J. McINTIRE
Introduction
Models of animal production
Conclusion
Footnotes
References
Economic analysis of trypanosomiasis has concentrated on the costs of vector control (ILCA, 1979; Jahnke, 1976) and on returns to animal production after control. Such studies assumed that animal production was near zero in trypanosomiasis-infested zones and that benefits from control would be roughly equal to the difference between animal production in areas without trypanosomiasis and no animal production. It was further assumed that production, once established, would be managed on a large scale with public support, or even direct government management. Several aspects of the problem, such as the scale of operation, associated investments in animal production (eg, better nutrition), risk and product mix (meat versus milk) were excluded.
Different economic analyses are now required for several reasons. Policy emphasis has moved from ranch or pastoral development in cleared areas of sparse settlement to spontaneously settled areas with mixed farming. Associated with this move is a shift in focus from East to West Africa, where the economic context is very different. Third, recognition of the high costs of fly control by land clearing or by insecticides and of the conflicts of those costs with other national objectives, has induced interest in trypanotolerant animals, with consequent emphasis on nutrition and management. Fourth, the scale of production, objectives and production costs in agro-pastoral systems are different from ranch or pastoral systems. The economic problem is no longer to achieve the highest return on a flexible investment made by government; rather, it is to find production techniques giving the highest return on a fixed investment (the existing agro-pastoral herds) held by herders with few other investment opportunities. Fifth, the international trade context has changed dramatically; producer prices in Africa, which were once well below world levels, are now in many instances well above them. This fact destroys much of any argument for an aggressive government price policy to encourage production.
The overall shift in emphasis has important implications for research strategies. To oversimplify grossly, the problem is no longer land-use planning consequent upon disease control. It has become the detailed study of factors determining productivity in different environments, the identification of treatments raising productivity and the insertion of those treatments into existing production systems. All have implications for relations with national research programmes and for the particular roles of ILCA, ILRAD and cooperating institutions in the Network.
This paper presents background to some of the economic analyses done in the Network. The first two sections define some models of animal production. The third presents results of experiments with the ILCA microcomputer herd model (von Kaufmann et al., 1987) about some important problems in the Network and projects returns to gains of animal production in areas where cattle are common. It uses those projected returns as guides to overall research resource allocation within the ATLN, with some attention to the roles of the various partners. The final section is a summary and conclusion.
The basic notion is that animals are a capital stock providing income to their owners. Income consists of cash sales, plus the values of meat, milk, manure and draft power. The owner can realize his capital immediately by selling the herd, or he can consume income over time as a flow of proceeds from offtake. In a commercial beef herd, the basic decision is when to sell the animals, given their immediate cash values and their expected future values.
Most economic models of animal production (a recent one is Rosen, 1987) have a simplified herd structure and discuss only the cash income of commercial ranchers. (An exception is the biological model of Konandreas and Anderson, 1982). They assume well-functioning marketing, veterinary and banking systems, in which costs of market information and health services are low and in which animal capital has a cash opportunity cost.
Assumptions in typical herd models must be modified for African production systems. First, it is necessary to assume a varied herd structure. In pastoral systems, mixed milk and meat herds are the rule. Second, kind income is important, especially milk consumed by herders. Third, costs of information and of health services are high with respect to product prices. (This does not mean that markets do not function well, only that distance and poor transport raise the prices of services.) Such costs are a tax on the prices received by herders and constitute a barrier to aggressive government price policies. Fourth, production is subject to risks from fluctuations in pasture availability and diseases.
Static and dynamic models
There are two general models for economic analysis of animal production: static and dynamic. Static models are for one year. Taking annual return to a breeding female as the criterion of profitability, that return is calculated as a function of female liveweight, calving rate, calf liveweight and viability and milk production. It is then possible to model response to some treatment, usually nutrition or health. This is done by estimating the treatment effect on some parameters (eg, the effect of dam nutrition on calving rate) and recalculating the model. Other variants allow estimates of the effects of price policies, and of the returns to factors of production, such as land and labour.
A second model is dynamic. The uses of dynamic models are to examine systems over time, to investigate risky production systems and to illustrate the effects of extreme events. Dynamic models are calculated over many years (between 5 and 30) to simulate a breeding herd. The outputs from such models are the same as those from static models: incremental values of output, sometimes expressed as rates of return to factors of production. Dynamic models can be modified to produce frequency distributions of production outcomes. This can be done stochastically - that is, in a model with random variables - or deterministically - that is, in a model in which variables are fixed at their means.
Table 1. Notation for model of animal production.
|
Indices |
|
|
i |
age/sex class in herd |
|
t |
year during the simulation |
|
Variables calculated for each age and sex class | |
|
A |
numbers in herd structure |
|
W |
LWs of animals |
|
D |
mortalities |
|
S |
sales |
|
I |
intake of pastures |
|
s |
sale rate |
|
sl |
slaughter rate |
|
Prices and costs | |
|
pw |
market price of LW |
|
pm |
market price of milk |
|
pc |
pastures cost |
|
pg |
grain price |
|
Reproduction and intake parameters | |
|
r |
calving rate per breeding cow |
|
f |
first age at which heifers enter breeding herd |
|
I |
intake of pastures as % of LW |
|
k |
calf growth as a function of milk consumption |
|
Milk production | |
|
m |
annual milk production per breeding cow |
|
mo |
annual milk sales (kg) |
A model of a mixed-cattle herd
The ILCA model is summarized using the notation in Table 1 for cattle (the model can be applied to small ruminants as well). Capital letters usually represent totals and lower case letters represent rates or coefficients. The model has sections for herd structure and growth, feed intake, output prices, milk production and technical changes.
