African Journal of Science, Technology, Innovation and Development ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rajs20 Efficiency and productivity analysis of pineapple farmers in the Akwapim-South District of Ghana: A distance function approach Yaw Ofori-Appiah, Edward Ebo Onumah & Freda Elikplim Asem To cite this article: Yaw Ofori-Appiah, Edward Ebo Onumah & Freda Elikplim Asem (2021): Efficiency and productivity analysis of pineapple farmers in the Akwapim-South District of Ghana: A distance function approach, African Journal of Science, Technology, Innovation and Development, DOI: 10.1080/20421338.2020.1857544 To link to this article: https://doi.org/10.1080/20421338.2020.1857544 Published online: 03 Feb 2021. Submit your article to this journal Article views: 76 View related articles View Crossmark data Citing articles: 1 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=rajs20 African Journal of Science, Technology, Innovation and Development, 2021 https://doi.org/10.1080/20421338.2020.1857544 © 2021 African Journal of Science, Technology, Innovation and Development Efficiency and productivity analysis of pineapple farmers in the Akwapim-South District of Ghana: A distance function approach Yaw Ofori-Appiah *, Edward Ebo Onumah and Freda Elikplim Asem Department of Agricultural Economics and Agribusiness, University of Ghana, Accra, Ghana *Corresponding author email: yoforiappiah@gmail.com; traviochenko@yahoo.com The study sought to analyze the efficiency and productivity of pineapple farmers in the Akwapim-South District of Ghana. The paper employs the stochastic translog distance function technique to analyze the influence of inputs on outputs whilst highlighting the technical efficiency and its determinants of the pineapple farmers. A total of 135 respondents were randomly selected using the multi-stage sampling technique. The results reveal that all the input variables (fertilizer, labour, cost of pineapple suckers and intermediate cost) are significant and have a positive influence on the productivity of pineapple production. The shadow share of other crops is estimated to be negative relative to the production of pineapple in the output mix. The mean technical efficiency score is found to be 85%. This implies that pineapple farmers in the Akwapim-South District have the potential to increase their output level by 15% using the current technology available to them. The results further indicate that age, number of extension visits, education and FBO (farmer-based organization) membership significantly and positively affect the technical efficiency of pineapple farmers. The paper recommends that government policies should ensure inputs are available to farmers at subsidized prices. Furthermore, regular education of extension agents and properly structured extension and FBO service delivery systems should be put in place to enhance the productivity of farmers. Keywords: stochastic model, output mix, shadow share, farmer-based organization Introduction (Gatune et al. 2013). Many producers and shippers were The development of most economies is based on three greatly affected by this change which virtually collapsed major sectors, namely the agricultural, service, and indus- the export of pineapple products. Farmers who success- trial sectors. In Ghana, agriculture continues to be a domi- fully managed to switch to the new variety were faced nant force with a contribution share to GDP of about with initial difficulties including lack of planting materials 18.3% (GSS 2018); hence, agriculture plays an essential and agronomic practices which led to low exportable role in the overall growth of the economy. Ghana has a yields with a high investment cost. The worst casualties comparative advantage in the production of pineapple as a result of this structural change was that smallholder which positioned the country to become a paramount pro- farmers who were not financially stable to invest in ducer and supplier of economical pineapple products to MD2 production lost their source of income (Jaeger EU markets (Jaeger 2008). The fruit industry through its 2008; Kleemann 2011). During the peak of this critical exportation to the EU yielded about €562 million to the period, the favourable comparative advantage that economy of Ghana between 2000–2013 (Eurostat 2013). Ghana had on the global market was no longer vibrant Pineapple production received support under the Econ- enough to boost the sector’s competitive nature as well omic Recovery Programme (ERP) in the 1990s and has as to compete in the European markets with other export- since contributed significantly to the development of ing countries, especially Costa Rica. Ghana’s revenue. The aim of the programme was to diver- In Ghana, smallholder farms dominate the agricultural sify export products in order to lessen Ghana’s depen- sector with a farm sizes of about 1.2 hectares and farming dence on its few primary commodities. activities characterized by the low use of improved tech- The fruit and vegetable contribution to Ghana’s export nology. Most crops have low yields with pineapple pro- rose significantly from the early 1990s and reached a peak duction yielding as low as 60% of its potential yield of more than US$30 million in the mid-2000s. During this from 2002 to 2008. The low productivity (an average period, the pineapple sector employed small-scale farmers yield of 60 Mt/Ha out of a potential yield of 100 Mt/Ha) and workers from the planting stage through to harvesting, can be attributed to a lack of improved planting materials collection, processing and exporting. The successful nature and minimal use of inputs, especially fertilizer, due to the of the pineapple sector during this period created employ- high cost (MoFA 2010; ISSER (2017)). Pineapple farmers ment and led to poverty reduction (Gatune et al. 2013). in Ghana are often not able to meet the high global However, Ghana experienced a series of crises start- demand for their produce compared with their counter- ing from the 2005 production year that has contributed parts from other African countries and Costa Rica. The to the decline in volume of pineapple in the export indus- failure or inability of pineapple farmers to achieve their try. There was a major structural transformation of the desired levels of output can be attributed to diverse and pineapple industry in Ghana due to the change in varietal varied factors which greatly affects their productivity preference for MD2 as compared to the Smooth Cayenne and efficiency levels. A multi-output feature is observed Supplemental data for this article can be accessed here