Price competitiveness of smallholder rice farmers under cooperative irrigation schemes in Coast and Morogoro regions ,

Tanzanian produced rice is not competitive because of high production costs. To increase competitiveness will require improvements in production efficiency. This study was conducted to analyse production cost efficiency using cross sectional data (2013/2014 agricultural season) from 200 farmers that belong to four cooperative irrigation schemes in Coast and Morogoro regions. Two stage sampling and translog stochastic cost frontier were used for sampling and cost efficiency analysis, respectively. Production costs were estimated quantitatively using the enterprise budgeting technique. Study data suggest that total costs of production are US$315/MT. Rice output and prices for labour, fertilizer and irrigation water significantly affected costs of production. Unit cost of production were reduced by increasing rice productivity (p<0.05). Production efficiency was significantly influenced by farming experience, planting methods, frequency of weeding, degree of specialization and source of purchased inputs. These factors accounted for 82% of the variability in costs of production (p<0.05). A major conclusion is that production efficiency is reduced by high production costs relative to rice output produced, input prices, source of purchased inputs and other agricultural practices. Use of labour saving technologies, the purchase of inputs from irrigation scheme cooperatives and greater economies of scale resulting from increased specialization can increase profits.


INTRODUCTION
Rice in Tanzania is the second most important food crop after maize, being grown by 18% of the farming households and more marketed than maize (URT, 2013).The quantity of marketed rice is approximately 42% of the total production while that of maize is 28%, thus being more commercialized than maize contributing to 2.67% of the GDP (EUCORD, 2012).The average production in the country is lower than the actual rice demand, *Corresponding author.E-mail: kangilej@gmail.comor chollima@iwayafrica.com.Tel: +255 755 248598.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License evidenced by decreasing rice food supplies in the country.Rice milled equivalent kg/capita/year in 2010, 2011and 2012 were 36.88;31.71 and 25.29, respectively (MAFAP, 2013).Furthermore, the decision of the government of Tanzania in 2013 to exempt the Common External Tariff (CET) of 75% in importation of 60000MT of rice justifies the existence of the supply gap.
In irrigated rice production, smallholder farmers are the main driver with exception of few large scale producers (SAGCOT, 2012).Thus, smallholder irrigated rice production is imperative to the rice sector and the country in general.On the other hand, price competitiveness that is, the ability of locally produced rice to compete with imported rice in the market is also important for growth of the sector and smallholder rice farmers.Low price competitiveness of produced rice in Tanzania gives room for imported rice from various countries in the world by reducing profit margin to smallholder farmers and market for domestic produced rice.For example, milled rice from Pakistan is imported in Tanzania at a Cost Insurance and Fleight price (CIF price) of US$390/MT and retailed at a price of US$500/MT, while domestic producers breaks even if they sell milled rice at US$605/MT (MAFAP, 2013).
Smallholder irrigated rice production is done in irrigation schemes which are managed either by irrigation scheme associations or cooperatives (cooperative irrigation schemes).Irrigation water is pumped using electrical and diesel engine pumps (pump-fed canal system).Some of the irrigation schemes are under flood recession and gravity-fed systems which does not necessarily need the pumps.
The irrigation scheme association is the group of farmers using water for irrigation in an irrigation scheme, having their own leadership and enforcing formal and informal rules such as social sanctions (URT, 2010).The irrigation scheme association is more concerned with allocation of water, operations and maintenance of the irrigation scheme with minimal or no involvement in marketing activities.
In cooperative irrigation schemes, the cooperative board manages the scheme on behalf of its members (farmers).Marketing activities are done in addition to maintenance, operation and management of the scheme.Marketing activities are usually done on the side of inputs.Available evidence indicates that, farmers rarely engage in collective marketing on the side of outputs.Farmers keeps their own stocks and does private marketing (Kilimo Trust, 2014;RCT, 2015).On the side of inputs, the cooperative board purchases in bulk and sells to members or facilitates on getting suppliers of inputs to members.
Underutilization of the cooperative irrigation schemes by smallholder farmers has been a concern of the government to an extent of either privatizing some of these irrigation schemes or running them in partnership with foreign investors (BMG, 2012;EUCORD, 2012).Smallholder farmers in these cooperative irrigation schemes have been failing to contribute to the payments for irrigation water supply, thus left the fields to other farmers who may not necessarily be members of the irrigation scheme cooperatives.
This situation may be linked to high production costs which results into low profit margins as Zaal et al. (2012) associate high production cost leading to low price competitiveness in irrigated rice production with the capital intensive nature of irrigated rice production system.Smallholder farmers in cooperative irrigation schemes, are also capital intensive in their rice production system due to high use of capital inputs (purchased inputs).The government claimed underutilization of the cooperative irrigation schemes and the failure of some smallholder farmers to pay for irrigation water supply may be due to the ascertained low price competitiveness of produced rice or any other factors at the cooperative management or smallholder farmer level.
The present study investigated smallholder farmers' decision making level in cooperative irrigation schemes on use of production inputs whether follows the efficiency path of the farm production resources.Efficiency path shows the degree of achievement in the allocation of the available inputs and output produced in order to attain a high degree of efficiency in cost, revenue or profit.Literally, efficiency is the ability of a decision making unit to obtain the maximum output from a set of inputs (output orientation) or to produce an output using the lowest possible amount of inputs combination (input orientation) (Kumbhakar and Lovell, 2003).Dzeng and Wu (2013) recently defined efficiency as the goal oriented concept for determining the best scenario to use the lowest input or reach the highest output.The study focused on investigation of efficiency in cost.The inefficiency which is associated with the loss of productivity due to inability of farmers to use production inputs in their optimal proportions was also investigated.
The study also ascertains the level of production costs since the reported high production costs in irrigated rice production leading to low price competitiveness is not specific for this type of irrigation scheme management.

