Agriculture is the basic source of food, and farmers are the basic food producers. Farmers are remarkably diverse people, ranging from near-subsistence peasants in India and Guatemala to corporate businessmen in California and Sao Paulo. Nevertheless, private agriculture is a markedly homogeneous tndustry in the kinds of decisions that must be made day in and day out and in the kinds of uncertainties that condition those decisions. The corporate soybean farm in Sa~o Paulo or the rice farm in California has more in common with the wheat-growing peasant operation in the Punjab than with U.S. Steel or Volkswagen of Brazil.
In a substantial part of the world, agricultural decisions are made within a collective environment-from North Korea through China to Vietnam and in Eastern Europe and the U.S.S.R. Perhaps half the world's farm households are part of collectivized or communal agriculture, and yet these households, like their private counterparts, must still make many decisions that are not made by higher authorities. Much of farmers' daily work is done at their own initiative, and the incentives that induce them to work in a timely and careful fashion strongly influence the quality and quantity of agricultural output. In both private and collective agricultures the decisionmaking environment is conditioned by incentives to work. Identifying the factors that influence the size and composition of agricultural output is impossible without an understanding of the decisionmaking environment of the farm household.
This chapter addresses four questions about the food production system. First, what are the objectives for the sector itself, as opposed to the broader objective of providing food to meet consumption requirements? Answering this question involves understanding why the agricultural sector is different from the steel or transportation industry and what the social and analytical issues are that flow from these important differences.
Second, how do farmers make decisions? Only with a decisionmaking framework that incorporates the full range of factors influencing farm households is it possible to address the behavior and performance of the food sector as a whole. Most farm households are characterized by joint consumption-production decisions, but not in a tightly defined subsistence setting. Farm households base their consumption and production decisions on farm input prices, cash and food crop output prices, the prices of consumer goods from the market, the opportunity cost of their members' time either in outside labor markets or on farm production (including household work), and demand for leisure. The full context of household decisionmaking is essential to understanding how food production will change when external circumstances change.
Third, what government interventions are available to change household decisionmaking and thereby change the performance of the food producing sector? Understanding how interventions will affect decisionmaking is more important for agriculture than for any other sector, for the government has very few interventions available that can directly alter domestic food production. State-run farms and public exhortations to farmers to step up output still must deal with the reality of millions of day-to-day decisions in planting, tending, and bringing in the crops. An old saying holds that governments do not grow food; only farmers grow food. But governments can import food, subsidize fertilizer, make agricultural research a priority, or "purchase" food surpluses at gunpoint. For better or worse, the farmer's fate is tied to government policy, and the government's fate, or at least the success of its food production plans, hinges on the willingness of farmers to go along.
Fourth, what are the elements of a successful agricultural development strategy? Of the wide variety of possible government interventions, what combination will reinforce the achievement of sectoral goals and simultaneously serve the broader set of food policy objectives? This question raises somewhat different issues from those of traditional output-oriented agricultural development literature, for the food policy analyst is concerned with the intersectoral and consumption consequences of a production strategy as well as the impact on yields.
What does society want from its food producing sector? The answers used to be cheap food and cheap labor to foster industrial development and the earning of foreign exchange to buy the capital machinery to make it possible. A host of new complexities, however, has been added. A concern for rural poverty, unstable world markets, and the importance of efficient growth in the rural economy as well as in industry now make the question much more difficult to answer. A major lesson of postwar development experience is that agriculture is site specific. What works in one location may not work in another, even in the same country, because the ecological setting is different or because farm households face different constraints on their decisionmaking. The food production issues important to the policy analyst begin with understanding why agriculture as a sector is so different from other industries and why agriculture itself is so heterogeneous from farm to farm and even from field to field.
Five features set apart the agricultural sector from other productive sectors of an economy: its large contribution to national income, the large number of participants, the peculiarities of the agricultural production function, the role of the agricultural sector as a resource reservoir, and the importance of home consumption of output. These features are more evident in traditional societies, and their distinctiveness tends to erode during the process of economic modernization. Indeed, perhaps the most striking characteristic of agriculture is its almost universal tendency to diminish in importance relative to other, more rapidly growing sectors of the economy. There are healthy and unhealthy declines, however. An urban-biased development strategy that neglects agricultural investments and incentives can force an agricultural decline. The alternative path-rapid growth in both agricultural and industrial sectors-leads to a relative shift in the agricultural sector's importance because citizens with higher incomes consume relatively less agricultural produce. Finding the path that fosters growth in agriculture and industry is the goal of the analysis here.
The Size of Agriculture in Gross National Product
A large proportion of economic activity is provided by agriculture in most poor societies. The agricultural sector contributes as much as 70 percent of national product in a few countries just emerging from centuries of traditional economic organization. Half the output in many developing countries is still produced in agriculture. If related industries are also counted in, as these develop rapidly in the course of modernizing agriculture itself, the share of this broader agribusiness sector seldom declines to less than one-quarter of national economic output, even in advanced industrial societies. In very few societies do consumers spend less than one-fifth of their incomes on food alone. If other agricultural activities-the input industries and the production of industrial raw materials-are also added in, the continued importance of agriculture is obvious.
When agriculture contributes half or more of gross national product, rapid growth in average per capita incomes is very difficult to achieve unless rural incomes are rising. From a macroeconomic perspective in poor countries, rapid and efficient increases in agricultural output will be essential to meeting overall growth goals. From a growth perspective, simple arithmetic makes agriculture more important than other sectors.
Number of Participants
In many countries, 60 to 80 percent of the population still lives in rural areas, earning a livelihood directly or indirectly from agriculture. In industrially advanced economies many of these people move to the industrial sector while still engaged in agriculturally related jobs-producing fertilizer, canning tomatoes, or stocking supermarket shelves. But in nearly all developing countries a majority of the population lives in the countryside. The overwhelming predominance of the rural population has three important consequences for understanding agricultural decisionmaking: most farms will be small because large numbers of people must share the arable land; millions of individuals will each behave according to particular decisionmaking environments; and much of the world's poverty and its human welfare costs will be in rural areas.
In most countries, if the available arable land were divided equally among the farm population, the resulting average farm size would be small by comparison with U.S. or European standards. Farms of less than a hectare would characterize China, Bangladesh, and Java, and even Japanese farms average only slightly more than a hectare. The average in India would be about i to 2 hectares, and in Africa and Latin America farms would tend to be less than 20 hectares. Average farm size is well over 100 hectares in the United States and well over 50 hectares in the United Kingdom.
The available farmland, of course, is usually not divided equally among all the potential farmers. The conditions of land tenure and the size distribution of farms are important characteristics of a country's agricultural decisionmaking environment. Although exact farm-size distribution is a subject for analysis in each country, its general pattern is important in judging the likely degree of poverty and the income distribution consequences of growth strategies for the rural sector.
A country with a unimodal distribution of farm sizes-a large number of small, family-operated farms capable of supporting the family members above a subsistence level, with only a fringe of smaller and larger farms around this modal norm-has the potential to use agricultural development strategy to reduce rural poverty at the same time that it increases food production. Countries with bimodal distributions of farm sizes-many very small farms on a minority of the land with a few very large, estate-like farms that occupy most of the arable land and produce most of the food surplus available for urban markets-face much more difficult dilemmas in reducing rural poverty while using traditional output-increasing strategies of agricultural development. Such land tenure patterns are common in Latin America and are one reason that land reform issues are so much more prominent there than in Africa or Asia.
The circumstances under which farm households have access to land for growing crops have many ramifications beyond the obvious insecurity and commonly weak bargaining position of tenants and renters. Landownership provides an asset which farmers can use to obtain credit for inputs or investments in farm capital. It provides greater incentives to save from household income. The conditions of tenure frequently affect the willingness of landlords to invest in land improvements and the willingness of tenants to use yield-increasing inputs at socially efficient levels. The discussion of food production strategies engages this issue in the context of the broader objectives of food policy.
Growing food is a decision-intensive undertaking. What crops to plant, what inputs to use, when to plow, to seed, to cultivate, to irrigate, to harvest, how much to keep for home consumption, how much to sell and how much to store for later sale are the decisions that occupy the daily routine of most farmers. Agriculture is truly unique in that literally millions of individuals and households are making these decisions themselves or in consultation with relatively small numbers of neighbors, friends, or partners. In Brazil, India, Indonesia, Nigeria, and even China, influencing agricultural production decisions to increase food output is an entirely different process from changing decisions about how much steel or cement to produce. In each of the countries-indeed, in most countries-a dozen or so individuals could take direct action which would lead to a 10 percent increase in steel output in a year or so. Their decisions would be decisive.
Nowhere, not even in socialist countries, can a similar small group of individuals decide to raise food production by 10 percent. To be sure, a small group of planners or the president and the cabinet can decide they want food production to rise by 10 percent. They can tell the food logistics agency, the ministry of agriculture, the newspapers, and agriculture extension agents that they want food production to rise by 10 percent. But they cannot increase food production 10 percent by themselves. They must also convince the millions of farmers in their country to want to increase food production by 10 percent, and make it in their self-interest to do so.
Here is the true importance of the vast number of agricultural decision-makers. There are simply too many of them to reach directly either with pleas for cooperation or with police power. Farmers must see the benefits of higher yields for themselves; there are too many opportunities to let high yields slip beneath the hoe or in a late fertilizer application, even under the watchful eyes of a guardian. Farming is a subtle combination of skilled crafr and brute force. Brute force alone will not achieve high yields.
Farmers' decisions are likely to be altered only when they perceive the incentives to be favorable to the change. A heated and frequently sterile debate has been waged over the incentives needed to induce change in farmers. The elements range from pretty ribbons to raising political consciousness, from basic literacy to the availability of consumer goods for purchase in rural markets. The debate is nearly dead now, for the answer is largely in. Most farmers respond to opportunities to improve their economic and material well-being.
The evidence is overwhelming that farmers make economic calculations in considering their agricultural decisions. When the economic environment changes, their calculations change in directions predicted by economic models of producer behavior. Collective ownership of land and other implements, and collective decisionmaking with respect to basic cropping patterns and investments, can alter sharply the perception of risk from farming. If the returns are also shared collectively, the perception of reward for individual work and initiative is altered as well. To produce agricultural output efficiently, most socialist agricultural systems have found it necessary to maintain personal or household incentives that link farming effort to rewards.
The tendency toward economic rationality in farm household decision-making justifies the use of basic economic models to help analysts judge the efficacy of policy interventions designed to change the decisionmaking environment of rural households. Just as models of consumer decisionmaking with respect to food help organize analysts' research and policy design, so do producer models help organize the complexity of the farm environment into several issues that are central to food policy.
Of these policy issues, some are of special importance: the efficiency with which farmers allocate the resources at their disposal to produce crops, relative to alternative uses of these resources; the technical ability of farmers to achieve the maximum output from a given set of inputs; and the impact of alternative forms of land tenure on both the allocative and the technical performance of farmers. Each of these issues conditions the scope for effective government intervention. One of the policy levers most susceptible to effective government control is price policy for food crops and agricultural inputs. The role of prices in influencing the behavior of farmers is extremely important and depends on farmers' allocative and technical efficiency and on the form of tenure for the land they farm.
Characteristics of Agricultural Production Functions
The only way to produce output is to combine the necessary ingredients-the inputs or factors of production-in suitable proportions so that the overall process yields the desired product. One unusual feature of the agricultural production function-the technical relationship that specifies how much output will be produced from any particular combination of inputs-is the joint combination of labor and management. Knowing what the right inputs are, how to combine them, and how to tend the process is the major function of management. In farming, this management skill is frequently combinedwith the farm household's own labor power, which is also an important ingredient in growing food. Several other features contribute to the uniqueness of agricultural production functions. The most important are seasonality, geographical dispersion, risk and uncertainty, and sources of technical change.
No agricultural region of the world has an absolutely constant year-round climate. Winter and summer create distinct growing seasons in the temperate zones. Wet and dry seasons and monsoons create conditions in which planting is appropriate, harvesting is difficult, or some crops simply do not thrive. Climatic variations cause agricultural production to follow distinct seasonal patterns even in most tropical areas, but seasonality is not a fixed and rigid constraint. Rice will grow in the dry season if irrigation water is provided, and tomatoes will grow in Siberia in January under artificial lights in a warm greenhouse.
Seasonality is important to farmers because it is generally cheaper to let Mother Nature provide many of the essential inputs for agricultural production-solar energy, water, carbon dioxide, temperature control, and essential nutrients from natural soils. But it is not always economical to let nature dictate the agronomic environment. One of the major tasks of government policy is to invest in socially profitable interventions, such as irrigation and drainage, that increase farmers' control over the crops that can be grown in particular regions and seasons.
Seasonality also tends to place high premiums on the timely performance of such critical agricultural tasks as plowing, planting, cultivating, and harvesting. Even though the available labor pool may seem more than adequate to provide the required number of man-days per hectare over an entire year for all the crops being grown, significant labor bottlenecks may occur if certain tasks must be performed very quickly at specific times to ensure maximum yields. Such bottlenecks frequently induce individual farmers to mechanize specific tasks-plowing or harvesting-even when much rural unemployment exists over the course of the year. Furthermore, a tractor that pays for itself by timely plowing then has a very low marginal cost of operation for other tasks as well, and labor displacement can be much more widespread than would be indicated by the removal of the plowing bottleneck alone.
Three lessons are apparent. First, seasonal aspects of agricultural production frequently constrain yields because of input bottlenecks. Labor is most often the constraining factor, but fertilizer, seeds, credit, or irrigation water must also be available at specific times. When fertilizer reaches the village godown a month after the proper application time, it may as well not have arrived at all. Government authorities responsible for managing the distribution of agricultural inputs are frequently unaware of or insensitive to the extreme importance of the timely availability of inputs. Suppliers whose incomes depend on providing appropriate inputs to farmers when and where needed are much more responsive to shifts in weather, cropping patterns, and new technologies than are government agencies trying to allocate inputs available from a planned industrial sector. Modern agriculture that uses industrial inputs as the basis for high yields is a dynamic enterprise quite unlike factories. It requires smoothly functioning input and output markets for production in the sector to grow rapidly and efficiently.
Second, there are often very high private economic returns to eliminating seasonal bottlenecks in production. When these private returns are generated at least partly by higher and more stable yields of agricultural products, society is also likely to gain. But if the private gains come from displacing hired labor that has few alternative jobs, the social gains may be small or even negative.
The third lesson is the importance of viewing agricultural production in a seasonal context. Most agricultural data are published on an annual basis, and there is an inevitable tendency to think about the sector in terms of the same criteria used to evaluate the annual performance of the steel or cotton textile industry. Such an annual approach, characteristic of five-year plans, hides two important roles for government analysis and intervention-the appropriate provision of inputs when and where they are needed and the full analysis of the social impact of private investments to reduce seasonal bottlenecks in agricultural production.
Agriculture is the only major sector that uses the land surface as an essential input into its production function. Like seasonah.ty, this widespread use of land is due to the largesse of nature. It is simply cheaper to let farms capture the free solar energy and rain than to stack a hundred hydroponic "fieHs" on top of each other and provide the light, nutrients, and water from industrial sources. This wide geographical dispersion of agricultural production has an important economic consequence. Transportation becomes essential if any output is going to leave the farm for consumption by others or if inputs, such as modern seeds, fertilizer, pesticides, or machinery, are to be used on the farm to raise output.
In combination, seasonality and geographical dispersion create the need for a marketing system that can store the product from a short harvest period to the much longer consumption period and can move the commodity from the farm where it was grown to the many households where it will be consumed. Both functions require that the commodity change hands and that ownership be exchanged. This transaction can happen only when both parties agree on the terms of the exchange, that is, the price, for the commodity at the point of sale. In socialist economies the terms of exchange are usually set by the state, but all other marketing services must still be provided if the food grown by farmers is to be eaten by consumers. The role of marketing in price formation and the provision of food to consumers is the subject of the next chapter.
Farmers the world over talk primarily about two topics, the weather and prices. Cn these two variables ride the rewards for the whole year's effort in farming. A failed monsoon, a flood, or a hailstorm can wipe out the crop. A bumper harvest can cause large losses if the price falls too low. No other industry depends on the whims of nature and volatile markets as much as farming does. Farmers who repeatedly make good decisions in response to rapid changes in their economic environment tend to survive and thrive. Those who do not, frequently fail and move to urban areas in search of jobs. Or they become impoverished landless laborers dependent on the rural economy for their incomes and access to food. Socialist-managed agricultures can cushion much of the welfare shock to individuals by sharing risks, but rapid and effective decisionmaking remains the key to dynamic efficiency in agricultural systems.
The fact that weather is uncertain causes farmers to behave differently than they would if weather were always known. This general uncertainty usually leads farmers to choose crops that will resist the extremes of weather, particularly crop varieties that are more tolerant of weather variations, and lower levels of inputs than would be optimal in a predictable world because of the risk of losing the investment altogether. These individual farmer reactions to uncertainty spill over into the arena of policy concern, for the resulting crop mix and aggregate output might be quite unsatisfactory for meeting government goals.
