Chapter 3
Basic economic principles governing project appraisal and evaluation

Traditionally, the economic analysis of a project has been undertaken last in a series of studies covering the technical, institutionalorganizational- managerial, social, commercial-marketing and financial aspects (Gittinger, 1982). For the tsetse and trypanosomiasis problem, this approach has recently been formalized with the development of SITE analysis (Doran and Van den Bossche, 2000); SITE is a process for screening strategy options by the four criteria on which the acronym is based:

• Socio-economic

• Institutional

• Technical

• Environmental.

The various options for intervention are then scored and ranked according to these criteria, and conflicts between the results for the different criteria explored. The remit of this paper obviously falls within the socio-economic component. There are a variety of techniques for analysing the economics of interventions in the field of agriculture and livestock production, which have been summarized in the animal health context by Rushton, Thornton and Otte (1999), the possible approaches are also discussed, with specific reference to parasitic diseases of livestock, in Perry and Randolph (1999). The technique that has been most used in the past, and which is favoured by many of the authors in Perry (1999) is some form of social benefit - cost analysis. This can be underpinned as appropriate by the use of a herd model simulating output from the livestock population with the project being implemented and consequently with improved production parameters, and comparing this to the situation in the absence of the project. Integrating epidemiological with economic models is also very helpful, particularly for a vector-borne disease such as trypanosomiasis (see McDermott and Coleman, 2001). Perry and Randolph (1999) emphasize the need to:

• “integrate the products of good epidemiological studies into economic frameworks”;

• “integrate techniques for economic analysis and simulation models of animal production and health dynamics within a systems framework”.

Published textbooks on the evaluation of animal health programmes, such as Putt et al. (1987) and Dijkhuizen and Morris (1997) also support this approach. It remains the most practical tool for analysing and ranking projects according to the relationship between their costs and their expected impact.

At this stage it is appropriate briefly to review some of the main techniques used in benefit - cost analysis which are particularly relevant in the field of tsetse and trypanosomiasis control. The main steps in benefit - cost analysis are:

• quantifying the expected benefits of an intervention over time;

• quantifying the expected costs of an intervention over time;

• comparing these, coming up with a standard measure (net present value, benefit - cost ratio or internal rate of return) that makes it possible to

  - assess the intervention’s profitability

  - compare it, or rank it against other possible interventions with which it is competing for funds or which are alternatives for development in the same production sector;

• undertaking sensitivity analyses to examine how sensitive the result is to changes in key assumptions, such as the effectiveness of the disease-control measures, the rate of adoption of an animal health intervention or the growth of human and livestock populations in the project area.

Social benefit - cost analysis studies the effect of an intervention, usually described as a project, on society as a whole, so it takes into account all the benefits and all the costs, regardless of who spends the money or to whom the benefits accrue. In the tsetse and trypanosomiasis field the benefits tend mostly to accrue to livestock and crop farmers, while the expenditures are usually shared between donors, government and local farmers. While many analyses focus on the total social costs and benefits, increasingly, studies are looking at the effect of interventions from the financial viewpoint of the livestock keepers. Thus, the studies by Woudyalew et al. (1999) and Blanc, Le Gall and Cuisance (1995) calculate benefit - cost ratios from the farmer’s point of view. New ways of modelling benefits at farm level are also being developed (McDermott, Coleman and Randolph, 2003).

Time value of money

A key principle underlying the benefit - cost approach is assigning a lower weighting to future income/expenditure as against current income/expenditure^[1]. The rationale for this can be presented in a number of ways.

• Using money has an opportunity cost, which banks acknowledge by paying interest to customers for using their money, and charging it when customers borrow the bank’s money; in the public sector this opportunity cost exists because projects are competing for scarce public funds and allocating money to one project within a sector usually takes it away from an alternative use.

• In this case, we should select projects which provide a good return on money invested, as measured by the compound interest which the benefits add to the costs over time.

