Performance of Smallholder Soybean Farmers in Ghana; Evidence from Upper West Region of Ghana


 The economic importance of soybean towards poverty alleviation and food security is gaining wider popularity and common acceptance among smallholder farmers in sub-Sahara Africa, especially in Ghana. Commercial soybean cultivation is relatively new in Ghana; hence it has recently benefited from several productivities enhancing innovation/technologies. However, despite these efforts, productivity has remained low. This paper investigates factors affecting production efficiency among commercial soybean farmers, across the three commercial districts of the Upper West region of Ghana. A cross-sectional data collected from 271 soybean farmers were used to investigate technical efficiency of soybean production. The overall mean technical efficiency estimate is 59% with a scale elasticity of 0.89-indicating a huge scope for efficiency improvement. The result shows that, factors affecting technical efficiency are dependent on the farmer’s socioeconomic status. With the existing technology and production recourses, soybean farmers can improve their current levels of soybean production by 41% through the adoption of best production practices.

smallholder farmers using the translog stochastic production frontier in Northern region of Ghana. 118 Principal findings from the study indicate that farm size significantly influences farmer's 119 efficiency in soybean production. Other factors such as farmers experience, educational level and 120 access to credit had no influence on farmer's production efficiencies. The mean technical 121 efficiency of 0.61 was reported for farmers in the region which supposes that soybean farmers can 122 obtain only 61% of their potential outputs given the available inputs and technology. 123 In another study, Abdul-Kareem (2016) estimated the technical efficiency of cassava farmers in 124 the Savannah Zone of Northern Ghana and found a mean technical efficiency of 51%. Education 125 was found to correlate negatively with technical inefficiency, meaning educated farmers are more 126 technically efficient than the uneducated ones. Extension service also had a negative influence on 127 efficiency. However, farm size influenced technical inefficiency positively, implying that an 128 increase in farm size increases inefficiency. The study also found technical efficiencies across 129 gender to vary. Hence, the author recommends specialized extension officers training and the 130 establishment of farmer field schools to sharpen farmers' experience. considered more appropriate for examining the efficiency and productivity in agricultural 141 production (Aigner et al., 1977), hence the SFA is adopted in this paper for examining the 142 productive efficiency of soybean farmers in the Upper West Region of Ghana. 143 To empirically examine technical efficiency of soybean production, the general stochastic frontier 144 production model is estimated and the model for technical inefficiency proposed by Battese and 145 Coelli (1993Coelli ( , 1995 is adopted to investigate the technical inefficiencies of soybean production. 146 Aigner et al. (1977) and Meeusen and van den Broeck (1977) introduced the stochastic production 147 frontier based on parametric technique (Trujillo and Iglesias 2013) to estimate a firm's technical  (1) 155 where, Y denotes output, X and β are vectors of inputs and parameter estimates, respectively. is 156 the stochastic error or statistical noise outside the influence of the farmer. It is assumed to be half 157 normally distributed as N (0, 2 ) and independent of . Conversely, is a non-negative random 158 variable associated with technical inefficiency which is under the control of the producer. It is 159 assumed to be independent and identically distributed non-negative truncation of the N (0, 2 ) and 160 independent of . The TE as posited by Battese (1992) is the ratio the of frontier output to the observed output, 162 consequently TE is stated as: Since represent a non-negative random variable accounting for technical inefficiency and are 165 within the farmer's control, factors that affect such inefficiency are stated as: Where denotes parameters and are socio-economic and other factors under the control of the 168 farmer influencing efficiency. The variance parameters are expressed as: Farmers' individual efficiency levels are also obtained as: The translog stochastic frontier (SF) is thus represented as: The values of the explanatory variables in the translog SF model in equation (6)  production, as well as animal rearing, are the main agricultural activities [GSS, 2012].

192
Sampling procedure and size 193 A multi-sampling technique was used to sample 271 soybean farmers for the study (see Table 1 194 below). At stage one, purposive sampling was used to select three (3)

204
This paper used primary quantitative data collected through a formal survey using a structured  production and inefficiency are explicitly expressed in equations (7) and (8) as: where the output is the yield of soybean measured in kgha -1 and is a non-negative random 224 variable associated with technical inefficiency. The explanatory variables are explained in Table   225 2. Summary of demographic characteristics of soybean farmers as well as summary overview of 231 statistics of key variables used in the technical efficiency models are presented in Table 3. The  ... 0 H β β = = . This hypothesis is rejected at the one percent level suggesting that 269 in deed farm specific input factors included in the model jointly influence soybean output.

270
The second hypothesis in Table 4 states that the second-order coefficients in the translog  Reject null hypothesis at 1% Note: *** , ** , and * denote statistical significance at the 1%, 5% and 10% levels, respectively.