Equally important, farmers' reactions to weather variations have consequences for aggregate output. A late monsoon may cause millet instead of wheat to be planted, good rains may permit a second or third rice crop, and high temperatures and humidity can bring pests and diseases that force farmers to change crop rotations. As each farmer reacts, the adjustments can spill over into rural labor markets and cause serious shortages if planting must be done suddenly when the weather breaks or the harvest brought in before a flood. A particularly "dry" dry season may mean the second crop is not planted or harvested, and an important, perhaps critical, source of wage income is eliminated for many rural workers. The reduced harvest may not be the most important consequence of such a crop failure. A famine could result because of the failed income opportunities.
At an aggregate level, weather-induced variations for staple crop output are frequently 5 percent above or below normal. For countries so small that erratic weather patterns affect all farming regions simultaneously, as in Sahelian Africa and Central America, variations of more than 20 percent from one year to the next have been recorded. Because food demand tends to be quite inelastic in the short run as people shift other budget expenditures to maintain adequate levels of food intake, even small variations in basic food output can cause large fluctuations in market prices unless governments have substantial buffer stocks available for price stabilization or can arrange for additional imports in a timely fashion. In socialist economies, the availability of food rations can be sharply curtailed if a crop fails and planners have not anticipated the need for additional supplies from alternative sources.
In addition, fluctuations in aggregate production are magnified at the level of marketings (produce available for consumption by nonfarm households) because farm household consumption tends to vary somewhat less than production. Consequently, marketings vary considerably more than production in economies where a significant share of food production is consumed directly by the farm household. In years of poor weather, net marketings decline proportionately more than production. Similarly, in good years the percentage increase in marketings is usually substantially larger than the percentage increase in production. These wide fluctuations simply add to the government S difficulty in stabilizing domestic food prices and provisioning urban areas. A tendency to use food imports for these purposes is certainly understandable, but it frequently discriminates against food producers.
Price uncertainty also adds to the farmer's difficulty in deciding what crops to grow and how many inputs to use in growing them. Unlike the handful of manufacturers in large-scale industries, farmers are unable to set their output prices and then adjust production and inventory levels to meet the price targets. Unlike consumers, who know with near certainty the price they must pay for a given quantity and quality of a commodity at the time they buy it, farmers must make major input purchase decisions well in advance of knowing what prices their resulting output will bring. At the time many key farming decisions are made-the allocation of land to various crops, fertilizer applications, hiring labor for weeding-the farmer can only guess at the prices for the output.
Reducing weather and price uncertainties is an important task for food policy interventions. Dams and drainage ditches can reduce the impact of rainfall variations, crop insurance can provide a guaranteed income floor even if heavy investments are wiped out, and research on more adaptable but still high-yielding plant varieties can reduce the risks of new technology. Similarly, a government can reduce price uncertainty by providing better price forecasting information, by using import and export policy to provide a band of prices within which domestic price formation can take place, or by implementing a more aggressive floor and ceiling price policy with a government-operated buffer stock program. But these efforts to stabilize prices must be visible in market operations, not just in press releases and legislative actions. Most farmers have learned by painful experience that simple statements of government intentions to stabilize prices-or even to require them by law-are ineffective.
Technical change is the source of most productivity growth in the long run, as continued investment in traditional technology very quickly faces low marginal returns. Most farmers are inveterate experimenters and tinkerers, always on the lookout for a slightly better way of doing things, whether a different seed spacing or a modified shape to the plow. As late as the 1920s most of the agricultural innovations in Europe and the United States arose on the farm and were gradually diffused by word of mouth and by agricultural colleges. Such on-farm innovation continues, but the scientific revolution in agriculture has made the process of technical innovation much more knowledge- and capital-intensive. Very few farmers even in the United States have the resources to carry out significant agricultural research, and most of it is now conducted by publicly funded agricultural research centers and by a handful of large agribusiness concerns, which are involved primarily in developing hybrid seed technology, chemical technology (herbicides and insecticides), and agricultural machinery. The small scale of most farms and limited financial resources mean little important agricultural research is conducted by farmers.
Technical change in agriculture shares many of the characteristics of technical change in other sectors, especially the tendency for individual inventors to be unable to capture the full economic rents from their inventions. The economic returns to innovation are small from the private inventor's point of view unless the new agricultural technology can be restricted for sale by its inventor or approved licensees. But the social returns to innovation may well be very large. Both the sheer scale of investment required for modern agricultural research and the inability of private research companies to capture the full return to their inventions mean that public agencies should play a leading role in funding and in carrying out agricultural research.
Diffusion of new technology is also a matter of policy concern, especially because not all farm households have equal access either to the knowledge to use new technology or to the agricultural and financial resources needed to make it productive on their own farms. Credit programs designed to improve the access of farmers, especially small farmers, to modern inputs are an essential component of the input programs themselves. Some inputs are "lumpy" and cannot be used efficiently on farms of even average size in many parts of the world. Large-scale tubewells and tractors might contribute significantly to higher productivity even on small farms if institutional arrangements can be found to separate the ownership of the assets from the service flows that such inputs can provide.
This public role can be overemphasized; the evidence suggests that truly profitable innovations spread quickly no matter what the government does. Where the entrepreneurship exists and the economic environment permits, rental arrangements and tractor-hire services frequently emerge spontaneously. But the location-specific nature of much new agricultural technology, especially seed technology, means that large areas of a country may be bypassed by the diffusion process unless government research and extension workers are actively engaged in the on-farm testing and evaluation of new technology. Adapting a general agricultural technology to a specific seed strain or technique that fits individual farming environments is a major responsibility of local research and extension stations.
An important policy concern is the impact of technical change on agricultural employment and rural income distribution. Historical evidence shows enormous variation in both the short-run and long-run impacts of innovations. The issues cannot be addressed satisfactorily by looking only at an individual farm. Because agricultural innovations tend to be embodied in inputs that must be provided through markets, they have complicated effects on the entire rural economy and eventually on the urban economy as well.
Most technical changes alter the biological processes of plants and animals to increase yield or they improve the efficiency of the mechanical functions needed to tend those biological processes. Primitive agriculture uses natural biological materials and processes in combination with human labor and management to bring in a food crop or livestock product. Modern agriculture uses scientific knowledge to reshape the biological materials so that each plant and animal is more productive, and it increasingly substitutes machines for human labor. Biological-chemical innovations, such as hybrid seeds, fertilizers, and pesticides, tend to increase yields and save land. Mechanical technology can also have a yield effect when it permits heavy soils to be cultivated or water to be pumped to dry lands, but most mechanical technology is designed to make agricultural work physically less burdensome and to save on the labor needed to produce a unit of output.
Yuj iro Hayami and Vernon Ruttan have shown that biological-chemical innovations have tended to be discovered and introduced in land-scarce, labor-rich societies, such as Japan and Western Europe, whereas mechanical innovations were developed and used in land-rich, labor-scarce societies, such as the United States, Canada, and Australia. Such "induced innovation" suggests that each society develops an agricultural technology appropriate to its resource endowments and agricultural needs. Whether such society-specific innovation will continue to yield appropriate results in the context of a much more interdependent international agricultural system is a prime question for the rest of the century.
Because most new agricultural technology is embodied in a physical input-a bag of fertilizer, a new seed variety, a tractor, or an irrigation pump it can be effective in a farmer's field only if a purchase (or rental arrangement) is made. Several consequences flow from this simple fact. For small farmers to participate in the benefits of technical change, they must be able not only to use the input on their farms (combines, for instance, usually are too large), but also to purchase the input that carries the new technology. Where a new seed-fertilizer package has a 200 percent rate of return, even borrowing from a village moneylender at 10 percent per month may be profitable. But for the full benefits of modern technology to reach small farmers, a credit program accessible to the farm household with only half a hectare or less may be essential.
Equally important, since new technology is embodied in inputs, a marketing and distribution system will be essential for both socialist and capitalist farmers to be able to purchase the inputs. Many traditional agricultural societies have a long history of small-scale marketing of surplus output to urban regions in exchange for consumer items, such as cloth, kerosene, or pots and pans, needed by farm households. There is no similar experience with large-scale movements of inputs, such as fertilizer or modern seeds, to those same dispersed farm households. The embodied nature of agricultural technology means that farmers cannot just be told about it. The marketing system must also deliver the inputs when needed.
Complementary fixed capital investments are ofren required to achieve the maximum benefits from the innovation. Usually this investment takes the form of better water control, land leveling, and drainage. Better control of seed bed preparation may sometimes require tractors with modern implements. Combines or threshers may be needed for faster and more sensitive harvesting techniques to avoid shattering and other harvesting losses. Shorter-maturity cereal varieties are often ready to harvest while the rainy season is still under way and solar drying is difficult or impossible. In such cases mechanical dryers and added storage capacity are essential.
The Farm Household as Both Producer and Consumer
Truly subsistence households produce to meet their own consumption needs and do not use the market for either buying or selling. To such households price signals are not only irrelevant, they are unseen. Few such households remain in today's world, not because farm families no longer consume produce from their own fields but because most farm families now buy and sell inputs and output in rural markets. They are aware of and react to market prices in making a wide variety of household decisions. But most farm households still retain some or most of their farm production for home consumption, which is a further distinguishing feature of the agricultural sector. Few steelworkers or even textile workers take their products home for household use.
The need to make connected production and consumption decisions within a single household obviously complicates life for the farm household, for the value of additional time spent in food preparation or tending the children must be balanced against the productivity of an additional hour weeding the rice, driving the ducks, or tending the home garden. The opportunity to spend some of that time working for cash on a neighbor's farm or in a rural wage-labor market places a lower bound on the value of household-farm time, and the value of leisure ultimately places a limit on the willingness to work, especially at low-productivity tasks. But for households with inadequate land to grow surplus crops for sale and with limited outside employment opportunities, the marginal value of leisure time may be low indeed, and possibly near zero. Even tiny increments to output can be valuable for very poor households.
The importance of joint household-farm decisionmaking also raises complex questions for analysts in search of ways to organize data and research issues into manageable and comprehensible frameworks for analysis. These complex questions have recently become the focus of a revived interest in models of household economies. At one level, the "new household economics provides a powerful insight into joint decisionmaking about food production, food consumption, investment in human capital, and even fertility and other demographic factors. By showing how all these decisions are related to each other and to the economic environment surrounding the household, the household economics models provide analysts with a conceptual understanding of the complicated lives that rural people live.
At the level of full empirical specification, however, the household economics models have so far not been able to provide more than a hint of the quantitative significance of the internal decisionmaking relationships. This shortcoming is partly because precise data on actual time allocations within households are difficult to obtain, just as food distribution among family members is difficult to determine without having the observer bias the distribution itself. More important, judging the real opportunity cost of time is both conceptually and empirically difficult because the true value lies within the mind of the decisionmaker. Knowing whether the possibility of entering the wage labor market really influences the amount of time a mother allocates to raising children or the time family members spend in the fields and gardens is critical to using household economics models. But such knowledge may not be attainable. This question of the real opportunity cost of time arises several times in this chapter because of household labor's important role in agricultural production; it is a main factor in understanding how farm households respond to economic incentives and what their costs of production are when responding.
Agriculture as a Resource Reservoir
Much of the early literature on agricultural development was based on strategies that saw modern industry as the cutting edge of the economic growth process. In this context, agriculture served a relatively passive role of resource reservoir to be tapped as industrial needs required. Virtually all early models identified agriculture as the traditional sector that housed surplus labor which could be moved to higher-productivity industrial jobs at constant real wages as capital investment created a demand for their services. More sophisticated and historical models looked to agriculture to provide food surpluses for urban workers, capital surpluses to be siphoned into industrial investment, and an "expenditure" surplus that permitted purchases of output from the industrial sector. Open economy models also focused on agriculture's role in earning foreign exchange so that the modern sector could import capital goods.
These images of agricultural surpluses to be used by the industrial sector die hard. In a dynamic setting, where the agricultural sector itself is participating in rapid and efficient growth, many of the transfers are possible and desirable. But in the context of a static and traditional agriculture, such exploitation models lead to both agricultural and industrial stagnation. Arthur Lewis argued that agricultural and industrial revolutions always go together. From that perspective agriculture does play a unique role in providing resources for economic development. A healthy rural economy creates productive employment for a large population that might otherwise seek jobs in the overcrowded cities, while it provides opportunities for investing in new technology with some of the highest returns available in any sector.
Because of agriculture's extraordinary diversity and heterogeneity of decisions required daily from farm to farm and in the entire marketing system, the sector is unique among major productive activities. This diversity places a heavy premium on decentralized decisionmaking. Planning agencies are simply incapable of making the necessary decisions efficiently and rapidly. Attempts to do so have stifled agricultural productivity in a number of countries, especially the socialist economies that have attempted to incorporate their agricultural sectors into the framework of central planning. As noted earlier, collective ownership and decisionmaking offer important gains in some areas of rural life, especially reduction of risk for individual households and more equal distribution of assets and incomes. In both socialist and market systems, however, many decisions that affect farm yields and the productivity of inputs must be made on the spot, day in and day out, by the individuals actually performing the work. The pressures and incentives these farmers face to make the decisions efficiently vary widely according to the type of economic and social structure and the agricultural policies pursued. Given that wide variation, it is important for analysts and policymakers to understand how farm-level decisions are likely to be made in a given context and how they will change when the structure and policies change. This section explains the nature of the production decisions that need to be made and the choices of individuals in the agricultural sector as they work to improve their personal or household returns from farming.
With an understanding of the features that make agriculture a unique sector, analysts are ready to address the basic production decisions farmers must make to function effectively year in and year out: what crops to produce, what combination of inputs to use to produce them, and what total output to produce. These decisions are related to each other in an economic decisionmaking framework that provides a rationale for farmer response to changed incentives. This section focuses on each of the decisions individually and then combines all three to relate output decisions to changes in output (or input) prices in order to construct a supply curve. The supply curve is a very convenient conceptual and empirical tool which summarizes a great deal of complicated producer decisionmaking in a simple two-dimensional diagram. In combination with the consumer demand curve for the same commodity, the supply curve is an essential tool in economists' understanding of price formation in market economies, one of the topics dealt with in chapter 4.
From an often wide array of possible crops, farmers must decide what commodities to produce. Except for perennial tree crops and pastoral livestock systems, these choices about which products to grow-often called product-product decisions-are faced annually, sometimes even monthly, by farmers. To make such choices in a rational fashion, farmers must assess the opportunity cost of growing more of one crop at the expense of another.
The production possibilities available to a farmer are depicted graphically in figure 3-1. A production possibility curve, LOFA, is drawn to show the various combinations of two crops that are technically possible for a farm household to grow using its available resources on a given plot of land in a single season. In this example, a farmer could choose to grow only beans (OA kilograms) and no corn. At the other extreme the choice could be to grow only corn (OL kilograms) and no beans. Point F represents a farmer's decision to grow some of each, OD kilograms of beans and OE kilograms of corn.
A rational and knowledgeable farmer would consider only points actually on the production possibility curve. A point such as K, which is inside the curve, represents an output level substantially less than the available farm resources could produce. It is not uncommon, however, to observe a farmer operating at an interior point such as K. The reasons might include bad weather or a pest infestation, lack of knowledge about appropriate production techniques, or an experimental new technique that failed. Understanding why some farmers are not on the production frontier is one step in determining the constraints on expanding output.
The production possibilities are shown as a curve, rather than a straight line, because the farm household's resources cannot produce corn and beans equally well. If the two crops were perfect substitutes, the production p05 sibility curve would be a straight line. The greater the curvature, the less easy it is to substitute one crop for another. Nearly all crops are substitutable for others to some degree if suitable investments are made in providing an appropriate growing environment. Whether such investments should be made for a particular crop is a crucial issue for agricultural policy. If self-sufficiency
in corn is an important objective, it is possible to tear out rubber trees or tea bushes and plant corn. The decision to do so is only partly an agronomic issue. Policy and economic incentives are often the determining factors.
A farmer represented in figure 3-1 who wants to increase the production of corn, from point F to G, that is, by ~C, must give up ~B units of beans. This opportunity cost is shown by the slope of the production possibility curve. As drawn in the figure, this slope is measured in physical units (such as bags, bushels, or kilograms). But farmers want to know the relative values they gain and give up, not weights. For this comparison, they need to know unit prices for the output. Whether the farmer's decision to grow more corn and fewer beans produces greater revenue can be determined only by comparing the value of bean output forgone (~B.PB) with the additional revenue expected by growing more corn (~C.PC), where PB and PC are the sale prices of beans and corn, respectively. If the gain exceeds the loss (if PC . ~C > PB . ~B), the farmer will find point G more profitable.
The combination of corn and bean output that maximizes revenue 15 tangent to the highest possible isorevenue line. This line is shown as MN in figure 3 - 1. It represents the combined value of output of both corn and beans (PB . B + BC . C). Along this isorevenue line the total revenue is constant. The farmer prefers higher isorevenue lines to lower ones but is equally content with any position along a particular line (in the absence of differing variable costs and assuming the same risks for each crop).
