3.1   Level of Analysis & Transfers

A technology which pays at the farm level may not pay at a regional level. For example, planting trees in Nepal to prevent erosion deprives Bangladesh of silt. These are not included in a cost-benefit analysis, the purpose of which is to determine performance of resource use.

3.2   Long term and environmental impacts

Ecological critiques are concerned with damaging effects of unrestrained competition on public goods and consider all biological entities to have moral consideration. Physical outputs are only one sub-set of the system. Certain assumptions of economic valuation and environmental change are rejected, including:

• Money can exactly substitute for some given decline, in air quality, leave the individual no worse off than before.
• Different kinds of impact are commensurable.
• Environmental goods of equal value may be substituted for each other with no loss of welfare.

Long-term benefits of resource conservation are minimised in discounting calculations. Given that knowledge of ecological process is poor in detail, use of the environment should be moderated by an awareness of risk, a 'precautionary principle' (Edwards-Jones, 2000).

'Steady state' models, used for resource-conserving interventions, assume a requirement to maintain the resource base regardless of discounting. Decisions not to discount are made through policy, after considering all information that analysts provide, including opportunity costs and trade-offs ~ Christopherson, 1992.

3.3   Optimisation and Aggregate Effects

The availability and opportunity cost of each resource is used to determine which resource is to be optimised. Thus, if land is relatively cheap and labour expensive, one would choose to use labour more efficiently than land. But, should quantative indicators alone be relied upon to make decisions, or should resources be organised to satisfy strategic objectives? For example, where women are target beneficiaries, equity appraisal, may reveal that integrating trees into home gardens may be more appropriate than introducing windbreaks.

In the Java Social Forestry Program, it is precisely those participants most in need who tend to fall below the average level of returns. Aggregate economic effects can be mistakenly taken as evidence of overall programme success in meeting social welfare goals (Sunderlin, W., 1990). A need exists to stratify farmers' groups to assess allocation of resources, and differences in financial benefits.

3.4   Strategic decisions

Scale of Impact Assessment and type of analysis: Priorities for groups at each level ( household, local, regional, global) are not mutually exclusive, but indicators are chosen to match the objectives of the decision-makers, upon which adoption of a project depends. Indicators guide data collection and interpretation, such as micro or macro analysis, economic

Priorities at household and local levels feature household profits, deforestation, erosion control, sedimentation, soil fertility, water cycling. National and regional policy-makers' goals relate to productivity, income and foreign exchange, nutrition, food security, employment, distributions of income, land, and social structure. Donors desire rapid and widespread impact on sustainable food production and basic commodities in less developed regions. An example of global impact assessment is within the framework of an 'Alternatives to Slash and Burn' programme which (1: ICRAF, 2000,) includes increased carbon sequestration and enhanced biodiversity conservation.

3.5   Subsistence economies

Raynor, (1992) reports an AF system which is extremely productive and socio-culturally viable in the context of a subsistence economy, but not in the context of the transition to a market orientated economy. Pohn Pei Island, Micronesia, with an emphasis on community, expands the criteria of value in analysis. Community relationships lie outside the market system; their impact is real. Comparisons of other land-use options should be made within household livelihood strategies. (Reiche, 1992). Social welfare measured in terms of consumption and production may inhibit social sources of value by focussing narrowly on utility and individual choice. Simplistic cost benefit calculations fail to capture the goals and motivations of local participants.

3.6   Assessing Impacts on multiple levels

In order to assess the impacts of research in multiple dimensions and communicate that information to decision-makers and stakeholders Trade-Off Analysis (TOA) links inputs and outputs from the various disciplinary simulation models on a site-specific basis (Yanggen et al., 2002).

Scientists in AF programmes come from the disciplines of economics, anthropology, ecology, and systems agronomy. Constraints in implementing standardised methods exist with conflicting objectives of researchers. With several objectives at different levels, and alternative technologies to consider, a multiple objective modelling approach was advocated by Wojtkowski (1988), integrating socio-economic models with biophysical models. Minimum displacements from the ideal, within a hierarchal approach is advocated (Fawcett R. at al. 1997). 'Multi Criteria Appraisal' identifies a set of objectives, and weights criteria in order to identify solutions. It assumes a level of similarity between techniques, which in fact depend of the value judgments of those involved in the analysis.

The Participatory Rural Appraisal approach concentrates on capacity building for problem solving at the local level. The emphasis is on adaptive planning, and a focus on context specific thought rather than a search for universals.

4.1   Context

Socioeconomic analysis addresses four broad questions; Diagnosis and design of solutions as agroforestry alternatives to farmers' problems can be defined in the context of ecozone, farm, or plot levels, in order to determine (Avila, M, 1992):

• impact upon agriculture and the environment,
• land-use systems and constraints of labour and biophysical kind
• Identify household priorities
• market forces and incentives

4.2   Ex-ante analysis

Ex-ante data are useful in broadly defining objectives for regions. ICRAF's Natural Resources Strategies and Policies Programme 2000 (1) conducts ex ante analyses on potential interventions, to set research and development priorities. Good ex-ante economic analysis requires good data. Detailed monitoring is costly but decisions made upon inadequate analyses lead to high incidences of failure, more costly than relevant monitoring (Christopherson, 1992).