• Finally, society places a relative valuation on present as against future income; this is the social time preference rate. This rate tends to be high in poor societies where current needs are urgent, and lower in wealthy societies where the future is more secure.

In benefit - cost analysis this relative weighting of present as against future income (the implied interest rate or minimum acceptable return on money invested) is undertaken by using a process called discounting. This process is not just applied in commercial business ventures, but is an integral part of the project analysis process in public sector projects in all areas (see Gittinger, 1982 on agricultural project analysis; Drummond et al., 1997 for human health projects; Putt et al., 1987; Dijkhuizen and Morris, 1997; Rushton, Thornton and Otte, 1999 for animal health projects and discussion in Kristjanson et al., 1999). Discount rates used in agricultural and livestock analysis generally range from 8 percent to 15 percent, and in the field of human health they range from 3 percent to 5 percent (Acharya and Murray, 1997). With the exception of Budd (1999), whose objective was to present the global magnitudes involved rather than undertake an analysis over time, the economic studies of the trypanosomiasis problem cited above, have applied discount rates of 8 percent or over in their analyses. Since the use of discount rates penalizes future benefits as against present costs, the use of high discount rates has been debated in projects that are expected to have very long-term benefits or many “intangible” benefits that are difficult to quantify, in particular in the field of the eradication of infectious diseases in humans (Acharya and Murray, 1997). The authors conclude that it can sometimes be argued that the selection of human diseases for eradication should be undertaken without discounting using other, very stringent, criteria, and that a proportion of global health funding be set aside for this purpose. Nevertheless, costs should be discounted in order to select the most cost-effective options. However, other writers, even in the field of human health, conclude “technically and theoretically there are good reasons for discounting benefits as well” (as costs) and “discounting health benefits has been advocated as good economic practice in all guidelines on economic evaluation” (Glydmark and Alban, 1997).

As a consequence, it is recommended here that, when dealing with a disease which:

• mainly affects livestock and agricultural production, and

• occurs in a continent where there are huge and urgent alternative demands on finance,

we maintain the convention of using discount rates. In view of the inclusion of tsetse elimination, which would have very long-term benefits, among the options for dealing with trypanosomiasis, the discount rate used in the analysis below was 5 percent rather than the 10 percent which would more usually be applied in the livestock sector.

Discounting has important implications in comparing control and eradication options. This is particularly so in the field of tsetse control where some techniques, such as targets or ground spraying, can be used for either control or eradication. Furthermore, eradicated areas may need to be treated repeatedly because of re-invasion or failure to completely eliminate a tsetse population.

Figure 1 illustrates some of the implications that discounting has for decision-making on options for tsetse and trypanosomiasis control. In this figure, annually recurring expenditures over 20 years are compared to once-off expenditures incurred at the start of a 20- year period. The example used is of expenditures on trypanocides, which are costed at US$1.50 a dose, and then multiplied by the number of cattle per km², in order to obtain an annual total cost per km² at different cattle population densities. The once-off expenditures could equally well refer to annual recurrent expenditure on tsetse control, for example using pour-ons.

The figures on the y-axis show what the equivalent amount spent per km² at the beginning of the period would be. Thus, at about 20 cattle per km², an annual expenditure of US$6, or four doses of trypanocide, would be equivalent to an initial outlay on tsetse elimination of US$1 500 per km². If tsetse elimination cost less than this, it would be the more attractive option, however if it cost more, a very clear argument would need to be presented to show that it was economically justified. Obviously this model simplifies the situation, for example:

• it does not take into account the fact that the cattle population might be increasing during the period;

• it only looks at cost-effectiveness, implying that the two options have equivalent benefits over time, whereas tsetse clearance may be subject to re-invasion, annual control usually does not totally remove the effects of disease, drug resistance can gradually appear;

• it is based on a 20-year time horizon;

• it assumes that tsetse clearance is a once-off expenditure occurring in year one of the project, whereas it may take several years to achieve and be followed by some ongoing annual costs, for example the cost of barriers.

