Collection of Input and Output Data

Agricultural production is characterized by large numbers of firms at dispersed locations. In most cases, farms lack formal records of input use, particularly with regard to individual crops. Output records are somewhat more common, but usually this information is not expressed in the yield measures needed for economic analysis. As a result, primary farm surveys are expensive and time-consuming and place heavy demands on skilled manpower for monitoring and evaluating the survey data. In PAM-related work, the constraints of time and financial support for research usually mean that primary farm surveys are not possible. Instead, the analyst relies on secondary data in the preparation of representative farm budgets. Fieldwork remains critical to the construction of the PAM, but efforts focus on the verification of secondary data, the collection of information about current prices, and the introduction of modifications of input-output relations to account for technological change.

In most circumstances, prior data on farm budgets are available. The ministry of agriculture, producer organizations, and university researchers in agricultural economics often produce farm budgets, and their surveys can provide estimates of input and output quantities. Agricultural investment project proposals require economic feasibility analyses; estimates of farm-level costs and returns are usually included in this work. Extension service personnel might also have useful information about the input and output quantity requirements for particular commodity systems. If this information is not recorded in reports, it can usually be collected through visits to the agent. Finally, studies of comparable technologies in neighboring countries sometimes provide useful farm budgets.

Whatever the source of budget information, fieldwork usually begins with interviews of the employees who originally prepared the budgets. Such interviews are useful to disaggregate information about costs and returns beyond the level provided in published documents, to assess the extent of heterogeneity of production practices and the need for multiple budgets, and to gain initial impressions about the price and quantity effects of particular policy distortions or market failures faced by producers. Field informants might also arrange interviews with farmers and other informed observers of the local agricultural economy, such as providers of input or marketing services. Interviews provide supplementary information about prices and the efficiency of various input and

output markets. They can also cover the particular policy issues that motivate the research. Because the selection of expert informants is not random, care must be exercised in using responses to characterize the various representative systems. But this approach has the advantage of confining fieldwork to several weeks rather than many months.

Quantities

Output data (crop yields or animal productivity) may be obtainable from ministry statistical branches responsible for national production estimates. If these data are available as a regional time series, the analyst can obtain useful estimates of normal yields. Because yields reflect economically influenced levels of input use and agronomically influenced varietal performance, care must be taken in designating particular yields as normal. Similar considerations require caution in the use of experiment station data. Under experimental conditions, the profitability of production is usually irrelevant and cultural practices that maximize yield rather than economic efficiency are the norm. Experiment station yields thus commonly overestimate on-farm yield levels. Associating these yields with estimates of on-farm costs causes profitability estimates to be excessive as well.

Experimental data can be useful, however, in estimation of the relative advantage of a new variety or cultural practice. Experimental plots are often used to compare new practices to the traditional practice in a control plot. The application of relative premia to actual yield data from farm surveys gives an estimate of the expected on-farm yields from the improved practice. This calculation presumes that the new technology will exert similar effects on both control plots and actual farms. If the control plot and actual farm yields are very different, the presumption may be erroneous. In this circumstance, only replication of on-farm practice can indicate the likely yield benefits from new technologies.

Different sources of secondary information on input and output quantities almost always show some differences in estimates. Comparisons of secondary data can be assisted by use of cropping calendars, which list the alternative estimates of input and output quantities for each farming task. Comparisons with other information sources, such as crop yield surveys, and the results of field interviews can then be used to make judgments about the quality of information from each source.

If the differences among estimates reflect variations in survey quality (caused, for example, by small sample size or careless survey design), the poorer-quality estimate can be disregarded. But estimates may also differ because the commodity systems are not the same. If the differences in estimates arise because of variations in local economic conditions or technologies, the description of the representative commodity system must be made more explicit. For example, if comparisons of fertilizer use estimates reveal one high estimate and one low estimate, these differences might be explained by differences in the sizes of farms sampled in the two surveys. Explicit decisions then need to be made about farm size in the description of the commodity system.

