To assess the contribution of dairy farming to household food security and income, the primary data at the household level are required. A random sample of 200 households from four milking collection centers out of 10 with the population of 2000 people was targeted for this survey. These 10 milking collection centers (MCC) are registered in the Balkh Dairy Union and mainly located in the area of Dehdadi and Nahre shahi districts, 15 km east of Mazar-e-Sharif city. The sampling of farm households proceeded based on the field visit and preliminary discussion with Balkh Dairy Union on livestock production in Balkh province with emphasis on the milk production. The data was collected on farming activities with a focus on the dairy production, impact of the training and extension services on the level of milk production, and income composition of the households.
The criteria for selecting the sample were owing cattle and being registered by MCCs. Out of 2000 households living in the area 1500 households were providing milk at the time of survey since the survey was conducted in the winter time when the number of milk producers decreases. The households were randomly selected from the total 384 households in four MCCs. At the same time, a number of households were included in a survey that had dairy cattle but did not provide milk at the time of survey for some reasons. In order to show equal representation of sampled MCCs, 50 households were selected from the provided list by MCCs in each location. Due to cultural issues, the households were registered under the name of one male number of the families. Therefore, 50 names were chosen randomly in order to call their female milk producer for the interview.
The table shows slightly different percentage in interviewed milk producers with the max of 44 percent and a minimum of 39 percent although the numbers of interviewed people were the same in all MCCs. Difference in total population of milk producers at each MCC explains the difference in percentage. The standardized questionnaire for household survey was developed after preliminary discussion with milk collectors. The first hypothesis regarding contributive factors to production cost and income functions was made. A pre-test survey with 10 households was conducted to examine the accuracy of the questions before the official survey and the questionnaire was modified to comply with the real situation.
Table 3 Sample Composition
Milk producers registered and surveyed
MCC 1 MCC 2 MCC 3 MCC 4 Total
No. of registered Milk producers providing milk at time of the survey 129 113 113 125 380
No. of interviewed milk producers (random sampling) 50 50 50 50 200
% of registered milk producers interviewed 39 44 44 40 53
The questionnaire accounts for the changes in the households’ milk production and total income through questions covering such issues as level of milk production, production cost, farm and non-farm income.
The questionnaire was developed in English with no translation into local language since the surveyors were familiar with both languages.
The household questionnaire is provided in Annex 2. It comprised five sections:
The introductory section records general information of interview and its setting, including the name of interviewer and interviewee, the number of the interview, place and date of interview, and membership in the dairy cooperative.
Section A provides information about household composition and the following characteristics: number, name, education, and age of household members.
Section B contains questions about land ownership, type of land, cultivation, and the use of agricultural output.
Section C investigates milk producers experience, the number of families and paid laborers and time spent in different dairy activities, number and type of dairy animals, amount of milk produced and sold in winter and summer, price seasonality, calving seasonality, production input and cost, and capital goods are used in dairy farming.
Section D finally accounts for household’s total income/ revenue and its break down to different sources, the right of women to use income earned by them or other family members, and the impact of training and services provided by different organizations and NGOs on the level of milk production and production cost.
The researcher collected the data with the assistance of the second surveyor, within a period of two weeks in January and February 2016. At each location, the milk collector from MCC was met prior to the survey to explain the purpose of the research and receive the list of beneficiaries. Groups of five to 10 milk producers were gathered in each location in order to explain the aim of the survey and afterward the questionnaires were filled. In some places due to distance problem, households were visited individually. The female milk producers were considered as interview partners. In most cases they were cooperative for interviewing, only in few cases, the interviewees refused to cooperate and answer the questions. Overall, the surveyors were able to communicate smoothly with interviewees and observe that how the dairy activities have influenced the households life both economically and socially. In addition, semi-structured interviews were conducted with chief executives of two dairy processing plants in Mazar city. The quantitative and qualitative information from these interviews complement the quantitative records from the household survey.