FIGURE 1
Comparing annual to once-off expenditure

Note: Calculated over 20 years at a 5% discount rate

All of these factors could easily be taken into account in a comprehensive benefit - cost analysis, in particular the changes in cattle populations can be tackled using a herd model as outlined below.

Despite these limitations, the analysis is useful in illustrating the basic nature of some of the decisions which have to be made in the field of tsetse and trypanosomiasis control. Similar graphs could be constructed to show:

• the annual benefit per head of cattle which would be needed in order to justify a certain initial outlay on tsetse clearance - again using Figure 1, it implies that if the benefit is expected to be of the order of US$6 per year, the average cattle population per km² would have to be about 12.5 in order to justify a once-off expenditure of US$1 000 on tsetse clearance;

• the level of annual expenditure on tsetse suppression for which it would be more economic, if feasible, to switch to tsetse clearance - for example, if suppression costs US$30 per km², this would be equivalent to a once-off expenditure of just under US$400 per km²; if suppression were deemed to be only 50 percent as effective in controlling the disease as permanent clearance, this figure could be adjusted to just under US$800 (400/0.5).

Threshold values

In economics, as in other disciplines, it is often useful for the decision-maker to be able to define threshold values or cut-off points, above which a certain decision is appropriate and below which another becomes valid. In economic and financial decision-making these are often referred to as break-even points. They define the point at which a project “breaks even”, meaning that above this point the benefits exceed the costs; below this point the costs exceed the benefits. In the same way that the cut-off point for a diagnostic test can be adjusted to make it either more specific or more sensitive, in economics, the cut-off discount rate chosen can make it possible to give different weights to long-term benefits as against current costs. Also, as in other disciplines, the threshold value has to be interpreted by the decision-maker, and may often consist of a range of values within which it is felt that the result is doubtful. In project appraisal, these “doubtful” projects, are those which should be put at the “bottom of the pile” and only looked at when no better alternatives are found or when circumstances change, such as their score on another of the SITE criteria.

The threshold concept is particularly helpful in assessing the economic viability of different tsetse and trypanosomiasis control schemes. Some of the thresholds are:

• cattle population density at the start of a programme (as seen in Figure 1 this determines benefit levels and cost levels for “per head of cattle” control methods such as trypanocides and pourons);

• human population density at the start of the programme (influencing fly habitat and also helping determine benefit units, for example the potential for using draught power);

• for each area the cost of once-off tsetse clearance plus the ongoing cost of barriers, weighted by the risk of needing to retreat the area;

• the cost of the technically feasible ongoing tsetse suppression techniques.

These thresholds can be defined with some accuracy for a particular area or region with similar areas - but as everyone who has worked on the tsetse and trypanosomiasis problem knows, generalizing is very difficult. There are other criteria to be included, in particular human and livestock population pressure in neighbouring areas. It should be noted at this stage that on the benefit side these thresholds are, to all intents and purposes, the same ones that are used in the GIS filtering process in order to identify promising areas for intervention (e.g. Gilbert et al., 2001; Hendrickx, 2001; Hendrickx et al., 1999; PAAT, in prep.).

To complete this filtering process, benefit - cost analysis adds the possibility of summarizing much of this information in a single measure. The most practical for the purposes of this analysis is the benefit - cost ratio (BCR),^[2] which is expressed as:

Benefit-cost ratios have the added advantage that they can easily be adjusted from the above measure, which calculates the return on all monies invested, to measures that analyse the return to different groups such as farmers, livestock keepers or to investment, research, etc.

The following sections discuss how the information above can be treated to produce realistic and consistent estimates on the impact of the disease over time and in response to various interventions.