From careful comparisons of secondary information sources, a synthetic representative budget is constructed. This budget may use different sources of information for quantity estimates of each particular input and output. For example, one study could provide a particularly convincing estimation of average yields; another might be judged superior for its measurement of direct labor inputs; yet another could be the source of the most accurate measurement of water use and irrigation practices. The chief danger of synthetic constructions is that estimates from different technologies may be unwittingly blended to create a budget that is not representative at all. A second problem for synthetic budgets occurs when input measures are not consistent with output measures. The "best" estimate of fertilizer use might come from a study that showed relatively high yield estimates. Combining the fertilizer estimates with national or regional average yield data might result in a "representative" budget that overestimates fertilizer input relative to output. Profitability in the PAM calculations would be underestimated. Again, field visits and consultation with expert observers become necessary to verify consistency among the input and output measures of the representative budgets.

If secondary data for input and output quantities are absent, PAM analysis usually is not possible. Research resources must be devoted to the collection of such information rather than the compilation of budgets. But even in these circumstances, the analyst might be able to construct a representative commodity system with relatively little primary survey work. In this approach, secondary input data from available commodity systems are used as benchmarks for the estimation of input requirements of other commodity systems. Interviewers ask farmers or other experts for information about labor utilization and intermediate input use relative to the requirements of alternative commodities that have well-understood input-output relationships, assuming a plot of equal size for each commodity. This information is relatively easy to collect. With it, one can apply appropriate discounts or premia to the estimates from the alternative crops, providing a budget for a new commodity.

Prices

An equally important aspect of farm-level fieldwork involves collection of prices for inputs and outputs. Secondary data sources often provide some price data. Statistical offices frequently collect annual market prices of principal agricultural outputs, and secondary sources of budget data contain price data for inputs and outputs. A problem with direct use of these data arises when the base year for PAM analysis differs from the year used for the data from secondary sources. In addition, these prices might not represent expected market prices but instead might be the outcomes of peculiar demand and supply conditions.

For some inputs and outputs, market prices will not exist because the product is produced and consumed exclusively on the farm. These situations are particularly common in subsistence-oriented areas, where inputs such as manure and forages might never be traded on markets. In this situation, one needs a market-equivalent value for the product-the price at which the product would sell if a market existed. In many circumstances, this valuation can be based on comparison with a substitute commodity sold through markets. For example, animal feeds can substitute for forages, and the number of feed units contained in forage can be calculated and evaluated at the market price of a feed unit for animal feeds. Because substitution is rarely perfect-for example, animal feeds might not contain the roughage provided by forages-the search for market-equivalent values will often be an exercise in approximation. When substitute inputs are not available, market-equivalent values have to be estimated on the basis of the labor, capital, and intermediate inputs required to produce the input. The total costs of these inputs are assumed to reflect an implicit market price for the product.

Perhaps the most common nonmarketed input is family labor. Instead of receiving a wage payment, a family laborer shares in the net income of the farm. Each family member receives an implicit wage equal to the value of individual consumption and savings divided by the time devoted to the farm activity. Makers of budgets usually avoid such calculations by applying market wages to all labor inputs. If family labor does not earn the market wage (private profit is negative), at least some family workers could do better financially by leaving their own farms and seeking employment as hired laborers. The analyst then needs to develop a rationale for acceptance of a relatively low rate of remuneration, such as limited alternative employment opportunities or a desire for food security and a consequent unwillingness to rely on markets for basic foodstuffs.

This treatment is not entirely satisfactory. As discussed in earlier chapters, implicit wages would ideally reflect private marginal products, and divergences of the sort just described would become part of the explanation for differences between private and social costs of labor. But because family labor wages cannot be observed, market wages become a necessary substitute.

The determination of private market wages can be a complicated task. Nonmonetary incentives, such as meals or drinks, are often provided by employers. Because these items are a cost to the activity, the market-equivalent values of nonmonetary incentives are included in the calculation of labor wage rates. Market wages also reflect differences among family members in skill level, sex, and age, making it unlikely that a single wage rate will apply to all the labor inputs described in the budget. During slack seasons of the crop production cycles, wages might fall to a subsistence level or, in the event of a total absence of labor demand, temporarily to zero.