The slope of this isorevenue line is - PB/PC and represents the rate at which corn can be exchanged for beans in the market. This property is identical to the price relationship facing consumers with a given income, or budget, constraint. In fact, for the simple decisionmaking environment illustrated here, line MN is the farm household's budget constraint as well as its maximum revenue possibility. When this isorevenue line and the production possibility frontier are tangent, as at G, the slopes of both lines are equal. The production possibility frontier represents the physical tradeoff between corn and beans, or ~C/~B, and the isorevenue line represents the monetary tradeoff, or - PB/PC. These two ratios must be equal when the two lines are tangent. At this point, G in figure 3-1, the revenue of the production lost is just equal to the revenue gained because PC . ~C = PB . ~B. The marginal cost equals the marginal revenue, the standard economic criterion for maximizing profit.
The equality of the two slopes also has implications for consumer welfare. If the farm household were to choose between corn and beans for its home consumption, the highest indifference curve it could reach would be tangent to MN, its income constraint. The rate of commodity substitution in consumer decisionmaking is thus exactly the same as the rate of substitution in production (if marketing costs are ignored). No reallocation of resources in production or consumption can improve on this result without lowering output or welfare in some other part of the economy. Such a result is said to be a Pareto Optimum.
Relative prices are clearly a major factor determining important decisions of both consumers and producers. Since government policy frequently uses import or export controls to alter relative commodity prices, as well as exchange rate policy to alter the level of prices, it is apparent that farm-level decisions about how much of which crops to grow can be directly influenced by such indirect government interventions. Extension agents may be urging farmers to grow more corn, but if government price policy favors beans, many farmers will ignore the advice.
When the farmer has decided which crops to grow, the next decision is how to grow them. To a significant extent farmers can use varying combinations of factors of production, or inputs, to produce a given crop. When the inputs are labor and capital, these factor-factor decisions have important consequences for employment and income distribution in rural areas. The extent to which labor and capital might substitute for each other in the agricultural production process is represented graphically in figure 3-2. The curved line DGBA represents all the different combinations of labor and capital that could be used to produce, for example, 100 kilograms of output. Point A would be a relatively capital-intensive technique. Point D would use more labor to produce the same amount of product.
An infinite number of input combinations is theoretically possible on the 100 kilogram isoquant that shows equal quantities of output. In practice, however, only a limited number of combinations are likely to be important to farmers. Figure 3-2 illustrates four alternative techniques: hand labor (point D), oxen (point G), a small tractor (point B), and large mechanized equipment (point A). The isoquant connecting these points portrays the possible technical alternatives for growing 100 kilograms of rice.
The appropriate combination of labor and capital is determined by the prices of the inputs. A farmer cultivating with human labor, who contemplates using oxen, wants to know how much labor is saved and how much
oxen time is needed. Schematically, this is shown as a move along the isoquant from point D to G, to represent a change in the combination of inputs, with ~L less labor and ~C more capital. If labor and capital were priced so that the cost of the labor given up were greater than the cost of the additional capital used, that is, if ~L . PL > ~C . PC where PL and PC represent the prices of labor and capital, respectively, the farmer would find the switch to the more capital-intensive combination profitable.
With the prices for the two inputs known, it is possible to construct an isocost line connecting points of equal costs. This line represents the various possible combinations of labor and capital that have the same cost. Like the slope of the isorevenue line, the slope of an isocost line is the negative of the price ratio of the two inputs. In figure 3-2, where the isocost line is tangent to the 100 kilogram isoquant, at point G, the farmer has determined the least-cost combination of labor and capital that will produce 100 kilograms of output. At any other point on the same isoquant, it will cost more to produce that amount of output. When marginal costs equal marginal revenue (~L . PL = ~C . PC), a farmer is using the least-cost combination of inputs to produce a given level of output.
To produce any more output, the farmer would have to use more capital, more labor, or both. Each level of output has its own isoquant, represented in figure 3-2 by isoquants labeled 200 kilograms and 250 kilograms. A dashed isocost line is shown parallel to the first (the price ratio of labor and capital is the same) and tangent to the 200 kilogram isoquant. The point of tangency at G' represents the least-cost combination of labor and capital to produce 200 kilograms of output. In this example a farmer using the least-cost combination to produce 200 kilograms would employ relatively more units of capital than labor in expanding output from 100 kilograms to 200 kilograms.
Farmers make decisions about their techniques of production-their factor-factor choices-according to the price relationships that prevail for the factors relative to their productivity. Whether those choices are "appropriate" in a broader social sense depends on whether the prices and available technology that led to the decision reflect the total costs to society of the techniques chosen. If capital is subsidized, either directly or indirectly, farmers are likely to choose techniques that use more capital than otherwise. If labor policy tries to push up wages, fewer workers will be hired.
The actual prices for factors of production faced by rural decisionmakers-wages, the cost of capital and imported equipment-are influenced significantly by macro policy. Such policy is often made by government officials who have little knowledge of whether the resulting rural decisions produce appropriate or inappropriate technology choices in agriculture. In many developing countries macro policy is designed to keep capital cheap to favor investment, to raise wages to increase incomes of workers, and to provide direct or indirect subsidies to imported capital machinery, such as tractors or combines, to raise productivity in agriculture.
If unskilled labor is widely available in both rural and urban areas, however, these policies ofren have exactly the opposite effect from that intended. The number of jobs created for each dollar of capital invested is low, wages outside the large-scale, formal sector are depressed, and mechanized farming exists side by side with rural unemployment and extreme poverty. The economic incentives determined by macro policy influence thousands of decisions about how to plant, cultivate, and harvest crops. These decisions in turn influence how many workers can find productive jobs directly. The breadth of rural purchasing power, largely a function of agricultural prices and the choice of technology in agricultural production and processing, determines the indirect and second-round employment effects. Together, the direct and indirect employment effects reflect how dynamic the rural economy is and how widely shared are the benefits of growth.
Agricultural performance is linked to macro policy not only through farm-level decisions about which crops to grow and how to grow them, but also through the overall response of total farm output to the economic environment that determines the profitability of more intensive agricultural effort. Policymakers are concerned about the outcome of farmers' decisions, for they determine the level of food grain supplies, the availability of foreign exchange earnings from the agricultural sector, and incomes in rural areas. To understand how these decisions are made and how they affect such important variables of policy concern, a production function relating inputs to output is a convenient conceptual tool.
Various technical relationships, the prices of inputs, and the output price the farmer expects are all weighed in the decision of how intensively to use factors to produce output-the factor-product decision. The production function is the basic technical relationship used to analyze these issues, and it is illustrated by curve GEMH in figure 3-3. This simplified one-factor function shows the yield of rice per hectare which could be expected from applying different amounts of fertilizer. This function assumes that other factors of production (such as land) are fixed and that all increases in output are due to fertilizer, the variable input shown on the horizontal axis. The figure is drawn to show diminishing marginal returns, that is, each additional unit of fertilizer results in a smaller increment in output. If no fertilizer were applied, a yield of OG would be obtained. The physical maximum yield of OD would be attained with OB application of fertilizer. Informal talks in the countryside with farm households and agricultural research workers can give the analyst some insight into what these values might be.
Curve GEMH shows the rate at which fertilizer can be converted into rice at varying levels of fertilizer input. This conversion is nature's exchange relationship between rice and fertilizer. The exchange can also be made in the other direction, from rice to fertilizer. When farmers take rice to the market and return with fertilizer, they are carrying out a market exchange even if they use money as an intermediary for convenience. The rate at which farmers can exchange rice for fertilizer is also shown in figure 3-3 as the line OP. It reflects the ratio of the price of fertilizer to the price of rice. When fertilizer prices rise, the line becomes steeper, reflecting the fact that more rice is required to buy a unit of fertilizer. Inversely, if rice prices rise, the line becomes flatter as each bag of rice buys more fertilizer.
For most farmers the price line is more or less straight. Except for quantity discounts for large purchases and price premiums for very small purchases, rice and fertilizer prices are little affected by individual farmer decisions. Because prices are nearly the same whatever the level of use, the line OP can also be thought of as a total cost curve in this example, for fertilizer is the only input. (The generalization to many inputs provides similar insights but with more complicated mathematics.) Here costs are measured in the same units as output, and so any excess of output over input costs for a given level of input use means the farmer is earning a profit. When the cost curve is above the production function, net revenues are negative and losses occur.
In a riskless world where farmers maximize profits, figure 3-3 can be used to determine how much fertilizer a farmer should use and how much output would result. The maximum profit occurs when the distance between output and input costs is the greatest. This point can be found by shifting the price line OP upward in a parallel fashion until it is just tangent to the production function. The dashed line shows this point of tangency at point E, where total rice production is OC and fertilizer use is OA. An amount of rice equal to AQ must be exchanged for the fertilizer used (OA), thus leaving an amount of rice equal to QE left over to repay the farmer's labor and use of land. AQ plus QE add up to OC, which is total output.
Of course, the degree of risk faced by farmers varies enormously-collective agricultural systems often significantly cushion individuals against risk, while market systems expose small farmers to substantial risks that affect their use of inputs. Also, pure profit maximization is an extreme case of rational behavior not likely to be found in the complicated world in which farm households make decisions. But an alternative formulation of the decision-making framework of profit maximization can illustrate how farmers adjust their fertilizer use and output decisions in response to inappropriate starting points or changes in prices or technology: the farmer simply seeks to move in a direction that increases net revenue. By comparing the additional revenue any increased yield will bring with the cost of the additional fertilizer required to produce that output, the farmer can decide whether additional fertilizer is profitable. If the marginal cost of the fertilizer is less than the marginal revenue (if ~F . Pf < ~R . Pr) the additional fertilizer is profitable. The farmer will continue to use fertilizer up to the point where the slope of the production function equals the slope of the ratio of the price of fertilizer to the price of rice (~R/~F = Pf/Pr), which is reached at E in figure 3-3.
This is the same point found by maximizing profits in a single, all-knowing decision, but this time the farmer arrives at E by a more plausible route of incremental trial and error. When farmers compare additional costs with the expected additional benefits, a natural and rational way to make choices, their behavior approximates that predicted by these simple economic models. Consequently, thinking about how farmers will respond to changed economic or technical circumstances with this basic production function model is likely to provide analysts with considerable insight into what will actually happen.
Figure 3-4 shows how this framework can help in understanding likely farmer reactions to significant changes in the underlying technology available for rice production. The development of modern fertilizer-responsive seed varieties shifts the entire production function up, allowing more output to be produced even with the same fertilizer input. But something else has happened in the shift as well, for even at the same fertilizer-to-rice price ratio a larger application of fertilizer is now profitable. The optimal point is E" where OK fertilizer is used to produce OC" rice.
The increase in output is composed of two separate effects of the technical change. As figure 3-4 shows, there is a neutral increase in yields from C to
C' even when fertilizer use stays constant at the previously optimal level OA. This increase occurs because the production function has shifted upward from E to E'. Second, because of the fertilizer-using nature of the technical change, optimal fertilizer input shifts from E' to E" even though the price relationship between rice and fertilizer remains the same. Fertilizer use expands from OA to OK (for those farmers who can afford it), and output reaches its new optimal level of OC". A shift can also occur through simple learning. As they observe other farmers' results with fertilizer or experiment with small amounts themselves, farmers gradually shift their own production function and demand for fertilizer upward.
The availability of different technologies may also explain why some farmers appear to be "inside" the production function, as at point J in figure 3-3. As figure 3-4 indicates, such farmers may be using the traditional seed varieties either for lack of knowledge or for lack of access to the appropriate inputs needed to use the modern varieties efficiently.
The second major factor influencing the farmer's decisionmaking environment shown in figure 3-4 is the relative price of fertilizer to rice, for this relationship determines the economic incentives to use more fertilizer. In most countries these prices are heavily influenced by government policy. Figure 3-5 illustrates what happens when the price of rice increases or the price of fertilizer falls (the two are equivalent in this two-dimensional world, and hence only the price ratio is important here). As fewer units of rice must be exchanged for a unit of fertilizer, the farmer is encouraged to use more fertilizer to grow more rice.
As long as the farmer can convert one unit of fertilizer into enough rice to buy more than that unit of fertilizer, it makes sense to expand fertilizer use. When an additional unit of fertilizer fails to produce enough rice to pay for itself, the farmer has gone too far. The appropriate stopping point is where
the exchange ratios are the same, a lesson already learned. Lowering the amount of rice needed to buy fertilizer-lowering the price rati~normally leads to increased fertilizer use and higher yields (and vice versa). According to rhe theoretical model, farmers are expected to apply inputs more intensively in order to increase their output of a commodity when its price goes up, if other prices remain constant. This positive supply response can also be illustrated within this framework, and the result is shown in figure 3-6.
The upper part of figure 3-6 is constructed from the technical relationship between inputs and resulting output that is shown in the production function in figure 3-3. Because the price of rice in this example does not depend on quantities sold, the total revenue line for the farmer is a straight line, where the angle indicates the rice price itself. At higher prices the angle is steeper, indicating more revenue per unit of output. The total cost curve is constructed from the production function and a given price of fertilizer. At each level of output a particular level of fertilizer is required. The cost of purchasing this fertilizer determines the cost of producing that output. As in figure 3-3' the optimal output is at point E, where the excess of total revenue over total cost is maximized, that is, where profits are greatest. As noted before, this point is also where the slopes of the two curves are equal, or where marginal revenue (equal to the price of output) equals marginal cost (the slope of the total cost curve).
The lower part of figure 3-6 shows these marginal conditions directly as they relate to total output. The vertical axis now measures cost per unit, as well as the price per unit of output. Since the price at which the farmer can sell output is constant for all output levels, it can be represented by the horizontal line at PO . Both the average cost curve and the marginal cost curve rise as output rises because inputs are being used more intensively with lower marginal productivity. The higher marginal costs then pull up average costs. Because output greater than D is impossible from this particular farm with its available technology, marginal and average costs become infinite at that point.
The farmer's choice of output level in the lower part of figure 3-6 corresponds to the choice in the upper part (and to the choice in figure 3-3). This point is again E, where the marginal cost of increasing output equals the marginal revenue gained by producing it. This marginal revenue is the price of output-each additional unit of output sold by the farmer brings in revenue equal to the price of output-and so the farmer's best choice is where the marginal cost curve intersects the market price of output. This is an extremely important result. If market demand or government policy caused the price of output to rise from PO to Pl the optimal decision point for the farmer would change from E to E', and output on the farm would rise from OC to OC'. The farmer's supply response to higher price incentives-the relationship between output supplied and output price-is simply the farmer's marginal cost curve for producing additional output.
Note: The farm area planted in rice is held constant.
With all the provisos noted about the simplified nature of these two dimensional diagrams, the farmer's supply curve is the same thing as the marginal cost curve. Anything that shifts the marginal cost curve, for example, new technology, access to new irrigation facilities, or even the weather, will also shift the supply curve. Many of these shifts result directly or indirectly from government policy or investments, and so the interest of analysts in farmers' responses is clear. Because the supply curve summarizes so much of farmer decisionmaking in terms of two variables of great relevance to the rest of food policy-output and price-knowing more about the elasticity of supply for important commodities is the next step for food policy analysts.
Estimating Farmer Supply Response
Government policy influences the location of the supply curve through investments that lower marginal costs of agricultural production (or through unintended actions that raise costs). Policy also influences where on the supply curve farmers choose to be, as price policies alter the incentives to use more intensive techniques of farming to produce more output. "Cheap food" policies can suppress growth in farm production while increasing consumption, often requiring subsidized food imports to be effective. As governments look at the costs and benefits of such policies, an immediate question is whether farmers will respond with greater output if better incentives are provided or with less output in the face of reduced incentives. The answer will vary for the short run as opposed to the long run, as well as for areas where additional land can be brought into cultivation. Some environments, especially in Asia, must rely on yield increases as the primary means of raising output. In addition, the supply response for individual crops, where, for example, corn can substitute for beans, differs from that for aggregate agricultural production, where substitutions do not alter total output significantly and response must come through changed intensity of input use, including labor.
These concerns are empirical, not theoreticaL They can be addressed only by careful attention to exactly which question is being asked, coupled with specific statistical analysis of country or regional data. The empirical estimation of supply response functions is an enormous and complicated topic and can be surveyed only briefly here. Just as with sophisticated food consumption function estimation, the food policy analyst is likely to be less concerned with actual estimation techniques than with a solid sense of what the important issues are, when to distrust econometric razzle-dazzle, and how to interpret representative empirical results.
The lower part of figure 3-6 shows a positive relationship between price and quantity of output. A natural inclination might be to look for a series of observations on a commodity's price and its output and to plot them graphically or even to estimate a regression with output as a function of price. Sometimes this technique actually works, but often the result is merely a confusing cluster of data points or, even worse, a perceptible negative relationship between price and output. Does this mean that farmers are perverse and have a backward-bending supply curve and produce less as prices rise? Usually it means the analyst has identified a curve with elements of both a demand curve and a supply curve. This "identification problem" has a famous history in economics, and although the theoretical issues are resolved, it continues to bedevil empirical investigators. Without additional information about whether the supply curve or the demand curve, or both, are shifting, any estimation results are unclear. If no additional information is available, the identification problem is simply insoluble.