https://doi.org/10.1080/20421338.2020.1857544 African Journal of Science, Technology, Innovation and Development is co-published by NISC Pty (Ltd) and Informa Limited (trading as Taylor & Francis Group) 2 Ofori-Appiah, Onumah and Asem in agricultural production as a risk minimization strategy. vital issue from the view of agricultural development. Pineapple farmers grow other crops such as maize, yam, Measurement provides the necessary information for the cassava as well as other vegetables and this can be formulation of agricultural policies and making of good observed to a large extent in the Akwapim-South District. management decisions in the allocation of resources. The multi-output nature of farming carried out by pineap- Hence, it is significant that in an effort to improve the pro- ple farmers is also practised in other African countries ductivity of pineapple production in Ghana, a more prac- such as Rwanda (Kayitesi 2011). Pineapples that are tical and realistic method is used to establish and measure usually cultivated for export or fruit processing compa- the relative efficiencies of resource use among pineapple nies are not intercropped with other crops and this is farmers. This will help to ascertain if some gaps in pro- because the farmers have to meet strict phytosanitary cer- duction still exist that need to be addressed. The conven- tification standards such as the GLOBAL G.A.P. tional method of efficiency measurement used to estimate requirement. efficiency of farmers will not be the true reflection of the Many smallholder farmers in developing countries are multi-output nature of production. This study therefore making pragmatic efforts to increase their productivity seeks to provide empirical information on the multi- and augment their incomes. Even though efforts have output production efficiency of pineapple farmers in the been made to help pineapple farmers boost production Akwapim-South district of Ghana using the stochastic by helping them to improve their efficiency, the extent output distance function approach. of improvement is not well captured in the literature. This study assessed the productivity and technical Pineapple is an important crop in both the local and inter- efficiency levels of the farmers using the stochastic dis- national markets since it generates high levels of revenue tance function technique. By identifying and analyzing and foreign exchange for the country. Issues relating to the factors that influence the efficiency of pineapple pro- pineapple production need to be addressed to achieve duction in the district, the findings of the study could economic growth. The right estimation will give the assist policy-makers in designing effective future inter- right technique to tackle issues that this study seeks to vention programmes to help the sector to exploit its full address. Past studies investigated technical efficiency of economic potential. Following the introduction, the rest pineapple production in a single output orientation using of the paper is organized as follows: The next section pro- a stochastic frontier approach instead of a multi-output vides the theoretical concepts underpinning the empirical context which is common in pineapple production. The estimation model. The section thereafter gives a brief stochastic frontier approach has been criticized for its overview of the study area and the data description. inability to estimate efficiency when the output involved The penultimate section presents the results and dis- exceeds one. The problem of estimating multi-output cussion, while the last section concludes with recommen- can be solved by using the distance function approach. dations for policy interventions. Another benefit of the distance function approach is that it permits the usage of multiple inputs-multiple outputs Materials and methods technology without the availability of prices information Conceptual framework (Coelli and Perelman 2000). The conceptual framework of the production efficiency of In agricultural production in developing countries, the pineapple production together with the production of other measurement of production efficiency has always been a crops is presented in Figure 1. Agricultural production Figure 1: A conceptual framework. Source: Adapted from Kayitesi (2011) African Journal of Science, Technology, Innovation and Development 3 involves the conversion of inputs such as labour, pineap- Thijssen 2002). Using the multi-output-multi-input esti- ple suckers (planting materials), fertilizer and intermediate mating concept in the context of Ghanaian farming may costs into outputs. Efficient production is associated with be a better choice and this can be solved by using the good management which entails the most effective combi- output distance function approach. nation of inputs at the right time. The production process The production frontier is linked with the highest is influenced by inputs and external factors such as con- attainable level of output, given the least level of inputs straints in production. Land is expected to increase that is needed to produce a certain amount of output. In output when brought under production. The defined tech- other words, it is the venue or location for maximum nology for this study comprises the use of pineapple achievable output for each input mix. As noted by suckers, fertilizer and intermediate cost to which labour Farrell (1957) and Debreu (1951), a production unit is is applied to yield a certain level of output. When these considered efficient only if it is found on the production factors are combined under a particular production frontier. Technical inefficiency is however accredited process given a particular technology, an actual output is with the inability of the firm to produce the frontier achieved. Since the typical farmer generally produces level of output, given the amount of inputs available more than one crop, these outputs have to be shared (Kumbhakar, 1994). among the different crop types that the farmer is cultivat- ing. Differences in the use of inputs bring about different Stochastic function approach levels of efficiency. Usually, there exist a gap in agricul- The stochastic frontier was introduced by both Aigner, tural production; hence, most often actual output is Lovell, and Schmidt (1977) and Meeusen and Van den unable to meet potential output. This gap can be explained Broeck (1977). They specify clearly that external noise by the inefficiencies on the part of farmers and other risk affects production by introducing errors, thus making it factors such as weather, diseases and so on. Inputs are necessary to