Partial analysis - defining the “with” and “without” scenarios

The basic tool used in farm management in order to quantify the costs and benefits of a proposed modification to the production system is partial analysis, which is also sometimes called partial budgeting. It provides a useful framework for categorizing benefits and costs, and when the framework is completed it acts as a checklist, which applies particularly well to disease-control interventions (e.g. Putt et al, 1987; Dijkhuizen and Morris, 1997; Rushton, Thornton and Otte, 1999).

For trypanosomiasis the main items to be included under the four headings that comprise the partial analysis framework are shown in Table 1.

“With” and “without” project scenarios for benefits

Determining what the “with” and “without” project scenarios are is always difficult. On the benefits side, in terms of livestock productivity, it depends on studying before and after, or with disease and without disease situations, and should thus follow the same principles as an intervention trial in epidemiology. Swallow (PAAT, 2000), in his review paper, discusses the basis on which the production parameters with and without the disease were estimated in the various studies, distinguishing between the following approaches:

• longitudinal monitoring of herds, comparing parameters for individuals detected parasitaemic and those not detected parasitaemic;

• monitoring the health and productivity of cattle herds in similar areas distinguished by different levels of trypanosomiasis risk or challenge;

• monitoring livestock before and after control measures were undertaken.

TABLE 1
Partial analysis for tsetse and trypanosomiasis interventions

An analysis of these studies and discussion of the parameters obtained is outside the scope of this paper, however it will be important (see Chapter 5) to consider these issues when making recommendations on how to standardize the collection of data required for the economic analyses.

The importance of correctly assessing the “with” and “without” scenarios can be illustrated by following the series of graphs given in Figure 2. Taking the size of the cattle population as an indicator of benefit levels, Figure 2a shows the “null hypothesis” situation, i.e. that the cattle population would remain unchanged in the absence of interventions to control the tsetse and trypanosomiasis problem. This “no change” scenario is often unconsciously adopted in evaluations, forgetting that while the population growth rate might remain more or less the same for some years in the absence of interventions, the population itself is unlikely to be static.

FIGURE 2
Alternative “with” and “without” intervention scenarios for tsetse and trypanosomiasis control

Figure 2b illustrates the situation where interventions to control the disease yield the highest profits - where a population is declining in the absence of control, owing to the severity of the disease - but would increase if effective control measures were implemented. This was the case, for example, in the Yalé area of Burkina Faso (Kamuanga et al., 2001a) where there had been massive losses due to the disease, reflected in a huge decline in the population.

Figure 2c, however, illustrates a situation that is often encountered in West Africa’s moist savannah zone, where even in the absence of interventions to control tsetse and trypanosomiasis, the cattle herds are still growing. This has been the situation in Côte d’Ivoire, due perhaps to farmers’ use of trypanocides and to the presence of trypanotolerant cattle (Camus, 1981; Shaw, 1993; Pokou, Swallow and Kamuanga, 1998). A similar situation is found in parts of northern Nigeria (Shaw, 1986). In this situation, potential benefits are lower than under the previous scenarios.

Finally, Figure 2d can be seen as an extension of Figure 2c, showing what the situation would be if there were a production ceiling, usually imposed by an area’s livestock carrying capacity limit, itself determined both by the quality of the natural forage and by the proportion of land taken up for farming. In this case, production under the “with” and “without” scenarios converges and the effect of disease control is to enable production from cattle to reach its ceiling earlier on. Benefits under this scenario, although lower than under the others, may still be significant.

An issue which further complicates assessments of the impact of tsetse control strategies, is the possibility of using pour-on preparations that also affect ticks, and thus produce a wider range of benefits whose impact is difficult to compare to those of other tsetse and trypanosomiasis control strategies.

This discussion has not directly mentioned the issue of cattle migration, and more specifically immigration into areas that have been cleared of tsetse. A method for dealing with this issue, which seems to work well, is to take the cattle population affected by the project as being:

• those animals present in the area at the start of the project,

• plus any animals that migrate into the area during the course of the evaluation period,

and assume that both groups benefit from improved productivity, since the immigrants presumably moved into the area because they hoped for better conditions - whether better grazing or less risk from disease. This approach produces realistic results for actual situations and can be integrated into a herd model (Putt et al., 1989; Shaw, 1990, 1993).