METHODOLOGY Description of the study area
The study was conducted in Morogoro and Coast regions focusing on cooperative managed and rice specialized irrigation schemes.In Morogoro region, it was conducted in Mvomero district under irrigation schemes serving Dakawa and Mlali wards while for Coast region, the focus was in Bagamoyo district under irrigation schemes that serves Magomeni, Dunda and Vigwaza wards (Figure 1).

Data collection, sampling procedures and sample size
The study was cross sectional and involved collecting data from farmers in a single agricultural season, 2013/2014 using semistructured questionnaire.A sample of farmers was drawn using a two stage sampling method.The first stage involved selection of four out of six irrigation schemes using probability proportional to size (PPS).The PPS method involved listing all rice specialized cooperative irrigation schemes, identifying the number of beneficiaries (population size), calculating the cumulative population in each irrigation scheme and calculating the sampling interval (SI) that is,

Cumulative total population 2311 578
Number of clusters 4  , and probability 1 and probability 2 of each irrigation scheme.Probability 1 is the likelihood of selection for each sampled irrigation scheme and probability 2 being the likelihood of selection for each individual farmer in each of the sampled irrigation schemes.Four clusters (number of irrigation schemes to be studied) were used.The last step of implementation of the PPS method was to generate random numbers and select the one which is equal or less than the SI, this is sometimes known as Random Start (RS).The random number selected was 275.The first irrigation scheme was then selected by looking for the irrigations scheme whose cumulative population size exceeds this random number in which CUMKI was selected.The second, third and the fourth irrigation schemes were selected using the same criterion by considering the SI that is; 275+578=853 for the second, 853+578=1431 for the third and 1431+578=2009 for the fourth in which UWAWAKUDA, TEGEMEO/BIDP and CHAURU irrigation schemes were respectively selected (Table 1).The second stage involved obtaining 50 smallholder farmers from each irrigation scheme sampled in first stage using Systematic Random Sampling (SRS) making a sample of 200.

Analytical framework
Quantification of costs of production was conducted using enterprise budgeting technique.The technique involved quantification of input costs such as irrigation water, seeds, fertilizer, herbicides, pesticides and labour.Land was not included in the quantification of the costs since farmers are given free by the irrigation scheme cooperative and the farmer has no direct decision on land thus cannot easily be allocated to an enterprise.The government land rent is paid by the cooperatives thus it is not a direct cost to farmers of the irrigation scheme cooperative.Labour costs were quantified from rice production activities of field clearing, ploughing, hallowing, planting, field water management, weeding, fertilizer application, herbicides application, and pesticides application, birds scaring, harvesting and bagging.In situations where family labour was used, equivalent wage cost of working off-farm for a wage was used.Other cost items were purchase of bagging materials and transport costs.Number of people (labour unit) required to perform a particular amount of work was also estimated to enable the quantification of price of labour.The amount of work that can be done by one labour unit in one day is called man-day (one man day is equivalent to 8 working hours).Thus, the number of days spent on doing a particular activity in the field was estimated.The price of labour was then obtained by taking the total cost of labour divided by the number of days taken on that activity.
Farm level cost efficiency was determined using stochastic frontier analysis given its ability to decompose deviations from the efficient frontier into two components of inefficiency and error term (Aigner et al., 1977;Coelli, 1995;Kumbhakar and Lovell, 2003).
The production technology was implied by a Translog function forming a Translog stochastic cost frontier.It should be noted that, Translog stochastic cost frontier has three distinct terms; linear, quadratic and interaction terms.The stochastic cost frontier function was modelled in a four input framework as shown in Equation 1with fertilizer, labour, irrigation water and rice seeds.Cost of production was measured in Tanzanian Shillings (TAS) per hectare (ha) and per metric ton (MT) of rice produced.Rice output was measured in MT/ha; average price of fertilizer of both basal application and top dressing applications used in one agricultural season in TAS/kg; price of seeds used in either transplanting, broadcasting or direct seeding in TAS/kg; price of irrigation water used in production paid to the irrigation scheme cooperative in TAS/ha and the price of labour was obtained by taking the total labour cost of an individual farmer divided by the total man days used in production (TAS/man day).One man day is the amount of work that can be done by one labour unit in one day (8 working hours).