isolate the influence of these exogenous expected to improve the production process in order to errors events from the errors that cause technical effi- obtain optimum output given a particular technology. ciency. They suggested a function which incorporates an However, lack of these inputs and improper usage of error term introduced by external noise and technical inputs brings about a gap in output levels. The gap in pro- inefficiency. duction can also be explained by certain socio-economic The stochastic approach employs various functional inefficiencies such as experience and access to credit. If forms in establishing the relation between inputs and these determinants are lacking, they introduce some inef- outputs in the analysis of technical efficiency. The most ficiencies in the production process and cause output gaps common ones used are the Cobb Douglas and Translog in the long run. There are other factors that bring about production function. gaps in production which may be (1) within the control Various studies conducted on efficiency have of the farmer, including: farming activities such as fertili- employed the translog stochastic frontier production func- zer application and method of land preparation; (2) factors tion specification. The specification is mostly used in pro- beyond the control of the farmer, including: the effects of duction analysis. This specification does not assume statistical noise present in the model, bad weather, nature homogeneity or separability. An added advantage is that of soil, etc. These factors also prevent farmers from the model does not impose any restrictions on the elas- obtaining their potential output. This study emphasizes ticity of substitution on the input variables in the function. only the inefficiencies on the part of farmers and not the Another advantage is that it allows for the usage of many risk aspect. input variables in the function (Berndt and Christensen 1973). This functional form is flexible and does not Theoretical concept pose any significant parameter restrictions on inputs The translog output distance function was applied in this present in the function. The functional form has some dis- paper. In almost all the farming communities in Ghana, a advantages, such as the presence of multicollinearity multiple-output production feature is usually a popular between the input variables (Abdulai and Huffman characteristic of farming activities which is largely influ- 2000). It also requires larger sample sizes. enced by the farmers’ approach to becoming self-suffi- cient and minimizing risk simultaneously (Mensah and Brümmer 2016). As mentioned earlier, the typical Gha- The output distance function naian farmer generally produces more than one crop The output distance function evaluates the proximity of a with one dominant crop and other crops either on subsis- specific output level to the maximum achievable level of tence level or to buttress the dominant crop. Several output using the same level of inputs under a technically studies have been conducted using the conventional tech- efficient production (Mawson, Carlaw, and McClellan nical efficiency approach which uses a single output to 2003). The definition of an output distance function measure the performance of farmers. However, the use starts by establishing the type of production technology of the conventional technical efficiency framework to used by the firm to produce the output, P(x), whichM measure how farmers perform under circumstances in shows the set of all output vectors, y [ R+ , that can be K which they produce more than a single output may tend produced using the input vector, y [ R+. This can be to bias the estimates that would be obtained from the expressed mathematically as: regression, thus ignoring the allocative effect in relation M to the inputs and outputs (Brümmer, Glauben, and P(x) = {y [ R+ : x can produce y} (1) 4 Ofori-Appiah, Onumah and Asem The output distance function is defined as: Translog output distance function Estimating the distance from the frontier (distance func- Do(x, y) = min {u:(y/u [ P(x)} (2) tion in a parametric setting) requires the need to determine the frontier as well as the relationship between inputs and where Do(x, y) = output distance function. It is non- outputs. There is also the need to identify some form of decreasing, positively linearly homogeneous and convex multi-output production function P(x). The functional in y, and decreasing in x (Lovell et al. 1994); u = level form that is commonly used is assumed to be the translog of efficiency (it shows the extent to which the output production function because the incorporation of its vector must be expanded to get to the efficient frontier); squared and cross-product (interactive) terms introduces P(x) = the set of vector (y) that can be obtained from a high degree of flexibility. Also, it does not impose the input vector; y = outputs that are in the output distance restrictions concerning substitutability between inputs or function; x = input vector. outputs. The translog distance function with M (m= 1, The conceptualization of a multi-output–multi-input 2,… , M ) outputs and K (k = 1, 2,… , K) inputs, and production technology for a cross-sectional set of data for I (i = 1, 2,… , I ) farmers can be specified as: can be depicted in Figure 1 where a farmer produces two outputs (y1, y2) from an input set of vector (x) and ∑M 1∑M ∑M y* represents the production possibility frontier (PPF). lnD0i =a 0 + am ln ymi + amn ln ymi ln y2 ni According to the production theory, the PPF curve m=1 m=1 n=1 shows all the possible combinations of technically effi- ∑K 1∑K ∑K cient production points of the outputs (y , y ) that could + bk ln xki + bkl ln x1 2 2 ki ln xli be generated with the input vector (x) and still be found k=1 k=1 l=1 in the feasible production region P(x) which is bounded ∑K ∑M by the PPF. Based on principles of stochastic frontier pro- + dkm ln xki ln ymi duction theory, we consider an individual with frontier k=1 m=1 y1*y2*. The individual can be anywhere on or below (3) this frontier. Any point below frontier y1*y2* represents a suboptimal or an inefficient point. From Figure 2, According to O’Donnel and Coelli (2005) and Coelli and point A is considered inefficient while point B (or C) is Perelman (2000), the parameters of the distance function considered to be efficient. The distance that point A is in Equation (3) must in theory satisfy linear homogeneity away from point B (which lies on the frontier) represents with regard to output and regularity conditions, thus their inefficiency level in the production process. Coelli