The most common approach in the empirical literature is to assume that consumers react to actual prices in their purchase decisions and that farmers react to expected prices in their planting and cultivating decisions. As argued earlier, this distinction is an important characteristic of agriculture because of the time between the farmer's input decisions and the output results. If a way can be found to approximate the price the farmer expects when making input decisions, most of the identification problem is resolved. Many models simply assume the farmer reacts to the previous price received, with a time lag of one year or one season. The results of estimating these models with time series data are surprisingly useful, for they indicate how responsive farmers actually are to short-run changes in price signals. Naturally, they may adjust much more as time passes, but the simple one-year lag model captures their immediate response.
The longer-run response can be captured only with more sophisticated models of formation of expected prices or of lags in responding to changed environments. An innovative methodology for doing this modeling was first used in an agricultural setting by Marc Nerlove and is appropriately called Nerlovian distributed lag analysis. Reproducing the complicated algebra needed to derive an appropriate estimating equation is not necessary because the results are intuitively plausible.
A farmer's output is a function of the previous price of output and of output the previous year. This lagged output term serves as a proxy for all previous adjustments to prices (and even to other excluded variables). The short-run supply response to price continues to be estimated by the coefficient attached to lagged price, but the long-run adjustment is larger by a factor determined by the coefficient estimated for the lagged output variable. (The actual formula is: long-run supply response = short-run supply response / [1 - the coefficient of lagged output].) In a typical case, if the short-run supply elasticity is 0.1 and the coefficient attached to (log) lagged output is 0.6, then the long-run elasticity is 0.25, that is, 0.1/(1 - 0.6). A very common result of this type of analysis is that the long-run supply elasticity is roughly twice as large as the short-run elasticity.
This methodology for direct estimation of supply curves requires time series data with enough observations to provide reliable statistical results at the same rime that significant structural change or technological innovation is minimal. Most developing countries have undergone both in their relatively short histories, and most supply curve estimation is confounded by these likely but difficult-to-measure shifts in the supply curve itself. As figure 3-7 shows, rapid shifts in,the supply curve mean that both an apparently positive or a negative estimated supply response could be consistent with actual p05itive and quite inelastic short-run supply curves.
One approach to dealing with this problem, and other aspects of the identification problem as well, is to use cross-section data to estimate supply functions. As with the use of cross-section data for consumption analysis, this approach requires that the decisionmakers face reasonably comparable environments if the results are to have any direct policy relevance. At the
same time, the decisionmakers must face different prices that cause measurable differences in output if the statistical analysis is to succeed in estimating an accurate supply response to price. Multiple regression analysis can be used to control roughly for differences in environments, but many factors that decisionmakers consider are extremely difficult to measure.
Cross-section estimates are frequently made across states, provinces, and even countries in order to find significant price variation. With such crosssection estimates, the argument is usually made that the resulting supply response parameters reflect full and long-run adjustment in all aspects of the environment that are related to price, not just the short-run response of farmers to price when that environment is held constant. Since this includes such important variables as farmer knowledge, irrigation facilities, and well-adapted seed varieties, this distinction is quite important. In fact, several researchers have discovered very high supply responses to agricultural prices by using this methodology. Work by Willis Peterson, in particular, shows an aggregate agricultural supply elasticity of about 1.2, but the countries in the sample that generated this result are probably a century or more apart in their economic development and hence in the full adjustment of the environments facing their farmers.
An alternative approach to direct estimation of farmer supply response uses technical or engineering data to estimate the agricultural production function, specifying an appropriate functional form and then using conditions of profit maximization to trace out the marginal cost function and hence the short-run supply response. Either time series or cross-section data can be used for this indirect approach, and each presents its own set of difficulties. The obvious general problem with estimating production functions to derive supply response functions is that the approach assumes what was the object of the search, that is, the extent to which farmers actually respond to price changes. Such "normative" supply curves are useful for placing upper bounds on plausible farmer responsiveness, and they sometimes show that farmers are already using inputs such as fertilizer about as intensively as is profitable. They are a relatively poor basis, however, for making predictions about the impact of changed prices on aggregate supply unless no other information is available. Then, of course, the one-eyed man is king.
A recent volume by Hossein Askari and John Cummings surveyed agricultural supply response estimates available in the mid-1970s. The book has an extensive review of the Nerlovian supply response model and of many of the studies around the world that tried to use it to estimate supply elasticities for a wide variety of crops. Table 3-1 is extracted from their summary table and shows the shortrun supply elasticities for rice,wheat, and corn, by approximate numerical range, for the regions and countries for which reasonably reliable estimates could be located
|Grain||Range of supply elasticities and region|
|Less than zero|
|Rice||Uttar Pradesh,a Himachal Pradesh,b Gujarat,b Maharashtra,b Madras,b,c Kerala,b Egyptb,c|
|Wheat||Uttar Pradesh,a,c Madhya Pradesh-Berar,b,c Bombay-Sind,b,c Iraqb|
|Zero to one-third|
|Rice||Assam,b Bihar,a Mysore,b Punjab, West Bengal,b Tripura,b Pakistan,b Bangladesh,b Thailand, West Malaysia, Japan,b Philippines, Egyptb,d|
|Wheat||Mysore, Punjab,b Rajasthan,b West Bengal,b Maharashtra,b Himachal Pradesh,b Pakistan,b Hungary, Jordan,b Lebanon, Egypt,b,c United States|
|Corn||Punjab, Egypt,b,d Lebanon,b Sudan, Philippines, United States|
|One-third to two-thirds|
|Rice||Punjab,a Bihar-Orissa,a Peru,b Java, Iraq|
|Wheat||Uttar Pradesh,a,d Bihar,a Egypt,a,d Syria, Lebanon,a New South Wales, United Kingdom,b France, Argentina, Chile|
|Corn||Punjab, Hungary, Sudana|
|Two-thirds to one|
|Wheat||Gujarat,b Egypt,a New South Wales,a New Zealand, United States, Canadab|
|More than one|
|Rice||West Malaysia,a Iraqa|
|Wheat||Syria,a New Zealand,a Chilea|
Note: Unless otherwise noted, elasticities are short run.
a. Long-run elasticity.
b. Short- and long-run elasticity.
c. Pre-World War II.
d. Post-World War II.
Source: Assembled from Hossein Askari and John T. Cummings, Agricultural Supply Response:A Survey of the Econometric Evidence (New York: Praeger, 1976)
Askari and Cummings note that their summary table was prepared by weighting different, sometimes conflicting, results for the same crop and region and by using their own judgments about the reliability of alternative estimating procedures or the particular time series data employed. Table 3-1 gives only a hint of the full array of evidence gathered by Askari and Cummings; indeed, nearly 500 separate supply elasticities are presented in their appendix!
Two points are important. First, most of the supply elasticities reported for basic cereal grains are positive, with a median value likely to be near the top of the zero to one-third range. Although it is preferable to determine grain supply elasticities on the basis of local conditions and data, sometimes this is not possible. If forced at gunpoint to pick a number from comparative experience, the analyst will not be far from the mark with an elasticity of 0.2 to 0.3.
Second, there is a noticeable tendency for the supply elasticity to be larger in more-developed countries and in regions with longer commercial histories. This larger supply response is partly because farmers are more economically minded, but also because purchased inputs have a greater role in farm production. The decision to purchase inputs instead of relying on traditional household resources inevitably reflects a willingness to calculate costs and benefits. In such circumstances the actual supply response begins to approach the normative supply response predicted by the economic models used here to understand farmer behavior. As the role of modern agricultural technology widens, the supply responsiveness of farmers around the world is likely to increase.
The neoclassical decisionmaking approach outlined in this section can be very helpful to analysts in illuminating the underlying factors that explain trends in output of particular crops, in input use, and to some extent In incomes in farming areas. Lagging productivity growth in corn production might be traced to low fertilizer use. This could be explained by poor technology available to farmers or to poor incentives to use inputs intensively. The answers have direct relevance to agricultural policies, whether for budgetary resource allocations to research stations or for improved price incentives for important crops. A sense of the responsiveness of farmers to policy changes is needed to pursue the issue from identification of problems to appropriate policy responses. In the early stages of agricultural development when one or two primary food crops dominate both traditional farm production and policy interests, fairly simple supply response analysis can provide many of the needed insights. As agriculture becomes more commercialized and more complicated, a broader approach that examines the full decision-making environment of the farm household is likely to offer important additional insights.
Neoclassical production theory is very helpful for understanding the direction of likely farmer response to a changed economic environment at the farm level. This approach is crop-specific, and the empirical estimates of farmer supply responsiveness have concentrated almost exclusively on single commodities, as table 3-1 illustrated. Useful as this focus and the accompanying supply elasticities are to policymakers in contemplating either pricing changes or the macro connections between agriculture and the rest of the economy, several critical farm-level and broader rural issues are neither identified nor analyzed within this framework.
Commodity substitution, land and labor allocations to alternative crops and farm household tasks, and potential income-earning opportunities for farm labor in rural off-farm jobs are broad and important issues. They influence the income and welfare of the farm household directly while contributing indirectly to the commodity supply responsiveness of immediate concern to policymakers. These issues can be understood by analyzing the entire set of farming activities and important nonfarm activities within a decisionmaking framework that specifically accounts for the connections and opportunity costs across both crop and other income-earning activities.
Activity analysis, or linear programming, is a technique for analyzing entire farming systems. Based on the construction of a farming system tableau, it is a helpful and ultimately powerful technique for understanding these broader farm decisionmaking issues. The tableau can help analysts identify the aspects of farming systems that need much more careful attention and more rigorous analysis. Specialists or outside consultants can then be brought in for this work, and their efforts made productive and relevant by specific terms of reference based on the preliminary assessments from the tableau analysis.
Describing the Agricultural Sector
Describing the agricultural sector in statistical terms is complicated by agriculture's unique characteristics that set it apart from the rest of the economy. Annual production statistics by crop for the entire country are very important for describing overall crop balance, aggregate supplies relative to domestic demand and import needs, and gross input requirements to maintain aggregate rates of growth in agricultural output. Since ministries of agriculture, planning agencies, and central banks need these aggregate estimates to plan investment and foreign exchange allocations, their gathering and analysis has dominated the statistical description of the rural sector.
The individual farm household decisions that generate these aggregate results are made from a very different perspective, however. Each farm sits within a particular ecological setting (sometimes more than one) and faces an economic environment dictated by the technological opportunities for crop production in that setting and by the input and output markets nearby. To understand how farm households will react when new technology is developed, input prices are subsidized, or output prices are raised, policy analysts need much more specific information about these indiv~dual decisionmaking environments than can be gleaned from national aggregate data. Such information usually must be collected from relatively homogeneous agroclimatic zones at the farm level. The descriptions of representative farming systems, the variations in yield and prices across those farms, and the distribution of farm sizes within each zone will provide the analyst with
sufficient information to understand how various farm households are likely to react to policy or exogenous changes in rural conditions and how these changes together provide an aggregate view of farmer responsiveness to economic incentives.
The ecological setting in which crops are grown strongly conditions the farmer's choice of crops, the techniques used to grow them, the resulting yields, and how much is available for sale in outside markets. Every field has a unique ecological setting, and each farm is different from neighboring ones. Too acute an awareness of this ecological diversity is immobilizing to policy analysts attempting to define agroclimatic zones, however, because central tendencies and representative settings are swamped in the local detail. Finding an appropriate balance between aggregate national data and the specific crops grown in individual fields and farms is necessary to bring any sense of order to the chaos of nature. It is good enough to know in principle that every farm is different. In practice, finding a half dozen or so representative agroclimatic zones that are reasonably consistent internally and have significantly different cropping patterns from zone to zone will capture all the diversity the analyst can hope to cope with and still retain an overall sense of how the pieces add up to form the nation's agricultural output.
When the analyst has time to conduct original survey research or has access to raw data from village-level surveys, boundaries around agroclimatic zones can be drawn on the basis of uniform ecological settings. Most analysts will have to settle for less precise boundaries, however, and will use district or provincial agricultural data that have already been collected. These administrative units are usually responsible for collecting agricultural data. When the administrative borders correspond, even roughly, to agroclimatic zones, the analyst can shortcut the data-gathering process significantly. In most cases, major compromises over the definition of a true agroclimatic zone are worthwhile to make it fit the existing data.
Just such a compromise has been made in the example used throughout the discussion of data gathering and food production analysis. East Java, a major province of Indonesia, is a diverse region with a larger population (more than 30 million people in 1981) than most countries. It is not a single agroclimatic zone, yet its representative farming systems are significantly different from those of neighboring Central Java or Bali. Certainly much can be learned by looking at individual districts or even villages within East Java, for they vary considerably. But in view of the difficulty of trying to understand Indonesian agriculture and its production responsiveness on the basis of individual farm households or the thousands of districts or hundreds of thousands of villages, a focus on representative decisions and production patterns in East Java is an appropriate compromise.
Even when provincial data correspond fairly closely to agroclimatic zones, as they do in East Java, the data most useful to analysts may not be available from government agricultural offices. Because government departments and programs are often organized around single crops (for example, a rice improvement program), agricultural ministries tend to collect new information mainly on individual commodities. Having data on typical farms, however, is likely to be much more important for understanding how farmers react to policy changes. Since most key farming decisions involve tradeoffs and opportunity costs as farmers seek to increase their incomes, information is needed about alternative farm activities, not just about individual crop productivity and response.
Farm-level data can be limited to the major crops or they can include garden, household, livestock, and off-farm employment activities. Where land is abundant and labor is relatively expensive, as in much of Africa, data limited initially to crop activities should suffice, for most income tradeoffs will revolve around crop choice. If labor is abundant and land is relatively much more scarce, as in parts of Asia, off-farm employment activities are likely to be more significant in determining household incomes and consequently in assessing farmer response to agricultural policies.
Data from representative farms in a half dozen different agroclimatic zones provide the descriptive basis for setting up a farming system tableau. The analyst begins with the main crop activities for the zone and adds other components, such as livestock or off-farm employment, when they are needed to understand farmers' responses to agricultural policies. The tableau serves to organize the analysis of important rural issues. To link the tableau analysis to policy issues, the analyst also needs farm size distribution and price and yield data. Some of these data can be found in published sources. Others wiU require discussions with farmers and agricultural field staffs. First-hand knowledge gained while spending time in the countryside gathering data imparts a sense of the diversity from field to field and the actual environment in which farmers must make their decisions and is thus as important for the analyst as the tabulations themselves.
The combination and sequences of crops grown by representative farmers make up a farming system, and usually only a few different types are dominant in an agroclimatic zone. As with the zones themselves, any definition of dominant farming systems is somewhat arbitrary. Cropping patterns vary among farms because of differences in soils, irrigation, prices, and proximity to markets. Still, policy analysts can usually pick a few primary cropping arrangements within a zone and resist the temptation to describe a large number of farming systems in hopes of representing all the various patterns within the countryside. Too many data tend to obscure rather than clarify the issues before policymakers, who need to understand how decisionmakers in a few representative farming systems are likely to respond to alternative policy initiatives that will affect the farming environment.
The cropping system diagram shown in figure 3-8 is an efficient way to visualize farming systems. To capture the seasonality that is important in agricultural production, the horizontal axis is divided into time periods. The three-season agriculture of East Java is shown over a one-year period, broken down by month. Systems involving tree crops might cover a longer period. A temperate agriculture that is limited by the number of frost-free days would cover a much shorter time.
Each rectangle in figure 3-8 represents the percentage of total arable land
that a given crop occupies in the region over three major seasons. The height of the rectangle is calibrated to show the relative area planted to a crop, thus indicating the importance of the commodity in the region. Such a visual picture of the farming system is simple and readily understandable and helps identify important decision options and tradeoffs faced by farm households. Other useful seasonal data, such as rainfall, irrigation flows, temperature, or intensity of sunlight, can be charted similarly to the rainfall graph in the lower part of figure 3-8.
This simple presentation of a farming system reveals important crop competitions and complementarities within seasons. The interactions between weather and cropping systems are also highlighted and may indicate potential benefits of irrigation or drainage investments. Several charts of this type can readily present the broad dimensions of farming operations within a region.
Table 3-2 illustrates data on farm size distribution for the province of East Java. Many of the numbers in this table and the other examples are approximations rather than actual data, and the farming systems have been greatly simplified to illustrate the concepts more easily.
From table 3-2 it is apparent that more than half (54 percent) of the farmers have landholdings of less than 0.3 hectares, yet they occupy a total of only 18 percent of the land. In terms of production, the 30 percent of farmers with 0.3 to 0.5 hectares are more representative than the smaller farmers because they cultivate more than one-quarter of the land. Because farm-size distribution data identify different types of "representative" farmers-most representative in total number or in area farmed-the information in table 3-2 helps organize field visits. To think about the implications of policy changes for agricultural production, analysts need to understand the likely reactions of several representative farmers. The table identifies their characteristics; field visits then permit analysts to talk with these farmers about their decisionmaking.
|Size of farm (hectares)||Number (millions)||Percentage of farmers||Area (millions of hectares)||Percentage of land|
Note: The numbers illustrate general trends; they are not specific data representing actual conditions.