 
  Cost efficiency frontier function is homogeneous of degree one in input prices such that ,, P_fert is the normalized price of fertilizer; P_lab is the normalized price of labour; P_water is the normalized price of irrigation water and s  are parameters to be estimated.
The above model was estimated by single step procedure using Frontier version 4.1 (Coelli, 1996).The single step procedure was used in order to avoid bias as the result of misspecification that is always brought by the use of two step procedure.Single step procedure estimates in a single equation the parameters for the efficiency model, cost efficiency scores, value of gamma and sources of inefficiency in the production system.
Cost efficiency scores ranges from 1 to infinity in a cost efficiency Purchasing inputs through the cooperative -Purchasing inputs from other input providers + frontier model.However, in case there is interest of showing cost efficiency ranging from 0 to 1, the approach is to take the reciprocal of the cost efficiency in cost frontier.
The value of Gamma (γ) indicates the level of inefficiency such that γ=0 implies that, deviations from the frontier are entirely due to noise as there is no evidence for presence of inefficiency effects.The value of γ=1 would mean that, all deviations from the frontier are due to inefficiency (inefficiency effects are highly significant in the production system).Furthermore, the likelihood ratio test (LR test) is used to compare the fitted model which includes inefficiency factors, to a corresponding model without inefficiency factors.The hypotheses being tested are there is no inefficiency implying that smallholder farmers are 100% efficient (Null hypothesis) and there is inefficiency implying that smallholder farmers are not 100% efficient (alternative hypothesis).Sources of inefficiency were determined through assumption on the inefficiency error component.The inefficiency error component i U was assumed to follow a truncated normal distribution (Equation 3) with a mean as a function of the hypothesized sources of inefficiency in production (Battese and Coelli, 1995).
Empirically, the inefficiency model used was specified as shown in Equation 4; The value of 0.. 7 0 .....   implies that there is no inefficiency in the production system.The variables 1.... 7 ZZ are given in Table 2 and 07 ...  were parameters to be estimated.
The negative expected sign shows a decrease of inefficiency (increasing cost efficiency) and the positive expected sign implies that a particular factor increases inefficiency (decreasing cost efficiency).Farming experience as measured by years of rice farming experience is expected to reduce the level of inefficiency in smallholder farmers' production system.The more years a farmer spends in irrigated rice production the more skills are acquired in managing the production inputs, hence increasing efficiency.Good Agricultural Practices (GAPs) are important in ensuring efficiency in any agricultural production system.Proper ways of harvesting, planting, weeding and use of quality seeds are vital for smallholder irrigated rice production in cooperative irrigation schemes.The use of combine harvester or motorized rice thresher as mechanization method in harvesting activity is expected to reduce inefficiency in harvesting activity, thus improving the cost efficiency.Manual harvesting is expected to be inefficient in terms of cost, quality of rice and time, hence can lead to inefficiency in the harvesting activity.
There are three main methods of planting rice; transplanting, broadcasting and direct seeding (dibbling).Transplanting method of planting allows the farmer to plan apriori the spacing, reduces weeds and leads to high yields.It is expected that, as farmers in irrigated rice production use transplanting method of planting, their efficiency increases through yield increase and reduced weeds and seeds for seedling.The increase in frequency of weeding (number of times weeding is done from planting to harvesting) increases cost efficiency due to reduced weed infestation in rice.Improved seeds also increase efficiency in production.Degree of specialization, Total rice acreage *100% Total acreage of all crops , is the measure of economies of scale.Specialization allows farmers to enter into bulk markets as the average cost of production falls.Additionally, source of purchased inputs (whether a farmer It is hypothesized that, purchasing inputs through the irrigation scheme cooperative increases cost efficiency in their production system.