monotonicity and curvature. et al. (2005) note that a proportional expansion of the When specifying an output distance function, there is output A towards point B can be achieved by an a need to impose restrictions on the translog output dis- upward scaling by a scalar (θ) which must be minimized. tance function. This is to ensure linear homogeneity in outputs and the required restrictions are as follows: = 0A ≤ ≤ Homogeneity of degree +1 in outputs is expressed as:Do(x, y) 1, thus Do(x, y) 10B ∑M ∑M = 1 ≥ am = 1 and amn = 1 m = 1, 2, . . . , M (3.1)Do(x, y) , thus TETE o 1 m=1 n=1o ∑M d^km = 1 k = 1, 2, . . . ., K (3.2) m=1 Symmetry is then imposed as: amn = anm; m, n = 1, 2, . . . , M and (3.3) bkl = blk; k, l = 1, 2, . . . , K. (3.4) Evidently, the output distance function has outputs which are generally considered as part of the dependent variables and, above all, the output distance function has an unob- served (unknown) value. In addressing these two pro- blems, the output distance function is normalized by dividing all outputs by a reference output (example ym). The negative of the natural logarithm of Doi can be added to independent variables in the model, which then Figure 2: A diagram showing output distance function with two becomes our inefficiency term. Finally, in estimating the outputs (y , y ). stochastic frontier, the error term vi, which has a normal1 2 Source: Adapted from Mensah and Brümmer (2016) distribution, is added. African Journal of Science, Technology, Innovation and Development 5 Lovell et al. (1994) and Coelli, Perelman, and Empirical model specification Romano (1999) explained that one can impose the According to Coelli and Perelman (1996), in selecting a linear homogeneity property by normalizing the output functional form for an empirical parametric study, the fun- distance function by one of the outputs. It is observed damental decision to make is the choice of a functional = 1 form. The functional form for the distance functionthat homogeneity implies that setting u and substi- ym should preferably be flexible and simple to compute, as tuting it in Equation (2) infers the distance function can well as allow for the imposition of homogeneity. Since be obtained as seen in Equation (4): the translog functional form meets these three require- ments, it has been used by several authors in the distance D (x, y/y ) = D (x, y)/y (4) function estimation (Grosskopf et al. 1997; Lovell et al.o M o M 1994; Mensah and Brümmer 2016). A translog stochastic output distance function with Hence, the translog expression for the ith farmer two outputs, (y1, y2), and four inputs, x = (x1, x2, x3, x4), would be expressed in Equation (4.1) as: was defined as: ln (Doi(x, y)/yMi) = TL(xi, yi/yMi, a, b, d), −ln (y1i) =a0 + a1 ln (y2i/y1i) = (4.1)i 1, 2, . . . , N ∑K=4 + 0.5a11 ln (y2i/y1i) ln (y2i/y1i)+ bk ln xki k=1 The translog expression would be expressed fully in Equation (5) below: ∑K=4∑K=4+ 0.5 bkl ln xki ln xli k=1 l=1 M∑−1 TL(xi, yi/yMi, a, b, d) = a0 + am ln (y K=4mi/y ∑Mi) = + 0.5 dk ln xki ln (y2i/y1i)+ vi + um 1 i k=1 1M∑−1M∑−1+ (6)amn ln (ymi/yMi) ln (y2 ni/yMi) m=1 n=1 ∑ ∑∑ where y1i represents the value of pineapple produced inK K K + 1b ln x + b ln x ln x Ghana cedis (GH¢) per hectare by the i-th farmer andk ki 2 kl ki li k=1 k=1 l=1 y2i is the normalized output, which is equal to the ∑K M∑−1 ‘output ratio’ of the value of other crops (the value of + 1 dkm ln xki ln (ymi/yMi) (5) cassava + maize + plantain + yam + tomatoes + pepper) in2 k=1 m=1 relation to the value of pineapple produced by the i-th farmer. By rearranging the terms in Equation (4), the function The input factors (xi) specified in model 6 are: (Equation 4.1) can then be rewritten as: x1 = fertilizer (GH¢) x2 = labour (man-days) − = − x3 = cost of pineapple suckers (planting materials) (GH¢)ln (yMi) TL(xi,yi/yMi, a, b, d) lnDoi(x, y), (5.1) x4 = intermediate cost (GH¢) i = 1, 2, . . . , N α, β and δ = are the unknown parameters that are to be estimated. where –lnD0i(x, y) is equivalent to the radial distance Description of variables in the translog distance function from the boundary. Hence, by setting –lnD0i- function (x, y) = u and in addition capturing the noise effect by The dependent variable is the total pineapple output adding the term vi, a Battese and Coelli (1988) kind of which is measured in value terms (GH¢/ha). The indepen- the conventional stochastic frontier model is obtained as dent variables include the value of other crops (GH¢/ha), proposed by Aigner et al. (1977) and Meeusen and Van fertilizer (GH¢/ha), labour (man-days/ha), cost of pineap- den Broeck (1977) and specified as: ple suckers (GH¢/ha) and intermediate cost (GH¢/ha). The shadow share, which is the contribution of pine- − ln (y ) = TL(x , y y , a, b, d)+ (5 2) apple in the overall output, can be obtained by imposingMi i i/ Mi 1i . homogeneity constraints. This is defined in Equation (7) as: 1i = vi + ui (5.2.1) ∑ ∑ P = 1– b (7) where εi = is the composed error term and –LnD0i(x, y)= u represents the inefficiency (one-sided inefficiency term) where, ∑P= the overall contribution of pineapple and of the farmer and it follows some probability distribution. ∑β = the coefficient of other crops. The error term ui is a non-negative random term; vi= it The input variables were estimated on a per hectare depicts the events or occurrences that are outside the basis. They were standardized by dividing through by farmer’s control (random noise). their respective farm sizes. The standardized variables 6 Ofori-Appiah, Onumah and Asem were then normalized by dividing through by their was used to estimate the one-step model which clearly respective means. This implies that after taking the logs, identifies both the stochastic frontier and technical ineffi- the first-order distance elasticities can be interpreted as ciency model. partial elasticities in relation to the inputs. x1 = represents the cost of fertilizer and it comprises Study area and dataset both liquid and solid fertilizers. It was measured in The research was conducted in the Akwapim-South Dis- Ghana cedis (GH¢) per hectare. The cost of fertilizers trict of Ghana. The Akwapim-South District is located that were incurred during the production season was com- in the south eastern-part of the Eastern Region of puted and standardized by dividing through by its farm Ghana. The district covers communities such as size. Pokrom, Amanfrom, Yaw Duodo and Berekuso, among x2 = denotes the labour component of the input vari- others, with Aburi as the district capital. Of the total able and it was measured in man-days and it was then number of households in the district, 4475 – representing standardized by its farm size. Man-days were computed 48.2% – engage in agriculture (GSS 2014). The majority according to the rule that one adult male, one adult of the households engaged in agricultural activities in the female and one child (≤ 18 years) working for one day district (94.5%) are involved crop farming. The dominant (8 h) was equal to 1; 0.75 and 0.50 man-days respectively. crop cultivated in the area is pineapple (MoFA 2015). It is These ratios were adopted from Coelli and Battese evident that a significant proportion of the population in (1996); Battese, Malik, and Gill (1996); Onumah the district engages in some form of farming with pineap- et al.