Comprehensive assessment has been inconsistent (Scher and Muller 1990). Inconsistent methods among projects limit cross comparisons and hence collection of baseline data. Reiche (1992), confirms the practical value of developing standardised field-based bio-economic coefficients ('minimum sets)' for agroforestry technologies. Benchline yield data needs improvement, eg. data for trees is usually from forest plantations, and not dispersed trees.

Sector analysis, describing regional structure and variation of AF production, use and markets provide context, but are notably lacking (Christopherson, 1992).

A secondary data assessment at the beginning of the trial/analysis will enhance the overall quality of the evaluation (Mercer 1993). This alone, however, is not sufficient. The main objective of research and development is to improve efficiency and productivity of the technology. Complex AF technologies may have to be optimised locally in order to generate meaningful production coefficients.

Off-farm trials allows identification technical potentials but constraints exist to the use of experimental data. Growth and yield data do not reflect conditions in the field, commonly overestimating by 20% to 40% (Christopherson, 1992). Growth is related to soil profiles, but local soil may vary considerably from one plot to another. Nitrogen fixation may be best drawn from field data. Data on soil erosion, nutrition, digestibility and response to levels of management intensity may be better collected under controlled situations.

4.3   Assessment on farms

Technologies may be affected by land-use systems, labour constraints, or agro-ecological constraints. Realistic farm management and market data needed. Trials on farmer's fields better represent target households, their farming systems and local constraints to performance criteria.

Local variations include distribution of miccorhiza, willingness, and seasonal availability of labour. Tree species should be suited to the local area. Food preferences of animals and humans should be considered. Where agroforestry fields are close to the farm house, animals may be easier to watch. There is a need to understand the needs of the animal; close by the sea, salt may be licked off tree trunks, damaging bark. Laws may treat AF products as public commodities, excluding tenants from rights to harvest.

Because existing AF systems may be improved with greater ease and success than new ones, it is useful to evaluate their management and performance in all their complexity. Farmers may be surveyed during field reconnaissance, AF activities may be collected and compared to benchmark data, so far as limitations permit.

Case studies are useful for identifying distributions of a land, particular variables or social groups, practises, income, management techniques,and plot yields. Panel surveys can capture variations over time. Studies may be exploratory, or extrapolative, in which case they must be selected for representativeness. Understanding of the system is required in order to design surveys. Given the variability between farms, correlations between variables cannot necessarily be explained. Sampling can be difficult. Local extension people and farmers may not view local problems the same way. Focus groups and informal interviews with farmers could determine whether this is the case or not.

5.1   Prices

Not necessarily reasonable indicators of value. Prices signal scarcity, but only when markets for goods are functioning properly, and competition is imperfect if suppliers (or governments) have power to set prices.

5.2   Price Trends

Farmers respond to market prices. Seasonal variations and cycles may be determined, necessitating use of local price trends. Local diversification will have consequences for local and regional market supply and demand, entailing prediction of trends at all levels. At the subsistence level, diversification increases flexibility as well as resources, which is hard to assign meaningful integer value.

Labour Ascribing realistic labour costs is tricky (Von Platen, 1992). Co-operative labour exchange groups (eg Kombit in Haiti) and family labour supply labour a times of low demand, and enable jobs to be easily accomplished. Costs are not included when harvest is achieved with excess labour. As regards forest-pasture, a farmer may act differently from a company because living on the farm it is reasonable to charge oneself nothing considering opportunity costs. Thus in winter, when no other demands exist for labour, in temperate environments pruning may be done. Non family labour costs would reduce net returns.

5.3   Stock

Seedlings performance is variable, and may have hidden costs. Clonal material, though uniform, is susceptible to disease. Transport of seedlings has an injurious effect upon viability. Spatial, performance and management criteria depend upon stock. For example, Teak trees grown in Java from clonal material give less bushy trees, need less pruning, have less knotting, and permit more light for the understorey crop. Initial planting at low density of good seed may be cheaper long term. Tree nurseries may be necessary to kick start a project. Grazers Putting animals in with trees can increase basal area growth by 25% prior to canopy closure, and their presence or absence must be accounted for. Animals may have multiple uses, including power for tillage.

5.4   Non-marketable commodities

What is a product? Is a root stump, used for fuel wood a product? Minor, non-market things may be more valuable locally, because they are 'there now' and meet daily needs. Eg. banana leaves as plates.

5.5   Non wood forest products (nwfp)

NWFP meet household needs and market potential. Examples are; oils/tannins, waxes, medicines, spices, flavourings, honey. Street, D. (1992) relates how participants presented problems in quantifying costs and benefits of woodlot and borders. Comparisons of yield depend on comparative technology, for example, charcoal production and traditional kilns.