Production activities and cost
Activities in irrigated rice production utilizing labour start from field clearing or ploughing to bagging as shown in Figure 2. Activities in which the farmer is incurring high average costs in descending order were weeding, harvesting, planting and bird scaring.Furthermore, unit production cost was found to be TAS 504.43 kg -1 . According to Bank of Tanzania (BOT), the 2013 annual exchange rate was TAS 1599/USD, thus using this rate, this is equivalent to US$315.47/MT.

Maximum likelihood estimates of the cost efficiency model
Cost efficiency frontier function was estimated through single step procedure and the results are shown in Table 3.All linear terms (normalized rice output, normalized fertilizer price, normalized price of labour, and normalized price of water) were significant (p<0.05) and positive.
Price of fertilizer was the most sensitive variable influencing the total cost of production followed by price of irrigation water charged by the irrigation scheme cooperative.A 1% increase in prices of fertilizer and irrigation water was found to increase 2.216 and 1.996% of the total production cost, respectively.Price of labour was the least variable in terms of sensitivity as a 1% increase in price of labour was found to increase 1.159% of the total production cost.Furthermore, cost elasticity of rice output evaluated at the mean was 0.917.
In quadratic (squared) terms, only normalized rice output was significant and positive, showing evidence of some economies of scale.All interaction terms were not significant, but some were positive showing complementarity of the variables.Rice output and labour, rice output and irrigation water and fertilizer and irrigation water are complements in smallholder irrigated rice production system, implying to have zero elasticity of substitution.The constant term was 8.444, positive and significant (p<0.05) and the value of gamma (γ) obtained was 0.8208 indicating high variability in costs of production.
Furthermore, the Likelihood Ratio (LR) test for one sided error testing the hypothesis that, smallholder farmers are not 100% efficient was supported (The test statistic LR Chi-square [ 2 ] was 27.63 which is greater than the critical LR  2 of 3.84 read from statistical tables).

Production cost efficiency distribution
All farmers had efficiency levels above 1 indicating that, they are operating above the cost frontier.Results indicate that, 64% of all farmers were below the mean cost efficiency level and 36% above the mean (Table 4).The average efficiency level was found to be 1.139 (87.80%) indicating that, 13.9% of costs of production in irrigated rice production can be avoided without affecting the level of rice output.In other words, the mean cost efficiency of 1.139 implies farmers to experience a cost saving of 13.9% if they happen to achieve cost efficiency.Likewise, cost saving by attaining the average cost efficiency level is 10.31%, that is, and for attaining the minimum efficiency level, smallholder farmers will save 12.76% of the production cost, that is, respectively.

Inefficiencies in production under cooperative irrigation schemes
The inefficiency model results indicated that, farming experience, planting methods, frequency of weeding, degree of specialization and source of purchased inputs were significant (p<0.05).This implies that, they have a significant influence on cost efficiency.Type of rice seed variety planted and harvesting methods were not significant as shown in Table 5.All other variables except harvesting method variable had their expected signs.The positive sign indicate that, the variable under consideration increases the cost inefficiency in the production system, while the negative sign shows decrease in cost inefficiency (increasing cost efficiency).