(2018). ple farming being predominant. The communities that x3 = depicts the intermediate cost component. This were covered in the study were Pokrom, Nsakyi, Aman- refers to the cost of other items and services that were from, Oboadaka, Pepawani, Apatem, Kwasi Doi, Otiak- involved in the pineapple production process. Its unit of rom and Yaw Duodu with at least ten (10) respondents measurement was in Ghana cedis (GH¢) per hectare. from each community, making a total sample size of The intermediate cost comprises items such as the cost 135. The selection of Akwapim-South District as the of cutlasses, hoes, knapsack sprays, watering-cans, hired study area was appropriate for this study. items, transportation costs, among others of the i-th A multi-stage sampling technique was employed to farm during the production year. obtain a total of 135 respondents for this study. Pineapple x4 = depicts the cost of the pineapple suckers, farmers were the target population for the research. A list measured in Ghana cedis (GH¢). The various costs of of farmers was obtained from the extension officers in the pineapple suckers were then divided by their respective district. For the first stage, pineapple farmers were purpo- farm sizes to get the unit of measurement as Ghana sively selected from the list. This was followed by cluster cedis (GH¢) per hectare. sampling to group the pineapple farmers according to The summation of the distance elasticities of the input their various communities. This was done carefully to variables gives an indication of the returns to scale of the ensure that farmers from all communities in the district production function. This shows how output changes with were represented. The final stage was the random respect to input use. The returns to scale can be constant, sampling of 135 respondents (Figure 3). increasing or decreasing. In investigating how the socioeconomic character- istics of pineapple farmers influence their production, a Summary statistics heteroskedastic corrected inefficiency model was used, Table 1 presents characteristics of the sample data such as as suggested by Wang and Schmidt (2002). It is expressed household size, farming experience, among others. Before in Equation (8) below: estimating the model, outputs were aggregated into two categories (i.e. value of pineapple and other crops pro- s = exp {z d } (8) duced). The dataset contains information on the quantityui ij j of pineapple and other crops produced and sold, as well s = exp {v+ Z + a Z + a Z as the farm gate price received by farmers from exporters,ui a1 age 2 gender 3 extenV fruit processing companies and local traders. The other + a4Zexp + a5Zfbo + a6Zeduc + a7Zhsize crops cultivated by the farmers include cassava, maize, + a Z } (8.1) plantain, yam, tomatoes and pepper. In the study area,8 accred the minimum crop diversification was two and the where sui represents the variance of the one-sided error maximum crop diversification was five. term (ui), Zage = age of farmer, Zgender = gender of farmer, ZextenV= number of extension visits, Zexp = years Results and discussion of farming experience, Zfbo = FBO membership, Zeduc = Distance elasticities level of education, Zhsize = household size, and Zaccred = The results from the translog distance function model are access to credit. given in Table 2. A negative estimate for the distance elas- The parameter estimates of the output oriented techni- ticities implies a positive contribution to the production of cal efficiency and the determinants of its inefficiency were pineapple in the district. All the input variables were derived by jointly estimating Equations (5) and (7) using found to be significantly different from zero (0); labour, the MLE method with Stata Version 14 Statistical Soft- cost of pineapple suckers and intermediate costs were sig- ware. Thus, the maximum likelihood estimation technique nificant at 1% while fertilizer was significant at 5%. This African Journal of Science, Technology, Innovation and Development 7 Figure 3: Map of Akwapim-South District (the study area). Source: Geography Department (University of Ghana), 2020 means that all the input variables contributed significantly other crops using the specified inputs reduces the to the production of pineapple in the district. shadow share value of pineapple output by 7.2%, and As presented in Table 2, the marginal productivity of vice versa. The relatively low estimate of other crops’ labour was the largest and it shows the importance of this output elasticities reflects the low shadow share of other factor input in the production of pineapple in the district crops in the multi-output production setup. The majority and agricultural production as a whole. A positive result of farmers in our data cultivate pineapple as a cash crop for the other crops’ distance elasticity implies a negative mainly for export and fruit processing companies and shadow share contribution of other crops relative to the therefore its large shadow share to total revenue with pineapple output in the overall production, hence reflect- respect to other crops is consistent with the data and our ing a degree of substitution. Using the homogeneity expectations. The pineapple farmers who produce restriction, the share of pineapple in the total farm pro- mainly for export have to meet EUREPGAP G.A.P. and duction is computed as 0.928 (92.8%) while that of GLOBAL G.A.P. certification standards and therefore other crops is 0.072 (7.2%). These estimates suggest have to adhere strictly to phytosanitary regulations that a percentage increase in the production share of (Ninson 2012). Also, fruit processing companies expect Table 1: Characteristics of the sample data (number of observation = 135). Variable Unit of measurement Mean Std. Dev. Pineapple GH¢/ha 10,209.22 319.36 Other crops GH¢/ha 1,668.68 47.10 Fertilizer GH¢/ha 627.96 10.82 Labour Man-days/ha 17.65 0.34 Intermediate cost GH¢/ha 1,222.87 