A simple farm-size distribution table can also indicate a great deal about the income distribution of a rural region and the numbers of farmers whose net marketings are likely to be significant. Information on respective yields by farm size can also be added, although this is complicated by different cropping patterns and by differential access to inputs. For example, small farmers often use multiple cropping systems that are much more labor intensive than the cultivation of uniform stands of cereals. Further, they often achieve higher yields per hectare for a given level of fertilizer use, but credit constraints cause them to use less fertilizer per hectare than larger farmers.
A rough estimate of the number of people likely to be affected by hunger if off-farm employment is unobtainable can be made by adding the number of landless and near-landless laborers to the table. In the East Java example an additional 2 million families would be included. In areas where large landowners split their holdings among several tenants, a table showing ownership rather than operating patterns would reveal something of the role of landlords in the countryside. Accurate ownership data are usually difficult to assemble, but they are particularly important for assessing policy changes affecting tenancy or landlord-tenant relations.
Information on farm prices and yields actually paid and received by farmers brings the analyst face to face with the decisionmaking environment in which rural households operate. By talking with a range of farmers and with experiment station workers, the analyst can draw a rough picture of the relationship between input use and crop yields. With input and output price data added, the profitability of intensifying crop production can be estimated. In combination, the input, yield, and profitability data provide the analyst with a sense of potential gains in production from available technology or changes in the structure of costs and prices and of the importance of developing new technology.
Prices actually received and paid by farmers in the region are essential to calculating several price ratios that reflect the economic environment facing farm decisionmakers. In particular, the ratio of commodity prices received by farmers to the price paid for a key input such as fertilizer provides a rough assessment of how tightly the agricultural sector is being squeezed by low economic incentives relative to other regions and countries.
A second calculation compares regional prices with international market prices for each commodity. The comparison between rice prices in Bangkok and in East Java shown in table 3-3, for example, cannot be made with any precision without careful attention to quality differentials and possible exchange rate biases. But product and factor prices often vary enormously from country to country relative to the import or export price. These wide price variations can alert analysts to local price environments that are greatly distorted. In the Indonesian example in table 3-3, local corn prices may be
|-||Tons per hectare per crop|
|Technology||Shelled corn||Milled rice|
|Experimental yields per hectare per crop||-||-|
|Ratio of international to East Java||2.9:1.0||1.5:1.0|
|"Progressive" farm yield per hectare per crop||-||-|
|Average farm yield per hectare per crop||-||-|
|-||Dollars per ton (rupiahs converted at Rp 625 per US$1)|
|Prices||Shelled corn||Milled rice|
|World price, October 1980||$145a||$445b|
|East Java farm price, October 1980||145||280c|
|Ratio of world price to East Java price||1.0:1.0||1.6:1.0|
|World price of urea, October 1980||250d||250d|
|East Java farm price of urea, October 1980||115e||115e|
|Ratio of world price to East Java price||2.2:1.0||2.2:1.0|
|East Java commodity price/East Java urea price, October 1980||1.26||1.78|
|World commodity price/world urea price, October 1980||0.58||2.43|
a. F.o.b. U.S. Gulf port (no. 2 yellow).
b. F.o.b. Bangkok (25 percent brokens).
c. Rice equivalent at farm gate.
d. F.o.b. Near East (bagged).
e. Subsidized rate of Rp 72 per Kilogram.
from the United States; rice prices seem significantly lower than the potential cost of landing Thai rice in Indonesia. Urea prices are less than half the world price, indicating very large subsidies.
Wide deviations in local prices from international market quotations reveal the potential for significant economic distortion and waste through misalocation of resources. Some of this distortion may be planned to advance other government objectives, such as food security or income transfers top farmers or consumers, but frequently the distortions and wasted resources are an unforseen byproduct of government policies for the nonagriculture sectors. By identifying the magnitude and potential impact of these price distortions in the countryside, analysts can bring to planning sessions a much more informed and persuasive case for showing how government policies affect agricultural performance.
The exact calibration of yield and price data for a region requires fieldwork plus a bit of desk research to fit the data together. They can be assembled rather quickly, however, into the type of information in table 3-3. This table illustrates how much variation can occur across commodities even within the same region and for different types of farmers. It is especially important to be on the lookout for factors that create an adverse economic climate for small farmers. When appropriate allowance is made for credit arrangements with suppliers, village moneylenders, and purchasing agents, small farmers often pay higher input prices and receive lower output prices than do large farmers.
In the East Java example, the ratio of local to international experimental yields indicates that the development of appropriate varieties is much further advanced for rice than for corn. International experimental rice yields are only half again as high as East Javan experimental yields. In contrast, international experimental corn yields are nearly three times as high as in East Java, despite the fact that the ratio of domestic to international prices is much more favorable for corn than for rice.
If farmers are obtaining low yields for the major food crop in a country, there is a reason. Perhaps farmers do not have a technological package capable of producing high yields reliably in their ecological setting. Perhaps farmers do not judge the economic returns from high yields to be worth the costs, efforts, or risks of producing them. Or they may not know how to use the available (and accessible) technology in such a way as to make a reasonable proftt by producing high yields. No doubt any one of these reasons is important in many circumstances, and maybe all are relevant on occasion. A constraints framework developed by Arthur Mosher helps analysts organize these factors and identify the constraints on farmer behavior under existing circumstances.
For a particular agroclimatic zone, the Mosher framework organizes farm households according to their output per hectare for the dominant crop in the region. In figure 3-9, the horizontal axis represents the percentage distribution of all arable land within an agroclimatic zone, and the vertical axis measures yield per hectare for the dominant commodity, usually one of the cereal grains. The achievement distribution, denoted a, indicates the distribution of yields actually obtained by farmers, ranked left to right from the highest to the lowest. The slope and location of the a curve is determined empirically from the yield data. If all farmers had the same yields, the achievement distribution would be horizontal. Analysts would expect farmers' yields to vary because not all farm households have equally good managers, land quality varies, and not all
Source: Arthur T. Mosher, An Introduction to Agricultural Extension (Singapore: Agricultural Development Council, Singapore University Press, 1978), p.73.
farmers have access to specific knowledge about how to raise their yields. Thus the a curve slopes downward. When the height of the achievement distribution measures the yield and the base reflects the land area achieving each yield, the area beneath the curve represents the total production of the crop, shown as the shaded area in the diagram.
The technical ceiling, the t curve, indicates the biological maximum yield for that crop, the yield an experiment station could achieve in the region. (Yields technically possible elsewhere in the world can be above a region's t curve within a zone.) The technical ceiling curve probably also slopes downward somewhat because of varying soils or other biological reasons.
The economic ceiling, shown by the e curve, reflects the yield constraint imposed on farmers by various price and technical relationships, and by definition it lies below the technical ceiling. The economic ceiling represents the yield achieved when all inputs are used at their most profitable levels on average. Most farmers are risk-averse in the face of uncertain weather, many face credit constraints on how many inputs they can purchase, and only a few are likely to know with any precision what the profit-maximizing levels will be. For these reasons, the economic ceiling cannot be directly observed in the field.
Still some simple calculations can help the observer determine the yields that reflect the current economic ceiling. If an additional unit of fertilizer will produce five units of grain, but the grain-to-fertilizer price ratio is only 1:6, the rational farmer will not use additional fertilizer to raise yields. The marginal response of grain to fertilizer can be determined accurately only by estimating fertilizer response functions with farm-level data, but it can be readily approximated by asking farmers about their own experience with fertilizer and by comparing different farmers' fertilizer use and resulting yields. Local experiment stations also frequently have guidelines on fertilizer responsiveness in their locale.
With price data and information on farm-level yields relative to experimental yields in several representative regions, the analyst is in a position to plot the achievement distribution quite roughly and determine whether the economic and technical ceilings are tightly constraining or not. A wide variety of possibilities might result, of course, but figure 3-10 shows four quite different environments which capture much of the diversity of agricultural settings in developing countries.
A farming system with low productivity, tightly constrained by technology, is depicted in Area A. Since the achievement distribution is near the economic ceiling, most farm households are doing as well as can be expected, and little increase in output can result from changes in economic policy or from more aggressive extension agents. For food production to be increased, biological and engineering research and investments in land improvement would be needed to raise the technical ceiling. A large number of ecological zones characterized by marginal rainfall and poor soils fit this pattern. In
such areas high-yielding crop technology does not yet exist. In other, more hospitable areas which look like Area A, adaptive research has been neglected, and efforts to borrow appropriate technology from other regions have not been made.
Area B represents rational and knowledgeable farmers whose output levels are constrained by economic policies, especially those affecting prices for inputs and output. When the economic ceiling is far below the technical ceiling, policies affecting input and output prices, international trade, the marketing system, and land tenure are preventing rapid increases in productivity and higher yields.
In Area C a few farm households are taking advantage of the available technology, but many are not. When at least parts of the achievement distribution are well below the economic ceiling, efforts to provide education, credit, and extension services are likely to be important parts of a production strategy to raise the achievement distribution.
A high technology dilemma is represented in Area D. In parts of China and Japan rice yields are currently pushing against the technical ceiling on rice production. Raising the technical ceiling requires long-run investments to increase biological potential and is not a source of rapid growth in the short run. To raise rural incomes more rapidly, it might be necessary to diversify the farming system into higher-valued crops.
Policy analysts can use the Mosher framework to determine the position of a country's agricultural sector with regard to the technical and economic ceilings and the actual achievement distribution. Because the framework is crQp-specific, it is much more helpful in identifying issues for campaigns to raise the output of wheat or rice, for instance, than it is for dealing with the complexity of multicommodity farming systems. At the earJy stages of agricultural development this commodity focus is not a serious shortcoming because major productivity gains from new technology tend to be commodity-specific, and most agricultural development plans are already organized around particular commodities. At later stages in development, however, efforts to raise agricultural productivity usually encounter complicated tradeoffs among commodities as well as competition for farm labor from wage labor markets. Then other analytical techniques, particularly the tableau discussed below, illuminate these more complex decisions made by most farm households.
Modeling Farmer Response with a Farming System Tableau
The broad picture of a farming system shown in figure 3-8 illustrates the alternative crops grown in each season, but it does not explain why farmers chose to grow these crops. To understand these choices more fully, farming systems data can be organized into a tableau of information for economic analysis. The concept of a tableau is derived from early attempts to organize the activities of an economy into a consistent framework to show how national output was generated. The literature on mathematical programming and activity analysis has developed the tableau into a formal tool which can be used to calculate optimal solutions to farming or policy problems. Even without the formal analysis, however, organizing information according to the conceptual elements of a tableau is an effective way to portray the array of decisions that farmers face.
The basic elements of a tableau are presented in table 3-4. Each component is identified in a farming system context and is linked to its analogous relationship in a linear programming framework. The core of a linear programming model is the input-output coefficient matrix-the aij matrix which indicates the amount of input i (for example, fertilizer) needed to operate activity j (for example, a hectare of rice) that will produce, say, 2.5 tons of output. The inputs are listed in rows on the left, the possible crop activities are shown in columns in the center, and the availability of inputs is indicated on the right-hand side of the tableau. Each crop activity produces a net revenue, and simple programming models have as their objective the maximization of the sum of all crop revenues. This happens when at least one input is fully utilized and output cannot be increased further.
The analyst need not (perhaps even should not) be a linear programming specialist to use this tableau framework successfully. The importance of the linear programming technique is its insistence on consistency and accurate specification of variables and data. In short, the tableau helps develop a logical and consistent approach to the full range of farm decisionmaking.
|-||Crop activities (crop production technique)||-|
|Inputs||1 2 3 4 5 ...||Constraints|
|List of inputs needed for all activities(often called the input vector); not all activities require all inputs.||Input-output coefficient matrix (the aij coefficients that indicate the amount a of input i needed to operate activity j on a unit, or hectare, of land. The coefficients are linear and constant).||Input availabilities or resources available to the farmer (often called the constraints vector or the right-hand side)|
Maximize ->the objective function = net revenue
The objective function is the sum of the output for each crop activitv times the net revenue per unit of output for producing it. If no cash costs are incurred, net revenue per unit is equal to the price of output.
Table 9: Inputs for an Illustrative Farming System in East Java, Indonesia
- Crop activities 1 2 3 4 5 6 7 8 - Paddy rice - - - - -- Technique Technique Technique Technique - - - - Inputs 1 2 3 2 Soybeans Peanuts Cassava Corn Land (hectares) - - - - - - - - Season I 1 1 1 - - - - - Season II - - - 1 1 1 1 - Season III - -- - - -- - 1 1 Labor (days per hectare) - - - - - - - - Season I 340 400 435 - - - - - Season II - - - 400 200 175 100 - Season III - - - - - - 150 120 Fertilizer (kilograms per hectare) - - - - - - - -- Season I 0 500 1,000 - - - - - Season II - - - 500 0 0 500 - Season III - - - - - - - 0 Yield (tons per hectare) 2 3 3.5 2.5 0.5 0.4 20 1.0
In a tableau as shown in table 3-5, the crops a representative farmer in a region could grow during a year are listed along with the inputs required to produce each crop on one hectare of land in each season. The major commodity grown in the area usually can be produced with several alternative techniques, and each is listed as a separate potential farm choice. Multiple cropping and seasonality are built into the tableau by indicating various crops and inputs separately for each distinct growing season, denoted in this example as Seasons I, II, and III.
Each alternative crop and cropping technique is a potential activity, and table 3-5 illustrates the alternative activities available to a representative farmer in East Java. The labor inputs, in days per hectare, required for each activity are recorded in the appropriate time periods for the three growing seasons in the year. Similarly, the fertilizer rows list fertilizer inputs for each crop and cropping technique (that is, for each activity). Since the input data for growing these crops have been standardized for one hectare, the coefficient i in the land rows refers to the particular season a given crop occupies that one hectare of land.
The tableau represents the technical relationship between inputs and the resulting output, and hence it is a numerical approximation of the production function. In this example paddy grown on one hectare using 340 days of labor and no fertilizer yields 2 tons of rice (Activity 1). With 400 days and 500 kilograms of fertilizer, a yield of 3 tons can be produced on that same hectare of land (Activity 2). By increasing fertilizer input to 1,000 kilograms and labor to 435 days, the farmer can produce 3.5 tons of rice using Activity 3. These three data points (fertilizer input, labor input, and resulting rice output) lie on a multifactor production function analogous to the one-factor continuous function depicted in figure 3-3.
Farm households must make their cropping decisions in the context of their land, labor, and capital resources available for crop production. To reflect these constraints, the total number of hectares of land, days of labor the family can provide, and capital resources in the form of cash, oxen, irrigation water, or other inputs, are entered along the right-hand side of the tableau, as in table 3-6. The constraints, however, are not necessarily fixed. Family labor may supply 400 days of labor in a year, but if the family had sufficient capital resources and if there were a market for agricultural labor, the farm household could hire additional laborers and alter the labor constraint. Similarly, a credit constraint could be altered if borrowing were a possibility.
The simple tableau in table 3-6 does not capture all possible constraints, particularly biological and behavioral constraints. For example, if nematodes infest the soil, such crops as tomatoes must be rotated from field to field to break the cycle of nematode reproduction. Planting a crop in only one year out of three in a particular field presents a planting constraint. Behavioral constraints can also be important to the actual outcome.
|Season I||1||1||1||-||-||-||-||-||2 hectares||-|
|Season II||-||-||-||1||1||1||1||-||2 hectares||-|
|Season III||-||--||-||-||--||-||1||1||2 hectares||-|
|Labor (days per hectare)||-||-||-||-||-||-||-||-||-||-|
|Season I||340||400||435||-||-||-||-||-||400 days||Price of labor: $1 per day|
|Season II||-||-||-||400||200||175||100||-||400 days|
|Season III||-||-||-||-||-||-||150||120||400 days|
|Fertilizer (kilograms per hectare)||-||-||-||-||-||-||-||--||-||-|
|Season I||0||500||1,000||-||-||-||-||-||$1,000||Price of urea fertilizer: $0.25 per kilogram|
|Yield (tons per hectare)||2||3||3.5||2.5||0.5||0.4||20||1.0||-||-|
|Price per ton||$200||$200||$200||$200||$500||$600||$35||$160||-||-|
|Gross revenues per hectare||$400||$600||$700||$500||$250||$240||$700||$160||-||-|
|Net revenue per hectare||$65||$75||$15||-$25||$50||$65||$325||$40||-||-|
The prices farmers face for both inputs and output complete the picture of the decisionmaking environment of a farming system. Actual input prices paid and output prices received by farmers in the region-for each activity-are used to calculate the gross and net revenues shown at the bottom of table 3-6. Gross revenue per hectare is simply the yield in tons for each activity multiplied by unit values (prices per ton) for output. Cost data for inputs used in each activity, for example, seed, fertilizer, and pesticide prices, are needed to determine net revenue per hectare for each crop or cropping technique.
With these calculations made and the tableau complete, comparisons among various crop activities can readily be made. The economic conditions faced by farm households, their responses to them, and the structure of incentives are all revealed by the tableau, even in this simple version. Analysts can gauge the impact of alternative output prices for various crops, of fertilizer prices, availability of credit, and so on, and thus identify opportunities for policy initiatives that will reflect both the resource constraints facing farmers and their likely behavior in response to various economic alternatives.