DISCUSSION
Weeding was found to be the highest costing activity.The high cost on weeding is due to low level of mechanization on weeding.Smallholder farmers using herbicides for weeding were 34.5%, but as a supplement to manual weeding.The use of mechanized labour saving technologies in weeding is low.Similar result was obtained by Mdemu and Francis (2013) in their study in Kapunga irrigation scheme in Mbeya, Tanzania where weeding was the highest labour intensive activity due to low level of mechanization of this activity.High average cost of harvesting activity is due to 33% of smallholder farmers being using manual harvesting.Planting and bird scaring were 100% not mechanized.Rice output, price of fertilizer, price of labour, and price of irrigation water were significant and positive indicating that, increasing these variables in irrigated rice production will increase the total cost of production.This implies that, costs of production in smallholder farming system under cooperative irrigation scheme are more sensitive to changes in input prices and rice productivity.Interestingly, the unit cost of production was found to decrease by increasing rice output.This is because the cost elasticity of rice output evaluated at the mean, was 0.917 implying that, a 1% increase in production in terms of rice output will increase 0.917% of the total production cost that is, increase in production are higher than increase in cost of production.
The constant term implied that, expense on fixed factors of production are incurred regardless of whether the production takes place or not.This includes all production fixed inputs especially farm implements as their costs are spread over a long period of time and contributions to the irrigation scheme cooperative for maintaining the membership status.This is in agreement with Ghosh and Raychaudhuri (2010) and Hidayah et al. (2013) in their study of cost efficiency in rice in India and Indonesia, respectively.
The model results indicated further that, 82% of the variability in the total cost of production that is not accounted by the function is influenced by inefficiency factors in irrigated rice production under cooperative irrigation schemes and only 18% being due to random factors that are beyond or outside smallholder farmers' control.Furthermore, the LR test results indicated that smallholder farmers in cooperative irrigation schemes are not cost efficient.
Production cost efficiency distribution showed high efficiency level (87.8%) which is in agreement with Hidayah et al. (2013) who also obtained a high mean efficiency level (86.6%).The high mean efficiency level is due to high level of specialization allowing farmers to enjoy economies of scale as 56% of smallholder farmers in cooperative irrigation schemes use more than 50% of their land farming for rice production within which 40% are 100% farming rice only.Moradi et al. (2013) had similar reason for existence of economies of scale when found high cost efficiency levels for wheat production in Iran.
Farming experience had a negative influence on farmers' cost inefficiency.It is true that, as smallholder irrigated rice farmers spend more years in rice farming their expertise in combining resources increase thus minimizing the wastage on the use of production inputs which increases production cost efficiency.Maganga et al. (2012) in his study of cost efficiency of Irish potato farmers in Malawi found farming experience to be highly influencing the cost efficiency of farmers.Likewise, Audu et al. (2013) obtained the same results in the study of cost efficiency in cassava production in Nigeria.
Planting and weeding are important GAPs in smallholder irrigated rice production under irrigation schemes.Planting methods, transplanting in particular and more frequency of weeding done by farmers, reduced inefficiency since had negative influence on cost inefficiency.The result on frequency of weeding being influencing cost efficiency, contradicts the result by Maganga et al. (2012) in his study of cost efficiency in Irish potato production in Malawi who found increase in weeding frequency not significantly influencing cost efficiency.This can be due to the nature of the rice production system being more susceptible to weeds.
Degree of specialization as the measure of economies of scale had a negative influence on cost efficiency.Specialization permit producers to enter into big markets through expansion of output levels, spreading fixed costs which leads to reduced average cost per unit of output.Therefore, specialization in rice production increases cost efficiency of smallholder irrigated rice farmers.This is in agreement with Maganga et al. (2012) and Dzeng and Wu (2013).Maganga et al. (2012) found that, more specialized Irish potato farmers were more cost efficiency than their counterparts who were less specialized.Similarly, Dzeng and Wu (2013) in the study of construction industry in Taiwan found cost efficiency to be higher to firms focusing in building construction only than those are involved in civil and building construction.Furthermore, source of purchased inputs was found to influence cost inefficiency negatively.The situation of a cooperative member to purchase inputs through the irrigation scheme cooperative contributes to increasing efficiency in production that spurs cost efficiency due to reduced unit cost of input and accessibility of after purchase services offered through the irrigation scheme cooperative.
Conclusively, this study has revealed that, smallholder farmers' costs of production stands at US$315/MT which is higher than the farm gate prices of rice in other countries especially Asian countries.Example; Kilimo Trust (2014) report that, farm gate prices of rice in Bagladesh and India are US$175/MT and US$169/MT respectively.Weeding, harvesting, transplanting and birds scaring are activities highly costing farmers.In order to lower cost and make these activities efficient, it is important to use herbicides in weeding; motorized rice thresher or combine harvester in harvesting and transplanter and direct rice seeder machine in planting.
Costs of production was found to be more sensitive to changes in prices of inputs and outputs but with unit cost of production being decreasing by increasing rice output.All farmers had efficiency levels above 1 indicating that are operating above the cost frontier and experiencing loss of efficiency in production, hence lowering rice price competitiveness.It is thus plausible to use labour saving technologies, purchasing inputs from irrigation scheme cooperatives and attaining greater economies of scale resulting from increased specialization to foster rice price competitiveness.

Figure 1 .
Figure 1.Map showing wards served by cooperative irrigation schemes used in the study.

Figure 2 .
Figure 2. Irrigated rice production activities and their associated average costs.

Table 1 .
Rice specialized cooperative irrigation schemes in Coast and Morogoro regions (Ministry of Agriculture, Food Security and Cooperatives; Eastern Zone irrigation office).
*Cooperative irrigation scheme selected.

Table 2 .
Description of variables for inefficiency model.

Table 3 .
Maximum likelihood estimates of the cost efficiency model.

Table 4 .
Cost efficiency distribution of smallholder irrigated rice farmers.

Table 5 .
Sources of inefficiencies in production.