278
The third hypothesis that technical inefficiency effects are absent from the model at every level : 0 H U = , implying that the deterministic production function is desirable is also rejected at 1% 280 significance level. This indicates that the stochastic production frontier approach applies.

281
Presented in Table 5 is the maximum likelihood estimates of the translog stochastic frontier 282 production model. The results show that significant effects of seed, labor, farm size and fertilizer 283 on soybean output. These factors play an important role in determining soybean outputs among 284 farmers in the Upper West Region. Improving productive efficiency of soybean will require 285 enhancing farmers access to improved seeds and fertilizers. In addition, the square of seed, 286 fertilizer, farm size, and interactive terms for seed*labour, seed*fertiliser, inoculants*labour, and 287 seed*farm size are also significant at various probability levels, however, seed*farm size, had 288 negative effect on soybean yield. The positive and significant effect of seed*farm size implies that 289 land productivity increases with increasing use of seed. In other words, increasing seed and farm 290 size have complementary effects on productivity of soybean production. The negative effect of 291 seed*labour and seed*fertilizer implies that seed productivity in soybean production increased 292 with decreasing labour and fertilizer productivity. This suggests that after certain quantity of seed 293 is used increasing the use of fertilizer or labour do not necessarily increase overall soybean 294 productivity. Similar explanations go for inoculants*labour.
The Wald Chi-square statistic is significant at 1%, implying that the inputs jointly explain the 296 variation in soybean output. The high gamma value of 0.94 for soybean farmers in the study area 297 signifies the presence of technical inefficiencies in soybean production among the sampled 298 farmers. The gamma value indicates that about 94% of the variation in the output of soybean 299 farmers from the translog frontier output was due to inefficiency in production. That is 94% of the 300 variation in composite error term was attributed to the inefficiency component. The results of the estimated elasticities of the factor inputs at mean output values are illustrated in 305   Table 6. The data reveals a scale elasticity of 0.89 implying that soybean farmers in the Upper

306
West region of Ghana are operating within stage II of the production possibility curve often 307 referred to as the 'economic zone. It means that the output of soybean is increasing at a decreasing 308 rate and hence employing additional unit of all resources results in less than an additional unit of 309 output. it further implies that, soybean production in the study area is inelastic. In other words, a 310 percentage increase in input use leads to a lesser percentage corresponding increase in output.

311
Despite this, farmers are expected to use more additional inputs until maximum output is reached.  Elasticity of production (EP) 0.89 ***, ** and * denote statistical significance at the 1%, 5% and 10% level, respectively.

317
The elasticity of farm size (0.68) is positive and significant at 1% probability level. The coefficient 318 suggests that a percent increase in farm size leads to a 0.68% increase in soybean output, all things 319 being equal. Therefore, the result confirms the assumption that an increase in cultivated farmland

355
The determinants of technical inefficiency among soybean farmers is shown in Table 8. Negative  production was found to be a key factor influencing technical inefficiency in production.

379
The results further show that major factors that contributes to soybean output include land, labor, 380 fertilizer and seed. Positive effects were found with land, labor, seed and soybean outputs, whiles 381 fertilizer has a negative effect. To improve efficiency in soybean production in the Upper West 382 region of Ghana will require policies that enhance farmers access to land, labour and improved 383 seeds. Similar of such programmes include the planting for food and jobs programme being 384 implemented by the Government of Ghana, which include facilitating access to improved seeds 385 and fertilizer. However, such strategy can be enhanced by the government through private 386 partnership arrangements with seed companies should provide subsidized certified soybean seeds 387 for farmers to improve production. Finally, there is the need for policy to enhance access to easy arable lands through active engagements with the chiefs and elders in production communities 389 especially for women farmers. Availability of data and material 394 The datasets used and/or analysed during the current study are available from the corresponding 395 author on reasonable request.

397
Competing interest 398 The authors declare that they have no competing interest 399 400

401
The study did not receive any form of funding.

404
FAA wrote the proposal, collected the data, analysed it and wrote the paper. FA also participated 405 in the data collection and performed the analysis. FN, BOA and AM edited and reviewed the 406 manuscript. All the authors read and approved the manuscript for publication.

408
Acknowledgement 409 We express our sincere gratitude to the staff of the Department of Agriculture in the Sissala West,

410
Wa East and Dafiama-Busie-Issa districts of Ghana for their immense contribution during the data 411 collection process. We are also grateful to the entire staff of TUDRIDEP and PRONET-NORTH 412 as well as the smallholder soybean farmers in the respective districts without whom this study 413 would not have materialised. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.