5.2 Data analysis and statistical methodology
The principal motive of this section is to provide a comprehensive explanation of the procedure are done to analyze the quantitative data that are presented in the result part. This places an emphasis on two central issues in question. The first is to identify the correct form of underlying functions/equations to be estimated and second to describe the steps in order to employ the regression analysis and assess the following:
The contributive effect of production factors to total level of milk production
The proportion of different production factor cost in total milk production cost
5.2.1 Identifying the Functional Form
Determining the correct functional form for a given relationship, input and output, is almost not possible. The challenge is to select the best form for given task (Griffin et al, 1987, p. 220). In the work of Griffin et al (1987), twenty functional forms with their properties and algebraic forms have been presented. In addition, a group of criteria has been purposed in order to boost the selection of true functional form, which depend on the following:
1. Maintenance of hypothesis regarding objectives in the presence of theoretical and empirical bases
2. Availability of data and computing procedure
3. Data characteristics and conformance
4. Specific features of the application
Estimation of production output and production cost is the main concern of this study. Therefore, the functional form of both functions should be identified to run the regression. For estimating the household’s milk production, the linear approach cannot be used as it is more likely impossible that a farmer without milking cows produce milk. i.e. milking cows are the prior independent variable and no milk is produced without that but the empirical shows that the level of production can be positive even without this variable under the linear functional form. In addition, the first derivate of the linear production function indicates that the physical marginal product is a constant value which is not in line with the reality as the addition of one more cow does not result in constant increase in the level of produced milk. Therefore, if production function is not linear, other options should be considered. In the case of this study, we purpose multiplicative combination of production factors in which each production factor is not entering one by one but to the power of a number. This type of functions modifies the impact of each individual factor. We choose the Constant Elasticity of Substitution (CES) function as a generalization of the Cobb-Douglas function, which allows for any (non-negative constant) elasticity of substitution (Henningsen & Henningsen, 2011, p. 1). The formal setting of CES production function with two inputs is as follow:
9) y=F (αX_1^ρ+ (1-α) X_2^ρ )^(1/ρ) Where y is the output, X1, and X2 are the inputs quantities, α and 1-α are distribution parameters that sum up to one and determines the factors share, and ρ is substitution parameter that is used to derive the elasticity of substitution σ=1/(1+ρ) (Miller, 2008, p. 8).
If ρ → 0, σ approaches 1, CES turn to Cobb-Douglas form. For ρ → ∞ , σ approaches 0 and CES turn to Leontief and in the case that ρ → -1 , σ approaches to infinity and CES turns to linear production function (Henningsen & Henningsen, 2011, p. 1).
To use multiplicative form in order to apply the CES production function, we translate the dependent (output) and independent (input factors) variable to natural log and what we get is:
10) LnYPM_i=Ln β_0+ β_1 LnK_i+ β_2 LnL_i+β_3 Ln Intermediae factors_i+ϵ_i
It is still linear approach and the coefficient is no standing to power anymore but they are simply a multiplicative combination of production factors. We can afterward run the linear regression taking natural log to linearize multiplicative level approach to additive log approach.
However, the linear functional form can be considered appropriate for total households’ income and production cost function as it meets the criteria of this form of relation.
5.2.2 Estimating the Impact of Milk Production on Income with Respect to Incurred Cost
In order to estimate the contribution of dairy income to the total income of the households, we first estimate and measure the amount of production and incurred cost. Regression analysis was employed to measure and specify the contribution and significance of different production factors to total production and cost.
At first, it is necessary to conduct the estimated regression for milk production, which is simplified as follows:
11) LnYPM_i= Ln(β_0 ) ̂+ (B_1 ) ̂LNnomcows_i+ (β_2 ) ̂Lnfeeding_i+(β_(3 LN) ) ̂whw_i+ (β_(4 ) ) ̂Lnsta_i+ϵ_i
Where YPM is the yearly produced milk of household (i); nomcows is the number of milking cows, feeding accounts for different feeding staff consumed for milk production, whw is the working hours per week spend on dairying activities, and sta shows the estimated size of stalls where the cows are kept. These variables are determinants in the level of milk production. β_i is the estimated impact of one unit of explanatory variables on household yearly produced milk and ϵ_i is the error term.
The same regression is run to estimate the impact of production factors on milk production cost:
12) YPC_i= (β_0 ) ̂+ (B_1 ) ̂feeding_i+ (β_2 ) ̂vetc_i+(β_3 ) ̂lc_i+ (β_4 ) ̂a〖ic〗_i+(β_5 ) ̂n〖ic〗_i+ϵ_i
Where YPC is the yearly production cost, feeding indicate the yearly feeding amount, the number of times vetc have been used including vaccination, aic and nic are the number of artificial and natural insemination. β_i is the estimated impact of one unit of explanatory variables on the household yearly production cost and ϵ_i is the error term
We can then measure the gain from milk production by subtracting the yearly production cost from the yearly production of milk times fixed milk price.
13) Gain_i=(YPM_i*P (AFN/liter)-YPC_i