Survey Strategies

Rarely are postfarm activities given the degree of attention accorded the farm sector. Few developing countries have transportation ministries or ministries concerned exclusively with agricultural processing. Secondary data for costs and returns are thus scarce. The analyst will be forced to visit a wide range of government ministries and academic/ research institutions to seek out specialists with information about postfarm activities. Central statistical agencies may carry out annual or occasional censuses of manufacturing that include firms processing agricultural commodities. Ministries of agriculture may devote some attention to postfarm activities of the most important commodities. But in most cases, secondary data will not be sufficiently detailed to allow the construction of budgets.

Fortunately, limited primary surveys are usually feasible. Firms engaged in postfarm activities are much fewer in number than are farmers. Marketing firms are easily located (a necessary feature of their business), and larger firms often maintain written records of expenditures and receipts. Moreover, the inputs used in most postfarm activities (except processing) are relatively small in number. Survey efforts can focus on the activities that are most important in total postfarm costs and can summarize the less important activities with budgets of only two or three items.

Surveys of Small-Scale Firms

Direct field interviews are the most common way to compile budget data for small-scale transport or processing activities. Written records of inputs and outputs exist only rarely. Small-scale firms are usually owner operated and often utilize makeshift capital equipment. If these items are not available on markets, their current market-equivalent value-the amount the activities would pay if the input were to be purchased-needs to be determined. If capital equipment items (such as the principal machinery) are purchased, the activity operator might not be aware of current prices but might know the useful life and salvage values. Current prices must then be found through equipment suppliers or processors and marketing agents who have recently built or retooled their operations. New businesses can also be good sources of information about replacement costs for buildings and other infrastructural capital. If current prices cannot be obtained, lastresort approximations are generated by application of an index of inflation (producer prices or wholesale prices) to historical purchase prices.

Data collection for variable input use requires several strategies, particularly when written records are not kept. Because annual use of inputs such as fuel, lubricants, and labor usually cannot be recalled with accuracy, such data are most easily collected on a per day basis, a per hour operation of machinery basis, or a per unit of output basis. Eventually, all information must be converted into a common numeraire and a common time frame. But primary data collection should include whatever numeraires and time frames yield the most accurate responses. In each case, conversion ratios will be necessary to convert costs and returns to an annual or per unit output basis.

The final category of data involves outputs. Annual throughput might be well known by the activity operator, and this information is essential to determine fixed input costs per unit of output. Alternatively, the activity operator can be asked to describe output activity during an average week or month; in this circumstance, information about the number of weeks or months of operation per year is necessary as well. Actual or imputed market values of secondary outputs must be included along with primary outputs. In many small firms, output prices will be unavailable, because the activity operates largely on a custom basis and neither buys nor sells the commodity. This practice is recognized in the structure of the activity budgets by the assumption that the processor buys the commodity from the farm-to-processor activity. The custom processing fee is then added to the simulated commodity purchase price to obtain the simulated sales price for the output of the processing activity.

Surveys of Large-Scale Firms

Investigations of large-scale marketing operations amount to industrial firm surveys. These investigations almost always focus on processing. Because record keeping is quite detailed, surveys usually can obtain

precise estimates of costs and returns. Some of the data necessary for the development of representative budgets can be gathered from annual financial reports. These statements provide measures of outputs, stocks, and raw material (commodity-in-process) inputs. But other inputs will not be described in sufficient detail for the PAM. Even reported capital equipment values may be of limited use. Accounting statements usually reflect some depreciated value based on the historical purchase price of the fixed input, and this reported value might be an unreliable indicator of replacement costs. Tax laws also can alter the ways that depreciation and book values are reported. Data from the country's industrial census might be helpful if they are reliable and up to date, but direct interviews and questionnaires are generally needed.

Because industrial technologies are highly complex, site visits and trials of test questionnaires will be needed to develop a clear understanding of input-output relationships. The actual survey can be completed in a relatively brief time period, because the number of large-scale firms is usually small. Screening firms during the initial site visits can save time for the research effort by allowing a focus on representative firms that have accessible and high-quality information. If the analyst is concerned about bias in results, a random sample can be chosen. But differences in responsiveness mean that the budget constructed will be based ultimately on only a subset of this sample. Box 10.2 illustrates a typical questionnaire format used for a study of wheat flour mills in Portugal.