In the simplified tableau shown in table 3-6, the rational farmer's decision of what crops to grow can be found by inspection and a little arithmetic. In Season I paddy rice grown with Technique 2 yields $75 per hectare in net revenue, the highest of the three possibilities. In Season II peanuts yield $65 per hectare, but cassava returns $325 per hectare if it is left in the ground for both Season II and Season III. Since corn is the only Season III alternative, its $40 net revenue per hectare can be added to the $65 from peanuts in Season II for comparison with cassava's returns. The $325 cassava returns far exceed the $105 joint returns from peanuts and corn. The net-revenue-maximizing farmer thus grows 3 tons of rice in Season I, making $75 per hectare, and 20 tons of cassava in Seasons II and III, making $325 per hectare, for a total net revenue of $800 from the two hectares of land on the farm.
With somewhat more time and effort this solution could have been obtained from a linear programming model run on a computer. For this simple but still quite interesting and revealing example, the back of an envelope is easier and faster than the computer. As complexities are added, however, or other issues need to be addressed, recourse to formal programming models and solutions may be inevitable.
Formal programming models are necessary when the tableau includes more than about ten activities and ten inputs, counting different techniques and seasons. Such complexity arises when a wide variety of potential crops can be grown, as in warm regions with year-round irrigation, and when the timing of planting, cultivating, and harvesting is intimately connected with the mix of other crops being grown and with seasonal fluctuations in labor availability and in product prices. Full programming solutions can be very helpful in understanding how farmers are likely to react to changed circumstances In such complicated environments.
A second setting that requires formal solutions is the prevalence of buying and selling activities for intermediate inputs, such as fodder or feed grains, or for labor in different seasons. Modeling these activities efficiently requires knowing some tricks of the trade, most of which are presented in the volume by Raymond Beneke and Ronald Winterboer listed in the bibliography. "Grow or purchase" decisions and on-farm versus off-farm work are not merely issues for highly commercialized agriculture, however. Chinese agricultural planners are interested in simple linear programming models that would address such questions at the commune level (and even lower).
More complicated models also become important when the simple linear relationships presumed in the tableau begin to break down. Diminishing returns to fertilizer were captured in table 3-6 by using three separate production techniques for rice cultivation. The farmer is forced, however, to choose one technique or another rather than use each over its efficient, but limited, range. Similarly, all purchased inputs are available at a fixed price, and all output can be sold for a constant price. This fixed-price environment is probably a fair representation of the individual farmer's perspective, but analysts who worry about how the results look when aggregated into market totals need to know whether strong demand for fertilizer will raise its price or whether the market price for cassava will fall below $35 per ton if every farmer grows 40 tons of it.
For the analyst, formal programming techniques have two important uses which the simple techniques in the next section cannot address. The first is the opportunity to ask whether the full set of constraints facing the farmer is actually incorporated in the tableau being used by the analyst. As additional biological and behavioral constraints are imposed on the farmer's decision-making, the simple maximization of net revenue outlined above will no longer give the "right" answer. While some analysts view the outcome of this simple maximization as a test of the farmer's "rationality," it is closer to a test of the analyst's ability to see the world through the farmer's eyes and to model what the farmer sees.
Part of the farmer's perception involves the aggregation problems raised above. Most farmers know that the market prices for many specialty crops-fruits and vegetables-are very sensitive to the quantities supplied. Through years of trial and error farmers have learned not to overplant these crops despite their apparent high profitability per hectare. Programming models that have all of India growing watermelons are not unheard of, and they measure the modeler's rationality, not the farmer's.
The second major role for formal programming solutions is the insight they provide into the value of additional units of the inputs available in fixed amounts. Linear programming solutions calculate how much net revenue will increase for a one-unit increase in each constrained input. Because these values are like implicit prices that allocate among the various fixed inputs the total net revenue produced by the optimal solution, they are often called shadow prices. These implicit input prices generate the same optimal solution using cost minimization (with costs equal to shadow prices times input use) as would the original revenue maximization solution. Since the revenue maximization procedure is normally done first, it is called the "primal solution." The shadow prices attached to the fixed inputs constitute "the dual solution."
Shadow prices from dual solutions provide important insight into the scarcity values of the inputs available in fixed or partially fixed supply at the farm level. These inputs include land or fixed capital, rationed inputs such as irrigation water for which the farmer may pay a price but still desire more, and family labor, which may not have easy access to jobs in rural markets. The shadow price attached to each of these inputs (the dual of the amount of input used in the optimal solution) indicates the value at the farm level of increasing the availability of the input by one unit. Inputs in surplus supply-for example, farm household labor in Season II on the representative farm shown in table 3-6, where only 200 of the available 400 days are utilized growing cassava-will not contribute any additional revenue if more is available. In the absence of outside hiring activities, the shadow price of such labor is zero. Conversely, in Season I the shadow price of family labor would be positive if the 400 days of labor hired from the rural labor market were not available to this farm. In fact, the shadow price of labor in Season I is about $1.17 for Technique i and $1.13 for Techniques 2 and 3 (assuming family labor is not paid an internal wage in calculating costs).
Shadow prices are also helpful in determining the value of added water supplies since an appropriate specification of water constraints at the farm level can provide planners with a full picture of how the dual solution changes when water constraints are varied. Sometimes it is necessary to distinguish between social and private costs and benefits in this exercise, a proviso that is equally appropriate for determining the marginal value of labor. These dual values also provide insight into farmers' behavior, especially their ability to allocate highly productive fixed inputs to appropriate uses. Additional constraints may also be revealed. If fertilizer has a much higher shadow price for a particular farm than the apparent market price, when actual use is inserted as a constraint, then the farm household has insufficient capital to purchase it, the fertilizer market price is not actually applicable at the farm level, or the risks of using larger quantities are perceived as being very great. Determining which answer is the relevant one requires some field investigation, but the programming results have raised the right questions.
In collecting the data appropriate to the particular farming system, the first step is to select the number of seasons or time periods to include, and the appropriate number obviously depends on the possible crop activities for a region. Two or three seasons are frequently sufficient to capture the major elements of seasonality in the system, but in irrigated areas where farmers have substantial flexibility in choosing planting dates for key crops, inputs may have to be broken down on a monthly basis to reflect the full options open to the farmer. A breakdown into shorter time periods is rarely necessary.
Eight to ten cropping activities are usually adequate, including different techniques for growing the most important commodities. Interplanted crops can be specified as a single activity; for example, corn or rice as an early intercrop with cassava represents a single activity. The input requirements, yields, and revenue would pertain to the entire mixture. Despite the importance of home gardens for improving incomes and nutritition, they cannot be modeled satisfactorily without introducing extreme complexity in the tableau design.
Input requirements can be quickly estlmated from a few published sources and later verified and modified by talking to farmers in the countryside. Having the right input coefficients-the values for a~ that indicate the amount a of input i into activity i-is clearly critical to using the tableau successfully to represent farmer behavior. To ensure that the input data are representative, interviews in a region should focus on variables such as irrigation, soils, and farm size, that determine the dominant crop activities and farming techniques. A series of group interviews will enable the analyst to inquire about the expected range and representative values for the input coefficients. Large farmers tend to dominate group meetings, and the answers will be biased unless the analyst is careful to sound out small farmers as well.
Input coefficients can also be obtained from farm management Surveys, but since such studies tend to be village censuses, policy analysts can be swamped with irrelevant data. In addition, many tabulations average across discrete activities, rather than treat them separately. For example, if half the farms use oxen and half use tractors in growing cereals, separate data are needed for each activity, not an average coefficient for both.
Synthetic activities can be constructed to reflect crops or crop techniques that might be used by a farmer but currently are not. For example, to explore the implications of introducing a power tiller when no tillers are used in a region, it will be necessary to develop synthetic data on the productivity of tillers from other areas. Similarly, investment in an irrigation system could radically alter cropping possibilities. These synthetic activities are simply added to existing crop and input activities to assess likely farmer reaction to agricultural production programs based on new technology or infrastructure. Great care must be taken in this procedure, however, not to assume the answer by using a highly optimistic coefficient of input productivity or actual on-farm availability. The trick is to see the innovation from the farmer's viewpoint and to include it in the tableau in a realistic fashion.
Defining the appropriate geographic area to be covered by a resource constraint is sometimes an issue for policy analysts. Since the tableau is used for policy planning, not for planning an individual farm, the resources or constraints might plausibly apply to the entire agroclimatic zone. While some resource constraints, such as irrigation supplies, are more clear-cut at the regional level than at the farm level, no regional decisionmaker actually chooses particular crop activities and then invests time, effort, and resources in growing them. Socialist economies frequently plan allocations of area to various crops and inputs to the respective areas, but such plans are not self-fulfilling. Farmers must still receive the inputs and the information from the plan on how, when, and where to use them. If a late monsoon delays the planting of wheat, then corn or sorghum may be more appropriate. The tableau is built not to help planners make these types of decisions, but to analyze how farm households make day-to-day decisions. For this reason, it is preferable to design the tableau to reflect decision activities of typical farms.
Cost data for the inputs, such as fertilizer, pesticides, irrigation water, and seed, are straightforward entries which can be determined by talking with farmers or farm supply depot managers in rural areas. Cost data for the primary factors, such as labor, are more difficult to collect. Even though many farms rely primarily on family labor, opportunities may exist both to hire additional farm labor if needs are great and for family members to work off the farm for wages if they are not needed (or are not very productive) on the farm. Because of these alternative opportunities, wage rates in nearby markets are needed to evaluate farmer response to new technology, altered price policy, or investments in infrastructure. Questions with regard to wage rates by season, differential wages between sexes (and children), and forms of payment need to be noted as part of the early field interviews.
Input costs can be incorporated in the tableau in several different ways. For many purposes, supplemental side listings of price and cost data, as shown in table 3-6, and solutions by inspection of the tableau for the farmer's maximum revenue choices will be adequate. But when separate activities that show purchases, sales, and transfers are added to the tableau, the farmer's best choices of activities can no longer be found by simple inspection of the various alternatives. Mathematical programming techniques are then needed to find optimal solutions to particular farming system or policy problems. Using such techniques requires expertise and practical experience if the results are not to be mechanical or trivial.
Using a Farming System Tableau for Policy Analysis
Aggregate farm output for any crop is defined by two variables: yield per hectare and area harvested. Both variables are partially under the farmer's control and partially determined by weather and other exogenous factors. The opportunities available to the farmer in both dimensions are illustrated in the tableau in table 3-6.
The important decision for the farmer in Season I is not what crop to grow but how to grow rice since no other crop is competitive enough even to enter the tableau. The three alternative techniques shown are representative of a wider, perhaps infinite, set of possibilities. Technique i uses traditional technology with no fertilizer to produce 2 tons of rice per hectare. To grow two hectares of rice using this technique, the farm household must hire 280 days of outside labor, at $1 per day, to assist the 400 days of household labor available. The 680 days of total labor cost $680, assuming the household labor is paid the market wage. The gross revenue is $400 per hectare, or $800 in total, from which the $680 in labor costs must be deducted, leaving a net revenue of $120 for the farm, or $60 per hectare as the return to management and land during Season I if Technique i is used.
Techniques 2 and 3 use more modern technology with moderate and large applications of fertilizer, respectively. The added fertilizer, in combination with greater labor inputs, raises yields to 3 tons per hectare with 500 kilograms of urea per hectare and to 3.5 tons with 1,000 kilograms per hectare. With urea fertilizer prices at $0.25 per kilogram and labor costs at $1 per day, the added yields from Technique 2 more than repay the added costs, and net revenue rises to $75 per hectare or $150 for the farm. But the additional yield from Technique 3 is not worthwhile, and net revenues fall to only $15 per hectare. The fully rational farmer will choose Technique 2 to maximize net revenue, but the additional $60 in labor costs and $125 in fertilizer costs per hectare to earn an extra $15 in net revenue may look like a risky investment to some farmers who would thus choose Technique i because of its smaller cash outlays.
In Seasons II and III the farmer must choose among several potential crops. Rice can still be grown using Technique 2, but the yield will be only 2.5 tons rather than 3 because less water is available. Soybean and peanut yields are very small, but they bring high prices. Cassava can be planted in Season II, but it produces no revenue then because it occupies the land for two full seasons, thus displacing any further crop in Season III. As noted earlier, the farmer must compare the two-season cassava revenues with the combined revenues from a single Season II crop and a single Season III crop. In this example, only corn can be grown successfully in Season III, but the cassava revenues dominate all possible combinations of corn with other Season II crops. Consequently, the farmer represented in the table 3-6 tableau chooses to grow cassava over the second two seasons, earning a net revenue of $325 per hectare.
The total income for the farming system is $800, not including the $900 earned by family labor during the three seasons ($400 in Season I, when family labor supplies were exhausted and 400 days of hired labor were required; $200 in Season II working on cassava; and $300 in Season III working on cassava). No additional labor income is earned when family labor is in surplus supply on the farm. This $800 net revenue is the basis for calculation of shadow prices for the fixed factors of production, and it thus serves as return to the farm household's land and management skills. This amount may seem like a satisfactory return for the family, depending on alternative uses of their land and management skills (and whether they could get jobs paying $900 per year off the farm). For example, if land of similar quality were selling for $2,000 per hectare, the $800 net return to land and management represents a 20 percent rate of return on the value of land alone.
One decision not represented in the tableau but which is important both to farm households and to policy analysts is how much of the rice and cassava produced will be retained for home consumption and how much will be sold to market intermediaries to become available for off-farm consumption or export. The additional steps needed to incorporate this decision into the tableau make it sufficiently complicated that formal solution techniques are needed to find the optimal answer, but the analyst's intuition can provide an estimate even without formal models. If the farmer can raise cheaper food crops for home consumption while offering more of the expensive crops for sale, the retained net revenue for the household will be greater. This topic is treated in more detail in chapter 4.
Modeling the farmer's static choices of cropping patterns and input use is more a test of the analyst's ability to calibrate the tableau properly than a test of the farmer's rationality, although the two must be decided jointly. The usefulness of the tableau becomes apparent when parameters influencing the farmer's decision-making environment begin to change. How do farmers respond when output or input prices vary or technical change occurs? The tableau is designed to help analysts address these questions.
Most governments have the potential to influence the prices for basic food crops with import and export controls, floor and ceiling price policies implemented with buffer stocks, or even special foreign exchange rates. Such price policies are closely examined in chapter 4. Needed here is some sense of how farmers might react to them.
The tableau in table 3-6 can be used to address this question. If labor and fertilizer costs are held constant and rice prices are raised progressively by a policy intervention from $180 per ton to $325 per ton, the farm household represented in the tableau will change rice production decisions according to each new net revenue calculation. At $180 per ton, the farmer uses Technique i to produce 4 tons of rice. From $185 to $320 per ton, the farmer uses Technique 2 and produces 6 tons of rice. At $320 per ton, the farmer switches to Technique 3 and produces 7 tons of rice. All of this supply response comes during Season I from higher yields achieved by more intensive labor and fertilizer use, an intensity made profitable by higher rice prices.
If rice prices continue to climb, the farmer begins to look at Season II. Growing rice in this season had been absolutely unprofitable at $200 per ton, and it was even less desirable relative to the large income-earning potential of two-season cassava. If rice prices exceed $324 per ton, however, the farmer will switch to growing rice in Season II and produce an additional 5 tons of output from the entire farm, making a total of 12 tons of rice produced for the year. In addition, 2 tons of corn will be grown in Season III, and 40 tons of cassava output will be lost. This supply response can be plotted as a "normative" supply curve, as in figure 3-11.
This supply curve, with its abrupt jumps in output at critical price levels, does not look much like the smooth supply curve drawn in figure 3-6 in correspondence to the short-run marginal cost curve. The jagged supply curve generated from the tableau shows what farmers "should" do if they make precise calculations whenever prices change and then switch immediately to
he newly profitable technique and level of input use. Risk aversion and an understandable desire to experiment on a small plot or with partial changes in input use, behavior which is not formally incorporated in the tableau, will make observed supply changes in the face of rice price changes less abrupt and, no doubt, somewhat less elastic in the short run. However, the supply function in figure 3-11 is a good indicator of the likely direction of change when the farm household's environment is altered, and it provides an upper bound to the potential magnitude of the response.
Several features should be noted about such a normative supply curve generated from a farming system tableau. The illustrated supply curve is for one cropping system only. To obtain a national supply curve, two steps are needed: aggregation of individual farms represented by several tableaux into a regional total, and aggregation of regional totals into a national level. For example, if there are four representative farms per agroclimatic zone and six zones in the country, an assessment would be required of twenty-four different farming systems. This is a sizable number although many systems might show similar responses. In table 3-6, increases in rice output in response to rice prices are caused primarily by increased fertilizer use. The area planted to rice and other crops in the system is affected only after a substantial rise in rice prices. Knowing the general mechanisms that lead to supply response and the relative weights of the various farming systems in national farm production is usually sufficient for policy analysts to form rough judgments about farmer responses to output price changes.