20.12 Cost of pineapple suckers GH¢/ha 266.57 9.98 Education Years 7.36 0.25 Experience Years 14.01 0.56 Household size Number of people in a household 5.80 0.19 Age Years 48.11 0.65 Extension Number of extension visits 3.96 0.30 Source: Field survey 8 Ofori-Appiah, Onumah and Asem Table 2: Average distance elasticities and returns to scale. Variables Coefficient Pineapple 0.928 Other crops 0.072*** Fertilizer −0.188*** Labour −0.325*** Cost of pineapple suckers −0.291*** Intermediate costs −0.296*** RTS 1.10 Source: Field survey farmers who supply their produce to them to ensure good farms in Ghana using a stochastic distance function agricultural practices (GAP); hence, these selected approach was also 85%. farmers do not intercrop the pineapple with other crops The performance score of the farmers is distributed but rather grow them separately on a small scale. This over a range of a minimum of 0.32 to a maximum of finding is in line with the study of Mensah and 0.99 as seen in Table 3. This conforms to the study Brümmer (2016) where banana which is largely cultivated done by Kusa (2012) on agricultural productivity. This for export had a larger shadow share to total revenue with wide variation in technical efficiency scores indicates respect to other crops. This finding is however contradic- the presence of varying levels of resource utilization tory to the study of Ogundari and Brümmer (2011) who among the pineapple farmers. The district therefore obtained a larger shadow share for other crops relative shows an unequal distribution of pineapple farmers to cassava. This may be because the cultivation of across the technical efficiency score. Nevertheless, most cassava permits intercropping which could be the expla- scores are close to the mean efficiency score. The distri- nation for the larger shadow share of other crops. bution is moving to the right direction indicating that The result of the RTS is also presented in Table 2 and the graph is positively skewed. This implies that the was estimated to be 1.10 in absolute terms which is signifi- majority of the farmers have high efficiency scores. The cantly different from zero. This implies that at the sample high efficiency scores result from farmers receiving mean, the production system exhibited increasing returns assistance in the form of education and training from to scale. The result can be economically interpreted as 1% the fruit processing companies and other NGOs. joint increase of the inputs increases pineapple output by about 1.1%. This result is comparable to the findings of Determinants of efficiencies Idiong (2007) where returns to scale was about 1.57%. Results of the determinants of efficiencies among pineap- An increasing RTS of 1.10 implies that pineapple ple farmers in the Akwapim-South District are presented farmers in the Akwapim-South District are producing at in Table 4. the first stage of the production function and as such Age, number of extension visits, level of education output can be increased in the short term if inputs are and FBO membership have significantly positive effects increased. It is uneconomical for a farmer to settle at the on production efficiency. first stage of the production curve and therefore doubling Some efficiency variables, including gender, farming the amount of inputs that are used in the production experience, household size, and access to credit, showed process, will lead to more than double increase in a weak correlation and are statistically insignificant. output. Thus, farmers should ensure good agronomic This implies that there is no relationship between these practices to increase their total farm output. This will variables and technical efficiency and hence these vari- also result in a reduction of the average cost of production ables do not affect the technical efficiency of farmers in per unit of output to take advantage of economies of scale. the study area. The findings in Table 3 show that the coefficients for age is negative and this implies that technical efficiency Technical efficiency increases with increase in age. In other words, the effi- The results from Table 3 reveal that the farmers produce, ciency of the farmers increases with an increase in age. on average, 85% of the potential output given the current This empirical result agrees with the observation by Essil- technology available to them. This implies that an average fie, Asiamah, and Nimoh (2011) that older farmers have increase of 15% in output margins could be possible if more experience. Studies have shown that older farmers production inefficiency is eliminated. The efficiency seem to be more efficient than younger farmers, because score for the study conducted by Mensah and Brümmer of the good managerial skills they have acquired over (2016) on efficiency analysis of commercial banana the years (Binam et al. 2004). The study adds that Table 3: Technical efficiency score. Observations Mean Standard deviation Minimum Maximum 135 0.846 0.1482 0.3187 0.9943 Source: Field survey African Journal of Science, Technology, Innovation and Development 9 Table 4: Determinants of efficiencies. Conclusions This paper used a distance function approach to analyze Variables Coefficient P-value and explore factors driving production efficiency in this Age −0.064* 0.083 Gender 0.474 0.425 district. Farmers in our survey produced a mixture of Extension visits −0.297*** 0.003 crops (i.e. production diversification). Therefore, the Experience 0.083 0.115 output distance function was considered appropriate as FBO membership −0.949* 0.064 it permitted the exploration of changes in the levels of Education −0.247*** 0.004 outputs relative to the production possibility frontier Household size 0.049 0.724 Access to credit −0.260 0.534 (frontier output mix). The test reveals that the translog Constant 0.863 0.547 functional form is more appropriate for the study. The *, **, *** significant at the 10%, 5% and 1% respectively. result of the first-order input elasticities also reveals that Source: Field survey all inputs significantly increased pineapple production in the district. The shadow share of other crops is negative relative to the production of pineapple in the output young farmers are deficient in resources and might not be mix. The indication of increasing returns to scale in the able to apply inputs or implement and certain agronomic study area suggests that farmers can still increase their practices efficiently which may be key areas influencing output and productivity by improving on their inputs of efficiency. production. The coefficient of extension visits is negative; hence, The