Yields may be variable, unstandardised (Eg. Poles, fuel wood and commercial fodder), and depend upon maturity of the system. Where nwfp are not offered by agricultural production, comparisons with exportable agricultural produce are misleading in terms of value. Local market values for Af products may exist in competition with unmanaged harvests of primary forest, natural re-growth, or forests on common land. Estimates of poles can be made on stumpage prices, but more appropriate is soil expectation value Soil Expectation Value.

5.6   Non-market values

Many environmental and social impacts are not traded and therefore have no explicit market price. Soil conservation may, in addition to increased production, benefit sustained production. Such benefits, which are not product, may be added to the resource in analysis, but are difficult to calculate. Monoculture options do not take into account soil conservation, and comparisons cannot be made in this respect. Items difficult to quantify can be qualitively valued, and user -weighted. Examples are physical protection (sun, wind, animals, thieves), aesthetics, land reclamation, live-staking, and soil conservation, environmental impacts.

Common criteria for environmental impacts are extent and frequency, reversibility, possibility of reduction, social acceptance, legal limits, future developments in the same environment. Impacts that cannot be be converted into money are not supposed to signal that they are insignificant, but there is an undeniable tendency for this to happen (Edwards-Jones et al., 2000). The benefit due to micro-climatic changes and diversity, which offer protection from threats such as drought and pests, may be subsumed in yield increases, but data must be based on long-term monitoring. These benefits should not be counted twice by assigning values according to non-tangible benefits.

5.7   Analytic Efficiency

AF systems are complex, but analytical efficiency is improved with bio-economic modelling (Christopherson, 1992). The models can employ ex-ante or post-ante data. (Thomas, 1994). Production functions and budgeting data are integrated to indicate the relative magnitude of the outputs and financial trade-offs over time.

Alternative designs of the same system, contrasting AF systems, or monocultures may be compared. Users have been put off by the difficulty in providing numerous values needed to tune the models for specific soil, crop, tree, climate, and management regime. Development of common libraries of soil/tree/crop parameter values makes things easier.

5.8   Discounted cash flows & Long term analysis

Farmers' desire for immediate returns to labour is a case of high discounting. An argument for a low interest rates is that farmers have in fact a long term view, the only explanation for farmers' perpetual desire to purchase more land, accumulate livestock, or educate their children, even when such investments may not yield the highest returns. High interest rates for poor farmers with short-term planning horizons would screen out poor projects, perhaps. Low interest rates would enhance long term objectives, increase uptake of AF technologies, but if generally applied, may have adverse environmental consequences in agriculture, encouraging wide investment in resources with cheap capital.

Farmers in Mexico saw agroforestry as a way to diversify outputs and combine short-term returns, (i.e. crops), with long-term investment, i.e. timber production (Casey et al., 1999). The combined short and long-term investment makes it difficult to assign an appropriate uniform discount rate.

Comparisons of alternatives for opportunity costs can be made with either traditional or improved systems, with different results. For example, trees which are not N-fixers, Ash pasture treated with low inputs of fertiliser at about 200 units per ha. tends to optimise production. Comparison could be made with improved systems.

6.1   Evaluating Farmer preferences

A positive on-farm economic analysis provides a necessary, but not sufficient, indication of the successful introduction of an agroforestry project. Technologies should address problems in such a way that farmers will actually consider implementing them within their overall production strategy. Evaluation of any technology must begin with understanding relevance, impacts and implications at the household/farm level (Avila M. 1992). For example, farmers may have a preference for flexibility in planting schedules, having seasonal trends in available labour.

6.2   Ability to choose technical alternatives

Seven or eight sets of land use choices appeared to be the limit for those farmers in Casey et al.'s farmers survey (1999). They seemed to consider the entire system, and not just focus in on one attribute. Eight attributes seemed to be too much information for farmers to evaluate at once, and this to was narrowed to five attributes:

• The extra number of workdays to manage the system,
• Years of technical assistance the farmer will receive,
• Types of products from each system,
• Availability of seedlings
• The environmental effects

How choices are to be limited, is not clear. Analysis of what farmers are doing with trees, which trees, and rationale, provides information (Avila, M., 1992). Farmers could limit their own choice in a task of ranking attributes. Illiterate farmers, unable to scrutinise choices at their leisure, may be assisted with pictorial matrices devised for the purpose. Conjoint analysis permits the measurement of preferences on several attributes when surveying farmers.

6.3   Manageability or Risk

• Lack of resources, lack of (transportation, water, technical training, farmer and technical experts), and neglect of management regimes adds to risk of failure.
• A survey of local infrastructure is required, in addition to farm surveys.
• Limits to flexibility may exist in switching inter-crops under trees.
• Less perishable products, wood, allow the farmer to assume some risk management, harvesting and selling when prices are most favourable.
• Smooth operation of the project can hindered under political and social instability (Street, 1992).