“With” and “without” project scenarios for costs

Identifying the “with” project costs is usually relatively straightforward, since these mainly involve direct expenditure on a new disease-control programme. However, if one of the impacts of the project is to increase livestock numbers and/or productivity, this may involve extra production costs for livestock keepers and these need to be included in the extra costs.

More difficult to assess are the “without” project costs. The main issue to consider here is “how are farmers now, and how will they continue to manage the problem of trypanosomiasis in the future?” More evidence of how they do this is slowly accumulating. CIRDES, ILRI and ITC (2000) comment on farmers’ expertise in “integrated disease management” and state “The strategies that livestock owners adopt for production under trypanosomiasis risk have elements that take effect over the long-term, medium-term and short-term. Choices with long-term effects, especially regarding livestock breed and type, condition choices with medium-term effects, especially regarding transhumance and use of acaricides for tsetse and tick control. Similarly, choices with long-term and medium-term effects condition choices with short-term effects, especially the use of trypanocidal drugs.” Looking at the RTTCP countries, Van den Bossche and Vale (2000) discuss the widespread use of trypanocides, and state that “preference is given to the treatment of oxen and cows, i.e. the productive animals in the herd” and Doran (2000) points out that in the surveys conducted, trypanosomiasis challenge seems to affect calving rates, but not cattle mortality rates which may be masked by the effects of curative treatment. This tendency to prioritize on cows and oxen is very sound in economic terms. Looking at the economics of traditional cattle-production systems in West Africa, most of the output by value either consists of milk and draught power or is linked to herd growth. These in turn are a function of the health of adult females and draught oxen. Thus, taking a herd model and simulating the results of removing the effects of the disease in these two groups of animals deals with around 75 percent of the losses due to trypanosomiasis in many situations.

TABLE 2
Partial analysis for tsetse control in an area where farmers currently use trypanocides

Thus, taking into account “with” and “without” project scenarios in this way means that the relevant partial analysis framework for the introduction of tsetse control would be as given in Table 2.

In Table 2, the benefits under c) would be the added increase in output due to a switch from using drugs to tsetse control and under d) for the savings that livestock keepers would now be able to make on trypanocides. In this context, Pokou, Swallow and Kamuanga (1998) and CIRDES, ILRI and ITC (2000) did note that farmers in northern Côte d’Ivoire continued to use drugs in the tsetse suppression area, probably partly because they were not completely aware of the extent to which tsetse control has reduced risk, and partly because some isk was actually still present and animals were being sent outside the tsetse control area on seasonal transhumance. Other factors might be the usefulness of these drugs against babesiosis, and the fact that in many places, trypanocides are still among the few veterinary drugs which are widely available.

Other methodological issues

There are a number of other methodological issues in project analysis, which have relevance to the analysis of the tsetse and trypanosomiasis problem.

The distinction between financial and economic analyses should briefly be mentioned (see Gittinger, 1982 for a detailed discussion). This operates at two levels.

a) The viewpoint from which the analysis is made - an economic analysis usually embraces the benefits and costs to society as a whole, while a financial analysis tends more often to be undertaken looking at the costs and benefits to individuals, specific groups or organizations (e.g. crop farmers, livestock keepers, cattle traders, governments).

b) The prices used in the analysis - there is a convention of using “accounting” or “shadow” prices which attempt to adjust market prices so that they better reflect real resource costs; this is particularly the case for some prices such as foreign exchange rates, or agricultural prices that are fixed by government, accounting prices have been used in looking at tsetse and trypanosomiasis control economics, for example by Jahnke (1974) and Itty (1992).