mean technical efficiency of the pineapple frequent extension visits by extension agents is expected farmers in the Akwapim-South District was 85% given to enhance technical efficiency of farmers and thus their the current technology available to them. Thus there is productivity. Feed the Future (2015) posits that effective the potential to increase their current output by 15%. Ana- agricultural extension systems, which are characterized lyzing the determinants of technical efficiency indicated by quality and timely delivery of services to farmers, con- that efficiency of farmers increases with an increase in tribute significantly to growth and development in the age and farmers who have more extension visits are agricultural sector. Extension agents serve as a link more likely to be efficient. Highly educated farmers are between farmers and researchers and are channels more likely to be efficient. Farmers who have access to through which new innovations in farming methods are credit are also more likely to be efficient. introduced and communicated to farmers. A study con- Pineapple farmers in the study area should increase ducted by Onumah, Brümmer, and Hörstgen-Schwark the use of inputs since they are at the first stage of the pro- (2010) indicates that output increased by 9% for fish duction process and as such increasing these inputs has farmers who had at least one extension visit during the the potential to increase pineapple output and reduce the 2007 production year. The findings of Ogundari (2013) average cost. An increase in the usage of these inputs is also conform to the findings of this study. possible if the Statistics, Research and Information The coefficient of education is negative; hence, edu- Department of MoFA makes them readily available to cation has a positive effect of increasing efficiency. Edu- farmers at subsidized prices. The youth should be given cated farmers are more capable of obtaining new special consideration in such interventions since they information on inputs and outputs prices as well as are more likely to have challenges in accessing inputs. being more willing to use modernized technologies to The Extension Service Directorate of MoFA should improve their farm operations as compared to uneducated structure the extension and advisory services properly farmers who are conservative in their production methods. for pineapple farmers in the district so that its impact Educated farmers are also able to keep accurate records of can be better felt and in turn translated into higher effi- farm operations and ensure the optimal utilization of ciency of the various pineapple farms. The Extension various farm inputs through effective resource allocation Service Directorate of MoFA must also ensure the techniques. This conforms to the findings of Idiong regular training of its extension agents to keep them (2007), whose study suggested that educated farmers are abreast of current and best pineapple production practices. more capable of obtaining new information and utilizing Extension agents must teach these farmers the right agro- it to improve their farm operations and efficiencies as nomic practices to use in the field. Also, farmers are more compared to uneducated farmers who are conservative likely to boost their production with the adoption of these in their production methods. improved techniques and meet international standards in Participation in an FBO has a significant effect on order to survive competition. In addition, Extension increasing efficiency. This implies that farmers who are Agents can assist the strengthening of FBOs and encou- members of these organizations are more efficient than rage farmers in their districts to participate in order to those who do not participate. FBOs ensure collective derive the benefits of being a member. action in acquiring inputs for their members as well as sharing valuable information during meetings. It is there- fore not surprising that members perform better. Our Acknowledgements result is consistent with Attipoe et al. (2020) and Jaime We would like to thank the Ministry of Food and Agriculture (MoFA) of the Akwapim-South District for their support and Salazar (2011), who reported that farmers who are during this research as well as the anonymous reviewers for association members have higher technical efficiency making this work a success. We are also thankful to our families levels than those who are not members. for their support and patience. 10 Ofori-Appiah, Onumah and Asem Disclosure statement Farrell, M. J. 1957. “The Measurement of Productive Nopotentialconflictofinterestwasreportedbytheauthor(s). Efficiency.” Journal of the Royal Statistical Society. Series A (General) 120: 253–281. Feed the Future. 2015. “The U.S. Government’s Global Hunger ORCID and Food Security Initiative.” Agricultural Extension Policy Yaw Ofori-Appiah http://orcid.org/0000-0002-6900- Forum: Ghana. Accessed at https://ghanalinks.org/ 1656 documents/190376/199081/USAID+Agriculture+Policy +Support+Project+FY+2015+Annual+Report/64af59fd- 7e36-411a-8aa8-e98fe0b9d83f?version=1.0 References Gatune, J., M. Chapman-Kodam, K. Korboe, F. Mulangu, and Abdulai, A., and W. Huffman. 2000. “Structural Adjustment and M. A. Rakotoarisoe. 2013. “FAO Commodity and Trade Economic Efficiency of Rice Farmers in Northern Ghana.” Policy Research Working Paper No. 41.” Analysis of Economic and Cultural Change 48 (3): 503–520. Trade Impacts on the Fresh Pineapple Sector in Ghana. Aigner, D. J., C. A. K. Lovell, and P. Schmidt. 1977. “Formulation Rome: Food and Agricultural Organization of the United and Estimation of Stochastic Frontier Production Function Nations. Models.” Journal of Econometrics 6: 21–37. Ghana Statistical Service. 2014. “2010 Population and Housing Attipoe, S. G., C. Jianmin, Y. Opoku-Kwanowaa, and F. Ohene- Census. District Analytical Report: Akwapim-South Sefa. 2020. “The Determinants of Technical Efficiency of District.” Accessed at https://www2.statsghana.gov.gh/ Cocoa Production in Ghana: An Analysis of the Role of docfiles/2010_District_Report/Eastern/AKWAPIM% Rural and Community Banks.” Sustainable Production 20SOUTH.pdf. and Consumption 23: 11–20. Ghana Statistical Service. 2018. “Provisional 2018 Annual Battese, G. E., and T. J. Coelli. 1988. “Prediction of Firm-Level Gross Domestic Product.” Accessed at https://www2. Technical Efficiencies With a Generalized Frontier statsghana.gov.gh/docfiles/GDP/GDP2018/2017% Production Function and Panel Data.” Journal of 20Quarter%204%20and%20annual%202017%20GDP% Econometrics 38 (3): 387–399. 