Previous Page Table of Contents Next Page

5.7 Data Handling and Construction of Variables

Variables were constructed for the purpose of empirically investigating the research questions raised in this study. The data collected covered information necessary to formulate farm-level indices of socio-demographic, economic, and efficiency indicators comparable across different categories of farms. Variables were extracted from the questionnaire in an attempt to explicitly measure differences in capturing negative environmental externalities across farms. Dependent variables as defined in Chapter 4 include profit per unit of output (for the stochastic profit frontier model) and expenditure on pollution abatement per unit of output (for the environmental regression model). Independent variables include farm inputs and infrastructure per unit of output, input and output prices, farm-household socio-economic and demographic characteristics, a scale variable, and many other relevant exogenous variables.

5.7.1 Brazil Data

In order to estimate the profit function frontiers of livestock production in Brazil, four equations were defined, one each for broilers, swine, dairy, and layers. The list of variables used in the estimation is defined in Chapter 5 of Annex V (Camargo Barros, 2003)to this synthesis report. In general, revenue is calculated as net revenue minus variable costs. Calculations of revenue and variable costs differ for each livestock producer. Revenue includes sales of output to contractors (if under contract) and/or to buyers, and manure sales. Variable costs include all costs incurred during the production process such as costs on hired labor, feeds, day-old-chicks (DOC), breeding stocks, electricity, litter, transport, fuel, and others.

5.7.2 India Data

Data for Poultry. Two teams conducted the survey for poultry production in India, separately for each state: one team in Andhra Pradesh and another team in Haryana. The data collected were validated and checked for inconsistencies, and were encoded for analysis. The definition of variables used in the analysis of the poultry data is given in Annex 2 of Annex II to this synthesis report.

Estimation of revenue, cost, and profit per annum involves a complex procedure in the case of poultry industry, because of the differences in the production structure between layer and broiler units and the differences in the reference period used for different input items. For instance, a layer unit consists of a number of batches in order to have a continuous flow of income. Data on output and most of the inputs are collected for only the latest completed batch, which forms only a part of the size of the unit. The size of the unit is expressed in terms of the total number of birds in all the batches at a certain period of time (one cycle).

The number of birds existing on the day of visit is not an appropriate indicator of the size of the unit, as some sheds may be vacant the day after the completion of a batch. It is more reliable to count the size of the unit when all batches exist on the farm. Since each unit has a gap at a different time, information on the present size of the unit is collected separately. The duration of a batch is about 72 weeks for a layer unit and about seven weeks in the case of a broiler unit. Annual estimates have to be derived for both types of units. Further, the data available for a batch has to be expanded for a cycle. In order to derive annual output of a layer unit from the output from a batch, a multiplier is constructed as the product of two factors: one for converting the batch estimate (about 72 weeks) to one year, and the other for converting the estimate for one batch to the estimate for the entire unit. If x is the duration of the batch (in weeks) and y (in weeks) is the gap between two batches, the first factor is 52/(x + y). The factor for converting the estimate of the batch per year is the ratio of the size of the unit to the size of the batch.

Revenue from layers was calculated from sales of eggs (net after spoilage), meat from culled bird, and manure. Revenue from broilers comes from sales of broiler meat and manure. The value of manure is calculated as the total value of manure used either by the farm or by others, valued at market prices.

Data on inputs are collected for various reference periods. Information for some inputs like regular workers was on a per year basis, while other inputs like casual workers or feed was collected from the latest batch. Information on feed was collected for different stages of the bird for the latest completed batch. Cost items collected from the latest batch were annualized.

Data for Dairy. Description of variables used in the analysis of dairy data is given in Chapter 7 of Annex III to this synthesis report. Expenditure on pollution abatement was measured in terms of imputed value of manure, annualized expenditure on manure storage sheds, cost of transportation of manure from production to end-use point, cost of spreading manure in the field, cost of making dung cakes (if used as fuel), and other taxes and fees. Expenditure per unit of milk was used as a proxy for the pollution abatement expenses.

Net returns/profit from milk production activity was calculated by taking the difference between gross returns and variable costs. Gross returns were derived by adding up revenue from main products and by-products, that is, milk production (quantity produced multiplied by the price of milk) and imputed value of manure. Variable costs included feed and fodder, hired labor, veterinary services and medicines, breeding and extension services, transport of inputs, maintenance of building and equipment, and other overhead costs.

The price of fodder was derived by taking the weighted average of market sale price paid by the farmer for different kinds of green fodder and dry fodder fed to cows and/or buffaloes. It was expressed in rupees per kg of fodder. The price of feed was estimated by calculating a weighted average of market prices paid by the farmers for various types of concentrates and other feed supplements fed to the animal. This variable was measured in rupees per kg of concentrate feed. The annualized value of cattle sheds, fodder storage sheds, milk cans, buckets, measuring instruments, chaff cutters, and other equipment used in milk production was derived by annualizing the value of buildings and equipment.

Since most of the farmers did not employ hired labor for milk production activities on their farms, it was difficult to get exact information on the wage rate. Moreover, there were large variations in wages given to the farm workers in the study area. Therefore, wage rates used in costing hired labor was the prevailing market wage rate in the study area. It is important to note that in certain cases, there were very small variations in the wage rates within a given sample. Wage rate was expressed in terms of rupees per day.

Milk production is an important activity for small and marginal farmers, and family labor is predominantly used for most of the dairy farming activities. Detailed information were collected from all sample households on human labor employment in various milk production activities, such as bringing fodder from fields, cleaning cattle sheds, milking animals, selling milk, etc. The average family labor (minutes) per unit of output was computed from the data.

In order to explain variations in net returns per unit of output and/or environmental pollution abatement behavior across different farm categories, variables were included as determinants of inefficiency. Such variables included age, education, experience, membership in organizations, number of livestock/agribusiness training programmes attended during the last two years, distance to market, etc. Dummy variables such as access of the household to information, market, credit, technology, land tenure system, etc. were also included to capture the impact of transactions costs on the profitability of the dairy farm.

5.7.3 The Philippines Data

Households or production units were given identification numbers. Variables to be used in the data analysis were identified and set in advance through an inception workshop held in August 2002. The period of coverage was also pre-determined (length of cycle or 1-month production period). Variables were classified as household characteristics or production/farm-related characteristics, and production activities were classified into types for hog/broiler production, depending on type of output and technology of operations (inputs used). The variables used and their descriptions are listed in Appendix Tables 4.1 and 4.2 of Annex I to this synthesis report.

5.7.4 Thailand Data

A complete description of variables used for swine, broiler, layer, and dairy can be found in Chapters 8, 9, 10, and 11, respectively, of Annex IV to this synthesis report. The pollution abatement cost variable is defined as the marginal cost spent on abatement effort and the fixed costs of equipment and capital used for pollution control such as water treatment and drainage systems, biogas, manure storage facilities, incinerator, and other machines invested for the purpose of mitigating environmental externalities. It also includes labor cost and value of manure disposed. The explanatory variables of the environmental mitigation regression include operator's characteristics, proxies for transaction costs such as access to assets and information, farm characteristics, and locational characteristics.

Net profit is likewise defined as revenue from livestock products (swine, layer, broiler, and milk), both direct and by-product income, less variable cost. Revenue is calculated from sales of output and manure, and is normalized per unit of output. Variables with missing values were adjusted by substituting them with mean values of non-missing samples drawn from the same province.

Previous Page Top of Page Next Page