In practice, many economists end up producing a sort of “halfway house” midway between an economic analysis and a financial analysis, by making adjustments for over-valued exchange rates and taxes and subsidies while leaving most other prices at their current market values. The term “economic” tends to be used as the general term covering both approaches, and this convention is followed here. Most of the analyses conducted here are economic in the sense that they look at the benefits and costs to society rather than individual groupings, and financial in the sense that they are based on current market prices. However, as discussed at the start of Chapter 3, a number of studies have looked at the benefits and costs from the financial viewpoint of farmers and livestock keepers (Blanc, Le Gall and Cuisance, 1995; Woudyalew et al., 1999; McDermott, Coleman and Randolph, 2003). In addition, a number of studies have examined farmers’ willingness to pay for tsetse control, these were studied for a West African situation by Kamuanga et al. (2001b) and the various studies were reviewed by Kamuanga in PAAT (2003).

Dealing with risk and uncertainty is obviously crucial when looking at the possible outcomes and costs of tsetse and trypanosomiasis control. Sensitivity analyses are an effective way to deal with this, by studying the effects of changes in key assumptions and seeing how sensitive the project’s performance is to likely changes. As mentioned above, identifying the threshold at which a project becomes profitable, through some form of break-even analysis is another way of defining the project’s limits (e.g. with respect to disease incidence in the absence of control or minimum human and cattle populations necessary to generate sufficient benefits to make the project economically feasible).

The time horizon selected is also important, especially when comparing control and eradication options, as mentioned above in the section on “Time value of money”, page 11. The figure conventionally selected in benefit - cost analyses is 20 years and this has been used in the model runs below. Sensitivity analyses looking at 30 and 40 years are desirable, particularly if eradication is being considered - however, these need to be very carefully interpreted, since looking that far ahead into the future involves considerable speculation, and the assumption that current trends will continue can be enormously misleading.

Defining the project to be analysed

Finally, against the background of discussions on huge area-wide programmes to eliminate the fly over large sections of the continent, what is the rationale for trying to prioritize and select intervention programmes to control the tsetse and trypanosomiasis problem? The terms of reference for this paper were to produce guidelines for prioritizing intervention programmes on the basis of economic criteria. In economics, decisions are made at the margin, that is by comparing the potential additional benefit from a proposed change to the likely additional costs as shown in the framework for partial analysis (see Tables 1 and 2). In looking at the tsetse and trypanosomiasis problem, it is essential that individual projects are defined, analysed and ranked using each of the SITE criteria (see beginning of Chapter 3). The size of such projects should take into account the following.

• The project must be technically feasible - the area must have a defined trypanosomiasis problem, be of a suitable size for the most cost-effective control technique (such as a fourth-level river basin for area-wide tsetse eradication, see Hendrickx, 2001; PAAT, in prep.) or the zone should be covered by a defined group capable of concerted action (such as farmers with a particular problem and outlook, a development project or an administrative or extension structure).

• Funding for the project must exist - there is no point in analysing a project for which funds will run out half way through, since this will prejudice the outcome and render the initial appraisal invalid.

• The technical capacity to carry out the project must exist.

Thus, it is strongly argued that each individual project, of whatever size, needs to be assessed on its own merits, not, especially at this stage, for its contribution to a continent-wide super-programme. The issue of timing, in particular, is important here. It is recognized that, as stated by the PAAT Advisory Group at its 8th meeting in 2002, while we “resolve to reduce and ultimately eliminate the constraint of tsetse-transmitted trypanosomiasis in man and animals ... progress towards the final objective is best achieved through concerted efforts towards intervention in a sequential fashion, with the focus on those areas where the disease impact is most severe and where control provides the greatest benefits to human health, well-being and sustainable agriculture and rural development”. It follows that undertaking tsetse eradication work on the fringes of the tsetse distribution, where the tsetse habitat is already marginal, cannot be justified purely in order to accrue benefits which will only start very far in the future and in another part of the continent. However, as Chapter 4 shows, it is in some of these fringe areas in West Africa, that controlling trypanosomiasis in cattle does yield high benefits.