A particular base year is desired for the study, although some flexibility might be needed to accommodate the accounting procedures of firms. If the study is performed on a calendar year basis but the accounting year involves some different twelve-month period, data are requested for the year closest to the base year. When annual inflation rates are very high, budgets based on accounting year data must be adjusted. But in most cases, the accounting year estimates will not need to be changed much in order to approximate results for the calendar year.

Guarantees of confidentiality are also important for successful data collection. When asked for details about costs, revenues, and profits, firms are understandably sensitive about how that information will be used. But because activity budgets are built for representative firms, no individual firm's practices will be revealed in published results. Discussions with top-level management and presentation of copies of research project proposals during initial site visits enhance effective information collection.

Once the data for inputs and outputs are collected, the questionnaires are combined to generate a synthetic representative budget. Because

firms differ in capacity and level of annual throughput, no two firms will report identical numbers for input use and output. The first task for budget preparation thus is to group the firms into size categories, according to processing technology. The measures for size categories are usually related to hourly processing capacity, such as number of metric tons per hour. Choices about the range of capacities to include within a single group depend on the presence of economies of scale. If a flour mill with a 4 metric ton per hour capacity is roughly the same as two 2 metric ton per hour capacity mills, both types of firms can be grouped together.

The choice of representative capacity is made in concert with the valuation of fixed capital costs. Capacity can be an average of the capacities of the sampled firms, but the availability of information on fixed investment costs often requires more pragmatic choices. Because firms within a grouping will have plants of different ages, replacement costs for fixed inputs might not be estimable from questionnaire results. The analyst will then be forced to use capital stock evaluations from the more recently established firms (adjusted for inflation, if necessary). Discussions with engineering and construction firms can also be helpful to verify the representativeness of particular fixed capital cost estimates.

Evaluations of labor, intermediate input, and raw material input requirements can be determined by comparison of the responses across firms within the sample. The input quantities and values can be standardized by expression of the reported results per unit of raw material input or per unit of processed product output. When these variable inputs are used in constant proportions, the standardized estimates should be comparable across firms in the sample. But if the standardized estimates are very different, some causal explanation must be sought. The sample is reclassified and multiple budgets are developed to reflect differences among firms in production technology.

Differences in shift work across firms often are important to the size and cost of the labor force. Administrators and many skilled employees often do not work during second- and third-shift operation. High wages and overtime payments might be required for those who work second and third shifts and for weekend and holiday work. The analyst has to choose between simple averages of results across firms and specific responses in shaping the budget of the representative firm. In either case, choices about shift work must be consistent with the selected rate of operation.

Partial Budgeting

Once budgets are constructed, partial budgeting can be used to evaluate the impact of additions to or substitutions of equipment in the existing technology and alterations in the rate of capacity utilization of equipment. Additions and substitutions may be constrained by the technical compatibility of components. Often, technological change in processing industries requires complete replacement of existing equipment. But in some cases, particularly with storage and handling facilities, components of the processing firm may be operationally separable from one another. Interviews with engineers and processing technicians can be used to verify the viability of new budgets.

Changes in capacity utilization may be feasible for several reasons. Monopoly power among processors, government-induced distortions in hours of operation, and the costs of capital equipment can create conditions where operation times are well below technically feasible levels. Capacity utilization can also be influenced by the costs of collecting raw materials and distributing the processed product. A reduction in these costs, either by elimination of distortions or by introduction of a new infrastructure, may allow processing firms to expand their markets. Because fixed costs are prominent in total costs of processing, reduced utilization rates can cause observed processing margins to be substantially larger than their cost-minimizing levels. It is also possible that entirely different processing technologies would be chosen under more intensive utilization. If the analyst is examining a sufficient number of representative processing systems, this possibility can be evaluated by comparison of the costs of alternative techniques, each operating at full capacity utilization.

The maximum feasible capacity utilization rate will almost always be less than 100 percent. Equipment downtime is often mandated by maintenance and repair requirements, social constraints on operating at certain periods, or limitations of local market demand. Interviews with technicians and other industry experts are necessary before the maximum capacity utilization rate can be chosen. Introduction of an adjusted capacity utilization rate will influence the fixed input costs of the processing budget. Some variable input adjustments might be required as well; as mentioned, higher wages usually are required for night and weekend operation. But most variable inputs will vary in direct proportion to output. Because the budgets are already calculated in terms of costs and revenues per unit of output (or per unit of raw material input), the greater part of the original representative budget can be directly transferred to the budget that simulates costs and returns under higher rates of capacity utilization. Box 10.3 describes a partial budgeting analysis of full capacity utilization for flour mills in Portugal.