The farmer response to higher rice prices also illustrates the concept of cross-elasticities of supply, which are analogous to cross-elasticities of demand. In the tableau example, increasing the supply of rice from 7 to 12 tons by raising the price of rice above $324 per ton is done at the expense of cassava output. Cassava is also an important starchy staple, especially among the poor in East Java, and the expanded rice output might have negative nutritional consequences despite the larger supply of rice available. If the expansion of rice production increased hiring of rural labor and wages rose, the poor who eat cassava might be better off, because of their higher incomes, despite the smaller Supply of cassava available. If most of the added labor comes from the farm family household in Season II, the rural landless poor could then be significantly worse-off. At rice prices above $324 per ton, corn production in Season III expands and is thus complementary with rice production in Season II.
A third complication not reflected in the supply curve in figure 3-11 is the potential interaction between quantities and prices. In the example, the increase in rice supply generated by the higher prices does not cause rice prices to fall because government policy determines the rice price. This assumption might be reasonable for small changes in quantity or for very well-managed policies. In some instances, however, increased supplies may depress prices and cause farmers to reevaluate their production decisions.
Purchased inputs produced in the modern industrial sector contribute to higher yields on farms that have appropriate environments to use them. Differential use of purchased inputs, especially fertilizer, explains much of the differential yields for important food crops around the world. Average rice yields for countries in Asia, for example, are highly correlated with fertilizer applications. These applications in turn are directly related to the farm price of rice relative to the price of fertilizer. Countries not wishing to use higher output prices as a means of stimulating food crop production understandably turn to lower prices for fertilizer and other inputs as a way to increase food supplies.
The farming system tableau in table 3-6 can be used to examine the farmer's response to changed input prices just as to changed output prices. If rice prices are $200 per ton and the price of urea is $0. 125 rather than $0.25 per kilogram, the rational farmer would switch to Technique 3 from Technique 2. More rice is produced with the lower fertilizer price, and the farmer's net income increases from $75 to $140 per hectare in Season 1. Both income to the farmer and production of rice increase. This distinction is very important, for although subsidies can be used to lower input costs in order to raise production, their main effect might be to transfer income to farm households. Subsidizing farmer incomes can be entirely legitimate when it is intended, but the analyst should understand the distinction between the effects of subsidies on production and effects on income transfers.
This distinction becomes even more important when assessing the relative merits of product price supports as opposed to input subsidies. Within the assumptions of the simplified tableau of table 3-6, the rational farmer will use the same quantity of fertilizer in Season I whether the price of paddy is $200 per ton and the price of urea is $0. 125 per kilogram or the price of paddy is $400 per ton and the price of urea is $0.25 per kilogram. In both instances the ratio of input to output prices is the same. As shown in figure 3-5, this price ratio determines optimal fertilizer use and rice yields per hectare in a simple production function model. But the income effects of the two price regimes are vastly different. In the $400/$0. 25 example, the net revenue per hectare for the farm is $715 in Season I, but in the $200/$0. 125 example, the net income is only $140 per hectare. The more than fivefold difference in net income is likely to change farmers' notions of risk, their capacity to make investments in subsequent seasons, and their entire consumption bundle. Analysis of which price policy pays greater social returns requires tracing the employment effects of such sharply different income streams. This effort requires the macro perspective of chapter 5 to complement the farm-level analysis here.
Farmers react to input and output prices within the technical environment of their farms. Changing that technical environment by providing more effective water control, new biological technology, or better mechanical equipment can change crop output significantly even within a stable price environment. For example, if the farm household in the tableau in table 3-6 were suddenly to have access to free irrigation water in Season II, rice yields from Technique 2 might rise from 2.5 to more than 4 tons per hectare (with other inputs remaining constant) because of the reduced cloud cover and greater solar energy available in the dry season. Growing a second crop of rice thus becomes profitable in combination with corn in Season III, despite the low price of rice and high price of fertilizer. Such changes in crop productivity are the major justification for investing in irrigation facilities or more rapid technical change. The farming system tableau provides a quantitative framework for examining these productivity effects.
Analysts can also make rough estimates of the production impact of other investments in rural infrastructure. For example, the effects of a new rural road, which might significantly influence farm-gate prices (especially for perishables), can be assessed by calculating existing transportation costs and estimating the costs that would prevail with a new road. With an estimate of how much marketing margins will be reduced as a result of lower transportation costs, analysts can translate the lower marketing costs into higher output prices for farmers as well as lower input prices. The new prices facing farmers are inserted in the tableau, and the analyst solves for new net revenue. If the lower marketing costs are all passed on to consumers through lower retail prices while farm-gate prices remain the same, no production impact will be felt although consumer welfare may increase significantly.
These rather simple calculations are obviously no substitute for detailed investment project reports or for the sophisticated benefit-cost procedures developed by the World Bank, and others. With a general familiarity with farming systems in the region, however, a policy analyst using these techniques is likely to be much more creative (and demanding) in the use of consultants to identify fruitful development projects.
Farmers usually argue that they are "losing money and that government price supports should cover "costs of production." Policy analysts are frequently asked to calculate these costs as a basis for price policy. It is important to understand, however, that even an individual farmer does not have a unique cost of production and that the full array of a country's farmers have widely different costs. A tableau can be used to calculate production costs and to illustrate the nature of the serious conceptual and empirical problems with such cost-of-production calculations.
The out-of-pocket costs a farmer incurs are calculated directly from the tableau. For each activity, the costs of production divided by total yield result in unit costs, that is, the average cost per ton of output. These costs specifically do not include a return for the farm household's land or managerial skills, but some assumptions about family labor costs are usually included. These family labor cost assumptions may not accurately reflect the actual opportunity cost of that labor in its next-best employment possibility unless great care has been taken to understand the dynamics of the rural labor market when the tableau was being constructed. Depending on the inputs used and the crop activities chosen, each farmer's costs of production can vary significantly.
The supply curve in figure 3-11 (or its smooth theoretical equivalent in figure 3-6) illustrates that there is no single cost of production, even for one crop produced by one farmer. At a price of paddy of $180 per ton, the cost of paddy production is $170 per ton for producing 4 tons of paddy on the two-hectare farm. With the output price at $320 per ton of paddy, the production cost per ton is $196 for the 7 tons produced, and at $325 per ton the production cost rises to $202 per ton for 12 tons produced.
The supply curve for a farm crop is directly related to its marginal cost curve, that is, the additional cost of producing additional units of output. The point at which a rational farmer chooses to be on the cost curve (or the supply function) depends not only on the price of inputs but also on the absolute and relative prices of the various crop outputs. Even for a single crop on a given farm, the cost of production is a fiction; there is only a schedule of costs and outputs. These schedules vary by farm and by agroclimatic zone. Both conceptually and empirically, therefore, the search for a single cost of production is fruitless, despite the tendency of government procurement agencies and price control boards to justify their prices on just such a basis. Various estimates over a wide range can all be correct even if the numbers are generated from reliable farm surveys. There cannot be one right answer even with perfect measurements.
Even for a single farm household growing a single crop in the context of a given price environment, calculating the cost of producing that crop is difficult when the value of family labor (or land) is included in the cost analysis. When family labor must be supplemented in peak seasons with hired labor, using the market price for labor in calculating net revenues is appropriate. But how is labor to be valued when there is surplus labor in the labor market or when family labor availability generally exceeds labor requirements?
Very high implied costs of production result from using government-set minimum wage rates to value farm labor. But smallholder agriculture is rarely affected by minimum wage laws, although plantations often comply. Since minimum wages often bear little relation to labor productivity in agriculture and are almost always higher than market wages for unskilled rural labor, market wage rates are likely to be much closer to a correct value for labor, especially from the perception of the farmer needing to hire additional workers. Market wages often vary considerably by season, being relatively higher in the planting and harvesting seasons and relatively lower when crops are growing or when meager rainfall prevents intensive agriculture. Although the tableau example uses a constant wage rate across seasons, seasonal wage rates would more accurately reflect labor costs.
Even local market wages might overstate the actual opportunity cost to the economy of hiring an additional worker or of a family member's working an additional day on the farm. In regions with surplus labor, a farmer hiring additional labor will have to pay a wage, even if it is low. Mote people may be seeking jobs at this low rate than can be hired, but the market wage falls only so far because no one is willing to work for zero wages or for less than some traditional level that covers subsistence costs. In such situations, the opportunity cost of labor-what labor is worth in its next-best occupation, such as petty trading-will be less than the market wage. Thus, the market rate will overstate the true labor cost to the economy, the shadow price of labor in the programming terminology discussed earlier, even though the market rate must be paid by an employer.
It is important for policy analysts to understand the logic of shadow prices for analysis of the macro implications of farmer decisionmaking, but their relevance to what farmers actually do in the countryside is limited. Farm households make decisions on the basis of their perceived opportunity costs. Market wage rates are a reasonable approximation for valuing both family and hired labor if the analyst has good data on wage variations by season and knows whether rural workers can find jobs at these wage levels. If they can, the labor market provides a reasonable basis for valuation. If family workers cannot find jobs, the marginal value of a family member's time may be near zero. This situation commonly occurs when the costs of migration to find part-time or seasonal work outweigh any potential earnings.
If wage labor opportunities are near zero, it is rational for farmers to exclude labor costs in making their production decisions and to calculate their return to the combined resources of land, management, and labor. The cost data calculated in such circumstances then have considerable meaning for price policy because they cover only cash costs incurred in crop production. Crop prices that are below these direct cash costs will indeed cause losses
Social Profitability Analysis
Farmers make decisions that are privately profitable, basing their calculations on the prices they actually pay. From a macroeconomic perspective these prices may be distorted for a number of reasons, including surplus labor conditions, tariff barriers, and government subsidies or taxes. Net social profitability analysis reveals how the calculations would change if all the price distortions were removed.
If rice prices are held down by trade policy, more intensive rice production will show greater profitability in social terms than in the private calculations of farmers. Alternatively, a fertilizer subsidy will make rice production (and other fertilizer-intensive crops) appear more profitable to individual farmers than to society as a whole. Net social profitability analysis weighs all such positive and negative effects to arrive at an overall judgment on the social desirability of carrying out a particular project.
Private profitability calculations are unlikely to result in a nation's best use of its resources when monopoly elements in marketing or segmentation in the labor market give price signals to farmers that cause them to misallocate resources. In addition, price distortions induced by government policies frequently cause farm prices to diverge from the opportunity costs of inputs or output. Government subsidized credit to farmers, for example, may encourage them to purchase capital equipment, thereby displacing labor even in a labor surplus economy. Various government trade policies, including taxes, subsidies, and bans on imports and exports, can cause domestic food prices to be vastly different from those prevailing internationally. Subsidy and trade policies intended to protect domestic industry raise prices of many consumer and producer goods used by farm households, thereby reducing their incomes and distorting their allocation of resources.
To measure net social profitability, the price data in a tableau such as table 3-6 are adjusted in two ways. First, output and the inputs that are traded in international markets are valued in world prices to eliminate the transfer effects caused by government policies. Output is valued at the price a country must pay for imports of a commodity (or can receive for its exports) instead of the actual market price that prevails domestically. Similarly, an input, such as fertilizer, that can be purchased or sold abroad is valued at its international cost rather than at a subsidized (or taxed) market price. International prices measure the opportunity costs of growing various crops because countries have the option of purchasing or selling goods abroad, whether or not these international markets are competitive. Identifying the international price that is relevant for the comparison is not always straightforward, however, because of short-run fluctuations in many important world commodity markets. Rough guidelines for finding appropriate opportunlty costs are provided in chapters 4 and 6.
The second adjustment to the price data requires that domestic resources, such as labor, capital, and land, be valued to reflect their social opportunity costs within the country-at the value of the output forgone from not using these resources in their next-best alternative employment. If farmers receive subsidized credit at an interest rate of 6 percent when the government could otherwise have used the capital in a development project yielding a 15 percent social rate of return, the social, or shadow, price of capital would be 15 percent.
After the two price adjustments are made, social benefits-the value of the output for each activity at the adjusted prices~an be compared with the social cost-the opportunity costs to society of using the inputs. Whether the commodity is a desirable good for consumption is a separate social question that depends on the distribution of purchasing power and a society's attempt to provide a minimum consumption floor for basic goods and services. Once consumption of any commodity is contemplated or under way, society has an obvious need to obtain the supplies as efficiently as possible. If social profitability is positive (if benefits exceed costs), it is efficient to produce the commodity instead of trading for it. The calculation of social profitability can be done for different commodities, different techniques of production, and different regions by using the data in the tableau format of table 3-6.
At the private market prices for inputs and output shown in table 3-6, the farm household grew 3 tons of rice per hectare in Season I using Technique 2 and 20 tons of cassava per hectare in Seasons II and III. Suppose, however, that those private market prices resulted from two government policy actions that subsidized rice imports by keeping the domestic price at $200 per ton while import prices were $250 per ton, and that subsidized cassava exports by keeping internal prices at $35 per ton while the export price was only $20 per ton (the cassava example especially is hypothetical). What would farmer decisionmaking be in the absence of these government policies?
This is equivalent to asking whether the two activities chosen by the farmers are socially profitable as well as privately profitable. In fact, when the social prices are inserted in the tableau and the new calculations performed, rice production increases from 3 tons to 5.5 tons per hectare as the farmer shifts to Technique 2 for growing rice in Season II, corn production increases from zero to 1.0 tons per hectare in Season III, and cassava production drops to zero, for its social profitability is negative. Although simple, the example shows clearly how government policies that influence market prices can significantly affect total food supplies, the composition of output, and even employment in the rural area.
Social profitability can diverge from private profitability in dimensions other than output prices. Table 3-7 shows the calculations of social profitability if private wage rates are different from social wage rates and if fertilizer is subsidized, thus causing private costs to diverge from international opportunity costs. By combining the total effects of each divergence, the analyst can judge the overall social profitability of each activity. Equally important, table 3-7 shows the contribution of each component of the divergence-output price policy, wage policy, input subsidy policy-to actual social profitability (or loss).
The difference between the private and social calculations in table 3 - 7 is striking. In this example, cassava, which was the main income-earner under private prices, would have a negative social profitability because the decline in private profitability from a lowered product price (- $300) and from the additional cost of fertilizer (-$125) is not offset by savings on labor (+$62).
|Private profitability||Commodity price effect||Fertilizer subsidy effect||Wage effect||Social profitability|
Column 1: Private profitability per hectare as shown in table 3-6.
Column 2: Commodity price effect, assuming the social opportunity price of rice (paddy) equals $250 per ton and the cassava price equals $20 per ton.
Column 3: Fertilizer subsidy effect, assuming an international price of $0.50 per kilogram of fertilizer as opposed to a domestic price of $0.25 per kilogram.
Column 4: Wage effect, assuming an opportunity wage of $0.75 per day as opposed to a market rate of $1.00 per day.
Column 5: Social profitability per hectare, assuming social prices and the physical coefficients of table 3-6.
Rice would be much more profitable under social prices, even though less fertilizer would be used in rice production. Rational farmers facing social rather than private prices would also have a very different combination of crop activities with peanuts and corn replacing cassava. Rice production in Season II is no longer the most profitable crop after the fertilizer subsidy is dropped. Employment would increase under social prices from 1,300 to 1,720 days. In short, government policies can have a large impact, and using social prices in a farming system tableau permits these effects to be identified and quantified.
A strategy for agricultural development within a food policy framework is broader than a simple concern for expanding farm production, important as that is for other elements to be effective. Growth in other sectors, job creation, income growth and distribution, access of poor people to food, and household and national food security are also integral components of a production strategy. The vast literature on agricultural development strategies and more recent analysis of rural development strategies have devoted considerable attention to such issues. Food policy incorporates these sectoral perspectives into a macro policy context while addressing consumption concerns.
Among the most difficult issues for agricultural sector planners are those of food self-sufficiency and comparative advantage, which inherently deal with international markets, appropriate border price policies, and foreign exchange rate management. These issues raise many of the topics treated in this book, and an integrated discussion is not possible until the last chapter. But social profitability analysis is the primary conceptual tool available to analysts to address these questions. Its ability to illuminate the issues raised by concerns for food security, self-sufficiency, rapid economic growth, and reduction of hunger is instructive at this stage.
Improving the Social Profitability of Agriculture
A major role of agricultural production policy is to reconcile differences between private and social profitability, for farmers make their decisions on the basis of the market signals they actually perceive, not those used by analysts in a planning agency. The desirability of reorienting a nation's farming systems toward socially profitable patterns is obvious. By definition, such patterns lead to more efficient resource allocation and faster growth in output. They do not, however, necessarily solve short-run problems of unemployment, poverty, and hunger. Some of these problems were addressed by the consumption interventions outlined in chapter 2, and some must await the appropriate macroeconomic environment to be discussed in chapter 5. However, the production strategy itself has some potential for alleviating these short-run problems.