20publications/Annual_2017_GDP_April%202018% Battese, G. E., S. J. Malik, and M. A. Gill. 1996. “An 20Edition.pdf. Investigation of Technical Inefficiencies of Production of Grosskopf, S., K. J. Hayes, L. L. Taylor, and W. L. Weber. 1997. Wheat Farmers in Four Districts of Pakistan.” Journal of “Budget-Constrained Frontier Measures of Fiscal Equality Agricultural Economics 47 (1): 37–49. and Efficiency in Schooling.” Review of Economics and Berndt, E. R., and L. R. Christensen. 1973. “The Internal Statistics 79 (1): 116–124. Structure of Functional Relationships: Separability, Idiong, I. C. 2007. “Estimation of Farm Level Technical Substitution, and Aggregation.” The Review of Economic Efficiency in Smallscale Swamp Rice Production in Studies 40 (3): 403–410. Cross River State of Nigeria: A Stochastic Frontier Binam, J. N., J. Tonye, G. Nyambi, and M. Akoa. 2004. “Factors Approach.” World Journal of Agricultural Sciences 3 Affecting the Technical Efficiency Among Smallholder (5): 653–658. Farmers in the Slash and Burn Agriculture Zone of ISSER. 2017. “The State of the Ghanaian Economy in 2016.” Cameroon.” Food Policy 29 (5): 531–545. Legon: Institute of Statistical, Social and Economic Brümmer, B., T. Glauben, and G. Thijssen. 2002. Research, October 2017. ISSER, 246. “Decomposition of Productivity Growth Using Distance Jaeger, P. 2008. “Ghana Export Horticulture Cluster Strategic Functions: The Case of Dairy Farms in Three European Profile Study Part I – Scoping Review.” Prepared for Countries.” American Journal of Agricultural Economics World Bank Sustainable Development Network (WB- 84: 628–644. SDN), The Republic of Ghana Ministry of Food and Coelli, T. J., and G. E. Battese. 1996. “Identification of Factors Agriculture and European Union All ACP Agricultural Which Influence the Technical Inefficiency of Indian Commodities. Farmers.” Australian Journal of Agricultural Economics Jaime, M. M., and C. A. Salazar. 2011. “Participation in 40 (2): 103–128. Organizations, Technical Efficiency and Territorial Coelli, T., and S. Perelman. 1996. Efficiency Measurement, Differences: A Study of Small Wheat Farmers in Chile.” Multiple-Output Technologies and Distance Functions: Chilean Journal of Agricultural Research 71 (1): 104–113. With Application to European Railways (No. DP 1996/05). Kayitesi, R. 2011. “Factors Affecting Small Scale Farmers’ CREPP. Pineapple Production: The Case Study of Ngoma District, Coelli, T., and S. Perelman. 2000. “Technical Efficiency of Rwanda.” A Research Project submitted to Van Hall European Railways: A Distance Function Approach.” Larenstein University of Applied Sciences in partial fulfill- Applied Economics 32 (15): 1967–1976. doi:10.1080/ ment of the requirements for the award of Professional 00036840050155896. Master Degree in Management of Development. https://ede- Coelli, T., S. Perelman, and E. Romano. 1999. “Accounting for pot.wur.nl/192623. Environmental Influences in Stochastic Frontier Models: Kleemann, L. 2011. Organic Pineapple Farming in Ghana – A With Application to International Airlines.” Journal of Good Choice for Smallholders? Kiel Institute for the Productivity Analysis 11: 251–273. World Economy. Working Paper No. 1671. Coelli, T. J., D. S. P. Rao, C. J. O’Donnell, and G. E. Battese. Kumbhakar, S. C. 1994. “Efficiency Estimation in a Profit 2005. An Introduction to Efficiency and Productivity Maximising Model Using Flexible Production Function.” Analysis. New York, NY: Springer Science & Business Agricultural Economics 10 (2): 143–152. Media. Kusa, M. U. 2012. “Farmer’s Health and Agricultural Debreu, G. 1951. “The Coefficient of Resource Utilization.” Productivity in Rural Ethiopia.” Thesis., Norwegian Econometrica 19 (3): 273–292. University of Life Sciences. Essilfie, F. L., M. T. Asiamah, and F. Nimoh. 2011. “Estimation Lovell, C. A. K., S. Richardson, P. Travers, and L. Wood. 1994. of Farm Level Technical Efficiency in Small Scale Maize Resources and Functioning: A New View of Inequality in Production in the Mfantseman Municipality.” Journal of Australia. Models and Measurement Welfare and Development and Agricultural Economics 3 (14): 645–654. Inequality. Vol. 38. Berlin: Springer Verlag Press. 225. Eurostat. 2013. Eurostat international trade data. EU Fruit doi:10.2307/2683664. Import from Ghana (2000–2013). (Eurostat is the Official Mawson, P., K. Carlaw, and N. McClellan. 2003. Productivity Statistical Office of the European Union). http://ec.europa. Measurement: Alternative Approaches and Estimates (No. eu/eurostat. 03/12). Wellington: New Zealand Treasury. African Journal of Science, Technology, Innovation and Development 11 Meeusen,W.,andJ.VandenBroeck.1977.“EfficiencyEstimation Ogundari, K. 2013. “Crop Diversification and Technical from Cobb-Douglas Production Functions With Composed Efficiency in Food Crop Production.” International Error.” International Economic Review 18: 435–444. Journal of Social Economics 40 (3): 267–288. Mensah, A., and B. Brümmer. 2016. “AMulti-Output Production Ogundari, K., and B. Brümmer. 2011. “Estimating Technical Efficiency Analysis of Commercial Banana Farms in the Efficiency, Input Substitution and Complementary Effects Volta Region of Ghana: A Stochastic Distance Function Using Output Distance Function: A Study of Cassava Approach.” African Journal of Agricultural and Resource Production in Nigeria.” Agricultural Economics Review 12 Economics Volume 11 (4): 305–316. (2): 62. Ministry of Food and Agriculture. 2010. Medium Term Onumah, E. E., B. Brümmer, and G. Hörstgen-Schwark. 2010. Agriculture Sector Investment Plan (METASIP) 2011– “Productivity of the Hired and Family Labour and 2015. Accra: Government of Ghana. Determinants of Technical Inefficiency in Ghana’s Fish Ministry of Food and Agriculture. 2015. Republic of Ghana. Farms.” Agricultural Economics (Zemědělská Ekonomika) Accessed June 1, 2018. http://mofa.gov.gh/site/page id=1495. 56 (2): 79–88. doi:10.17221/38/2009-agricecon. Ninson, D. 2012. “Analysis of Pineapple in the Akwapim-South Onumah, E. E., J. A. Onumah, and G. E. Onumah. 2018. Municipal Area.” Department of Agricultural Economics “Production Risk and Technical Efficiency of Fish Farms and Agribusiness, College of Agriculture and Consumer in Ghana.” Aquaculture 495: 55–61. Sciences, University of Ghana, Legon, Ghana. Wang, H. J., and P. Schmidt. 2002. “One-Step and Two-Step O’Donnel, C., and T. Coelli. 2005. “A Bayesian Approach to Estimation of the Effects of Exogenous Variables on Imposing Curvature on Distance Functions.” Journal of Technical Efficiency Levels.” Journal of Productivity Econometrics 126 (2): 493–523. http://www.sciencedirect. Analysis 18 (2): 129–144. com/science/article/pii/S0304407604001216.