C. & Bhattacharya, A. (1992) Price Seyoun, E. T., Battese, G. E. & Fleming, E. M. distortion and resource use efficiency in Indian (1998) Technical efficiency and productivity of agriculture: A restricted profit function approach. maize producers in eastern Ethiopia: A study of Review of Economic and Statistics 74, 231-239. farmers within and outside the Sasakawa-Global Kyi, T. & von Oppen, M. (1999) An economic analysis 2000 project. Agricultural Economics 19, 341-348. of technical efficiency of rice farmers at Delta region Shapiro, K. H. (1983) Efficiency differentiates in in Myanmar. Stuttgart Grauer Verlag, Research in peasant agriculture and their application of Development Economics and Policy. Discussion development policies. J. Dev. Stud.19, 179-190. Paper No. 06/99. Wudiri, B. B. & Fatoba, I. O. (1992) Cereals in the Ladebo, O. J. (1998) Determinants of adoption of food economy of Nigeria. In Proceedings of the new technology among rice farmers in Ifo local Workshop on Recent Development in Cereal , Nigeria. ACTA Production in Nigeria (ed. S. M. Lawani and T. Universities Agriculturae et Silviculturae Babaleye) held at Dubar Hotel, Kaduna, 2-4 Mendelianae Brumensis 48. September 1991. Ibadan, IITA. Leibenstein, H. (1978) X- Inefficiency Xists – Reply Xu, X. & Jeffrey, S. R. (1998) Efficiency and to an Xorcist . Amer. Econ. Rev. 68, 203-211. technical progress in traditional and modern Lingrand, J., Castillo, L. & Jayasuriya, S. (1983) agriculture: Evidence from rice production in China. Comparative efficiency of rice farmers in central Agric. Econs. 18, 157-165.