To start, policy analysts need to understand what to do if social and private profitability calculations show substantial divergence. As shown in table 3-7, farm activities can have positive or negative social profits, even before a return to land is included. In some cases, private profitability greatly exceeds social profitability because a set of government policies promotes the inefficient use of domestic resources. Then crops are being grown for which a country does not have a current cost advantage. In countries especially worried about self-sufficiency in food production, governments might wish to absorb small resource costs in the interest of increasing food production at the expense of other crops. Such a decision involves assessing the tradeoffs between greater food self-sufficiency, the budgetary costs required to achieve it, the transfer of incomes from consumers to producers because of higher prices, and the efficiency losses that occur because of misallocation of resources in a narrow economic sense.
One of the most important roles of government is to ensure that the society's food supply is not subject to the whims of weather, international markets, or political blackmail. Food security is different from self-sufficiency, however, for in most countries domestic food production is even less stable than supplies available in international markets. Further, domestic food self-sufficiency within a generally interdependent world is an elusive concept. Does it mean self-sufficiency in a single staple grain, in all food, in all food-producing inputs (for example, feed grains for livestock or fertilizer for grain production), or even in all inputs to the input industries?
For a country bent on self-sufficiency, eliminating food grain imports is relatively easy. By raising grain prices high enough, consumption will be curtailed, production will be stimulated, and any import gap can be closed. Reaching self-sufficiency in this fashion, however, would surely be a hollow victory for food policy. Simply eliminating food imports does nothing to guarantee that poor people have enough to eat, and it may make matters much worse.
Most hunger is related to poverty, and so income generation through efficient employment creation is an important component of any strategy designed to improve household food security. Whether such employment creation leads to greater food self-sufficiency is obviously not a question of food supplies alone, but rather depends on the social profitability of increasing those supplies from domestic production. Policymakers may rightly value the higher rural incomes from increased domestic production somewhat more than the lower cost of similar food from imports and may feel that such production adds to the society's sense of food security. Beyond a 10 to 20 percent premium paid for that increased production, however, the wasted resources have very high opportunity costs for a poor country.
The argument that producing food should be a society's first priority until hunger is eliminated, after which diversification into cash crops can be permitted, has an emotional appeal when strawberries are being exported while landless peasants starve. The appropriate balance of crops-between cash crops and food crops and between cereals and legumes-has several dimensions in addition to the apparently simple question of whether all farm households are producing their own food first.
Several important crops are needed as industrial inputs, and increased cotton or jute output, for example, might permit more and higher-paying jobs in the industrial sector. The productivity of these jobs is a major factor in the social profitability of growing such crops. Farmers whose incomes are increased by growing cotton might be much worse off if forced to grow corn, and the greater home production of food might not offset the worsened poverty. At the same time, pushing tenants off the land on which they raised food for their families in order to grow export crops with mechanized farming techniques may contribute to significant rural hunger. The important issue, however, is not the nature of the crop being grown but the size of the income stream from growing the crop and the recipients of that income.
Efficient and widespread income generation is the most important role for any economy. From income stems the consumer S freedom to purchase a variety of desired goods; from lack of income stem restricted food purchases, hunger, and malnutrition. If a food system is creating many new jobs accessible to the ranks of rural workers dependent on wages for their livelihood, it is a success almost without attention to the composition of output. Well-paying jobs cutting sugarcane or carnations for export are superior to the hunger of marginal subsistence farming. It may be more desirable to have a well-fed, self-sufficient peasantry with adequate land to feed, clothe, and educate a family, but in many parts of the world, especially in much of Asia, this is not feasible. More productive jobs in the agricultural sector are then the only realistic escape route from rural poverty.
Rural jobs can be created in a wide variety of ways. They can come directly from a project that invests in a new agricultural endeavor, such as a sugar factory, an oil palm plantation, or an intensive livestock feeding operation. Many rural jobs are created, however, in a more roundabout fashion. Derived demand for labor is generated by the expenditure of income from basic agricultural production. Farmers consume some of their own produce directly with little downstream ripple effect on employment, but they also buy many goods and services from rural markets. Depending on the relative prices of those goods and services available for purchase in rural areas, and hence on the composition of demand, the secondary employment impact can be substantial. A strategy that pumps significant purchasing power into rural areas through incentive prices for agricultural produce can have a large, secondary impact on employment generation if other policies on industrial prices and wages are favorable to labor-intensive production of goods and services.
Incentive price policies also influence the second major source of employment in rural areas, wage labor on farms. An incentives-led agricultural development strategy encourages rapid growth in wage labor, and hence in jobs available to the rural landless, if macro policies and agricultural development programs do not offset this impact. Low interest rates, overvalued exchange rates, and direct subsidies to tractors and other labor-saving machinery, for example, can counteract the growth in demand for agricultural labor from higher agricultural prices.
Determining the appropriate degree of mechanization is very complicated. Tractors, for example, can be important to agricultural development in some circumstances. They might increase yields and absorb labor because more timely farm operations might make multiple cropping feasible. Such effects will make them privately profitable without the need for government subsidies. Just as subsidies for fertilizer induced the representative farm household in the tableau to use more fertilizer than was socially optimal, so subsidies to tractors cause farmers to use more tractors than is socially profitable. This is likely to increase their labor-displacing effects and reduce rural employment.
As the social profitability analysis showed, subsidies that distort prices have both production and income effects at the farm level. When important choices of production technique are also affected, subsidies and price distortions can have powerful consequences for employment and income distribution. Government policies in many developing countries have tended to price domestic food below international levels, to favor industry over agrIculture, to favor export crops over food grain production, and to favor capital-intensive techniques relative to the employment of labor. The price distortions resulting from such policies have important consequences for the level and composition of agricultural output, rural employment and incomes, and the degree and distribution of hunger.
Social profitability analysis can illuminate and quantify some of these consequences. Not all the information or insight needed to do this analysis is yet in hand. Which prices to use and how to incorporate other food policy objectives into the analysis are topics for the next three chapters. But the social profitability analysis of actual decisions farmers must make year in and year out provides the essential foundation for an agricultural sector strategy that will be consistent with broader food policy objectives.
ELements of a Production Strategy
Four major lessons emerge from the farm decisionmaking analysis and are likely to influence most production strategies, especially when they are incorporated into a broader and consistent food strategy. These lessons include the desirability of broadly based programs for small farm households, the need for government policy to foster appropriate price incentives that increase agricultural output and to generate rapid increases in rural incomes, the importance of technical change for raising productivity and keeping food prices to consumers within reasonable limits, and the efficiency to be achieved by using international markets both as a source of gains from trade and as a measure of opportunity costs in policy deliberations when short-run trade is ruled out for other reasons.
Countries that have emphasized broadly based programs for small farmers have been more successful in achieving both their production and their consumption goals. Bimodal rural systems with a few large, modern farms and many small farms have sometimes achieved agricultural growth, but most have perpetuated or even exacerbated widespread poverty in the countryside. This poverty is the major constraint on solving the problems of hunger and on using new technology to increase agricultural productivity in the long run.
The need to make rapid and multiple decisions in field after field and day after day has also made centralized state farms difficult to manage. Socialist economies have found collective or communal farms less productive than private plots except where farm managers and workers have receive4 clear incentives to improve efficiency of input use. In such circumstances, however, the incentives tend to skew the distribution of income and expose households to greater risk than if output were shared more equally. This tradeoff between incentives to produce efficiently and the distribution of returns, especially to poorer and highly risk-averse households, is not merely a problem for socialist economies, however. It contributes to the basic food price dilemma in market-oriented economies as well.
In these economies, farming systems with a large number of relatively small-scale farms have more effectively generated rural income and achieved a more equal income distribution than have systems of large farms. Given the decentralized nature of agriculture, government policies with respect to prices, both received and paid by farmers, are a crucial element in attempts to create a dynamic rural society. Even in the short run, price responsiveness can be important quantitatively. For the longer run, a continuation of "cheap food" policies, as pursued by many developing countries, is likely to have severely negative production effects. With an appropriate set of price incentives, a country can benefit from a decentralized system of management, where many individual farm households respond to changed economic conditions. Since farmers reap the rewards of good management, income incentives are important in all agricultural systems, both market-oriented and centrally planned.
The profitability of food production is related to technology as well as to prices. Outward shifts in the supply curve arising from technical change are more important to increasing agricultural productivity than are movements along it. Increasing productivity is a primary mechanism for sustaining longer-run profitability in agriculture without having to resort to higher food prices for consumers. For farmers to adopt new technology in response to price incentives, the improved technology must actually be available and appropriate to the ecological setting. Governments play an important role in developing new irrigation systems, fostering research to develop improved, locally adapted seed varieties, and investing in rural infrastructure and marketing to facilitate the flow of productive inputs and output. In the absence of such technical improvements in farming systems constrained by traditional technology, government policies can have only a limited effect in raising agricultural productivity.
Interaction with international markets provides a standard of efficiency for both domestic industry and agriculture. For poor societies, such efficiency is critical to the mobilization of domestic resources to cope with poverty and hunger. Attempts by countries to become completely autarkic and to insulate themselves from international prices have often led to severe price distortions and disincentives within agriculture, with a stagnant rural economy as a result. For every country that followed international price signals too closely and experienced roller coaster instability, there must be ten countries that have not read the signals closely enough and are saddled with inefficient and stagnant rural sectors. Both types of countries are likely to face significant problems of hunger in rural and urban areas, and both need to seek the middle ground defined by a broader food policy perspective.
This broader food policy approach includes reading long-run international market trends and using the signals to measure the efficiency of domestic price policy initiatives. It also includes careful attention to the domestic food marketing sector, which is the primary carrier of both price signals and food commodities from producers to consumers. For many commodities, price formation itself takes place in domestic markets, and these prices influence farmers and consumers and the options available to policymakers as they try to alter decisions made by both. Because of the marketing sector's role in generating and signaling prices, the food policy discussion is inevitably broadened to include other prices important in rural decisionmaking, especially wage rates, interest rates, and foreign exchange rates. Policy for farmers must fit within this broader marketing and macro context. At the same time, however, farm productivity fundamentally conditions the options available to policymakers to achieve a wide range of food policy objectives, including the reduction of hunger.
Much of the vast literature in agricultural economics deals with the analysis of agricultural production systems. The analytical base connecting neoclassical economics with rural resource allocation appears in the classic book by Earl 0. Heady, Economics of Agricultural Production and Resource Use (Englewood Cliffs, N.J.: Prentice-Hall, 1952). More recent and accessible treatments of how farmers allocate resources are provided by John Doll, V. James Rhodes, and Jerry West, Economics of Agricultural Production, Markets, and Policy (Homewood, Ill.: Richard D. Irwin, 1968), and by a set of essays collected by Bock Thiam Tan, Kamphol Adulavidhaya, Indirjit J. Singh, John C. Flinn, and Shao-er Ong, eds., Improving Farm Management Teaching in Asia (Bangkok: Agricultural Development Council, 1980). A thorough discussion of social profitability analysis as applied to agriculture is contained in Scott R. Pearson, J. Dirck Stryker, Charles P. Humphreys, and others, Rice in West Africa: Policy and Economics (Stanford, Calif.: Stanford University Press, 1981).
Issues involving risk in the agricultural production process are covered extensively in volumes by Jock Anderson, John Dillon, and Brian Hardaker, Agricultural Decision Analysis (Ames: Iowa State University Press, 1977), and by James A. Roumasset, Jean-Marc Boussard, and Indirjit J. Singh, Risk, Uncertainty and Agricultural Development (Laguna: Southeast Asian Regional Center for Graduate Study and Research in Agriculture, 1979). Both of these books require mathematical competence.
Several excellent books deal specifically with the application of linear programming to decision problems within agriculture. A useful introductory discussion is provided in chapters 2 and 3 of R. C. Agrawal and Earl 0. Heady, Operations Research Methods for Agricultural Decisions (Ames: Iowa State University Press, 1972). Researchers interested in solution techniques will find the analysis by Raymond R. Beneke and Ronald Winterboer, Linear Programming Applications to Agriculture (Ames: Iowa State University Press, 1973), an essential reference. Several examples of policy application of programming methodology are contained in the collection of essays on Pakistan edited by Carl H. Cotsch, "Linear Programming and Agricultural Policy: Micro Studies of the Pakistan Punjab," Food Research Institute Studies, vol. 14, no.1(1975). Finally, a state-of-the-art volume edited by Louis M. Goreux and Alan Manne, Multi-Level Planning: Case Studies in Mexico (Amsterdam: North-Holland, 1973), provides an illustration of how a series of programming tableaux can be connected across farms and regions.
There is an extensive bibliography on supply functions and the price responsiveness of farmers. The initial formulation and empirical application of the Nerlovian distributed lag adjustment model is in Marc Nerlove, The Dynamics of Supply: Estimation of Farmer's Response to Price (Baltimore, Md.: Johns Hopkins University Press, 1958). An excellent review of the earlier literature is provided in an essay by Raj Krishna, "Agricultural Price Policy and Economic Development," in Herman M. Southworth and Bruce F. Johnston, eds., Agricultural Development and Economic Growth (Ithaca, N.Y.: Cornell University Press, 1967). Recent case studies by the World Bank also illustrate various pricing principles, as in Lucio C. Reca, Argentina: Country Case Study of Agricultural Prices, Taxes, and Subsidies, World Bank Staff Working Paper no.386 (Washington, D.C., 1980), and Carl H. Gotsch and Gilbert Brown, Prices, Taxes, and Subsidies in Pakistan Agriculture, 1960-1976, World Bank Staff Working Paper no.387 (Washington, D.C., 1980). The best summary of empirical estimates of supply elasticities can be found in Hossein Askari and John T. Cummings, Agricultural Supply Response: A Survey of the Econometric Evidence (New York: Praeger, 1976). A more theoretical treatment of response is contained in the book by John Dillon, The Analysis of Response in Crop and Livestock Production, 2d ed. (Sydney: Pergamon, 1977). A provocative essay by Willis Peterson, "International Farm Prices and the Social Cost of Cheap Food," American Journal of Agricultural Economics, vol.61, no.1 (1979), and a collection of essays edited by Theodore W. Schultz, Distortions of Agricultural Incentives (Bloomington: Indiana University Press, 1978), offer new evidence on the aggregate price elasticity of supply for agriculture.
Although this book on food policy does not deal extensively with benefitcost techniques for investment projects, agricultural policy analysts will find helpful the manual on project analysis prepared by J. Price Gittinger, Economic Analysis of Agricultural Projects, 2d ed. (Baltimore, Md.: Johns Hopkins University Press, 1982). The Gittinger book, especially if supplemented with additional case materials from the Economics Development Institute of the World Bank, illustrates important production policy issues including shadow prices, discounting, and with/without calculations. Several methodological approaches to dealing with choices of production technique, in the context of case studies, appear in C. Peter Timmer and others, Choice of Technique in Developing Countries: Some Cautionary Tales, Occasional Paper no. 32 (Cambridge, Mass.: Harvard Center for International Affairs, 1975).
A huge literature exists on agricultural development strategies. Some of the basic contributions in this field are Theodore W. Schultz, Transforming Traditional Agriculture (New Haven, Conn.: Yale University Press, 1964); Arthur T. Mosher, Getting Agriculture Moving (New York: Praeger, 1966); John W. Mellor, The Economics of Agricultural Development (Ithaca, N.Y.: Cornell University Press, 1966); Clifton W. Wharton, Jr., Subsistence Agriculture and Economic Development (Chicago: Aldine, 1969); and Yujiro Hayami and Vernon Ruttan, Agricultural Development: An International Perspective (Baltimore, Md.: Johns Hopkins University Press, 1972). The book by Uma J. Lele, The Design of Rural Development: Lessons from Africa (Baltimore, Md.: Johns Hopkins University Press, 1975), summarizes much of the World Bank's experience in Africa trying to broaden production strategies to include rural welfare issues. Bruce F. Johnston and Peter Kilby, Agriculture and Structural Transformation: Economic Strategies in Late-Developing Countries (New York: Oxford University Press, 1975), discuss bimodal and unimodal agricultural development strategies and emphasize the importance of technical change within a broader strategic vision. This strategic perspective is developed further and broadened in its scope in Bruce F. Johnston and William C. Clark, Redesigning Rural Development: A Strategic Perspective (Baltimore, Md.: Johns Hopkins University Press, 1982). An aggressive small-farmer development strategy is articulated in Sterling Wortman and Ralph W. Cummings, Jr., To Feed This World: The Challenge and the Strategy (Baltimore, Md.: Johns Hopkins University Press, 1978). The role of agriculture as a resource reservoir is summarized in the introduction to Lloyd G. Reynolds, ed., Agriculture in Development Theory (New Haven, Conn.: Yale University Press, 1975).
A small book by Arthur T. Mosher, An Introduction to Agricultural Extension (Singapore: Agricultural Development Council, Singapore University Press, 1978), is extremely useful. Data collection procedures relevant for the Mosher framework are covered in a new manual published by the International Maize and Wheat Improvement Center (CIMMYT), Planning Techniques Appropriate to Farmers: Concepts and Procedures (Mexico City, 1980). The international data needed to supplement the analysis of this chapter can be found in many sources, but the monthly Food Outlook series of the Food and Agriculture Organization of the United Nations is among the most helpful.