Terms used in the herd model
The "untreated" situation is the set of conditions-animal productivity, prices, tsetse challenge-before the intervention to improve productivity. The 'treated' situation is the set of conditions after the intervention. Net income is the annual income from liveweight (LW) and milk sales, valued at market prices and including the value of milk consumed by the herder (manure, draft power, hides and skins are excluded for the time being). The incremental cash flow is the difference in net income between the treated and untreated situations. The discount rate is the opportunity cost of the herder's capital; it is usually defined as the return to capital in its best alternative use. The net present value (NPV) of the treatment is the value of the incremental cash flow discounted to the first year in which the treatment takes place. The NPV is a standard measure of the profitability of an investment.
Herd structure and growth
Let [A] represent herd numbers by age and sex class at some time "t", [W] a vector of animal LWs, [D] mortalities and [S] net sales. (Assume that no purchases of animals are possible, so that [S] => O. {Footnote 1}) For reproduction, let "r" be the annual calving rate of breeding cows in the herd and "f" be the age at first calving.
Pasture production and feed intake
The herd grazes a maintenance diet of pastures. Its rate of dry matter (DM) intake, "i", is a constant fraction of its LW; [I] is the herd's total annual intake (equal to i [W]) and "pc" is the unit cost of pastures. Pasture production fluctuates randomly, within a normal distribution. Mortalities and sales are set at mean values, but fluctuate with pasture production. For example, if pasture production is below average, then mortalities and sales are above average. The total cost of pastures, "q", is therefore a function of intake and is equal to pc (i[A][W]). The unit cost "pc" is calculated from the opportunity cost of pasture land, which could otherwise be used for crops.
Output prices
Let "p" be price paid per kg of animal LW, net of marketing and other costs, so that revenue per kg LW is equal to "p" and similarly for the milk price "pm". Prices of LW and milk are constant and do not vary with numbers of animals marketed {Footnote 2}.
Milk production
Milk production "m" is a constant annual quantity per cow. Total milk production per herd is therefore the number of cows times the calving rate times calf viability times "m". (It is assumed that milk let-down stops if the calf dies.) The herder can sell the milk, consume it, or leave it for the calf. Milk sold or consumed is valued at the market price. Milk left for the calf is transformed into calf LW at a constant rate, "k", equal to 7 kg of milk per 1 kg of LW until weaning.
Technical changes
Technical changes are modelled with respect to the treated and untreated situations. First, the untreated situation is modelled with survey data. Herd structure and productivity parameters are adjusted to produce a stable herd. The treated situation then introduces changes in productivity parameters, given the stable herd assumption and calculates the return to the technical change, defined as the NPV of the differential cash flow between the treated and untreated situations.
Equations
Suppressing the subscript "i" for animal class, the basic relation is
(1) a (t+1) = a (t) - s (t)- d (t) + b (t).
The number in any class of the herd at the end of the year is thus equal to the beginning herd, minus sales and deaths, plus births.
Letting a (cow, t) indicate the numbers of cows in the herd, then income from the herd is where the first term is the net value of animal sales across all categories of animals, the second is the net value of milk sales plus milk consumed by the herder and the third is the cost of pastures.
(2) [S (t)][W](p) + {[a (cow, t)]*(r)]m (pm)} - q
The model illustrates feasible solutions to animal production problems. The research problem is to identify treatments which increase productivity. The economic problem is to maximize production from the herd, along the lines of equation (2) over the simulation period, selecting the most profitable treatments. In the context of trypanosomiasis the basic question is: what are the benefits from productivity gains resulting from control?
Because data are sparse on the true effects of productivity interventions, it is necessary to estimate productivity gains by comparing herds in the same environment as a measure of what is possible. Differences among herds on productivity criteria (calving rate, mortalities, liveweights, etc.) can be said to represent the natural variation between herds in that environment.
Data sources and herd structure
Data are taken from preliminary results of the ATLN (ILCA, 1986). To simplify the presentation, a herd of approximately 100 cattle is modelled (Table 2). The herd structure and productivity indices correspond roughly to those at a humid site in west-central Africa with moderate trypanosomiasis challenge. Milk and LW prices are set at representative levels. Benefits from trypanosomiasis control include reduced calving interval, reduced calf mortality and greater LWs consequent on greater weaning weights. These benefits can be modelled individually and collectively.
Table 2. Data for experiments.
|
Herd size |
95 | |
|
Number of females |
25 | |
|
Annual calving rate |
60.0% | |
|
Age at first calving, years |
4 | |
|
Calf birth weights (kg) |
20 | |
|
Mortalities | ||
|
|
Adults, heifers, young males (> 2 years) |
5.0% |
|
|
Calves, 0-1 years |
10.0% |
|
|
Calves, 1-2 years |
5.0% |
|
Prices, CFAF/kg | ||
|
|
Milk |
300 |
|
|
LW |
750 |
|
Opportunity cost of 1 ha of pasture, | ||
|
|
CFAF |
5,400 |
|
Average herd value, 10 year period, | ||
|
|
Untreated situation, 000s CFAF |
11,622 |
The riskiness of treatments
Investments in trypanosomiasis control are subject to risks from changes in prices and in animal productivity. To simulate those risks, random variations in prices and animal productivity are introduced into the model. Each random variation is associated with what is called a "state of nature". Each state of nature corresponds to an NPV. The distributions of the NPVs are used to assess the "riskiness" of the treatment in different states of nature, as well as the probability that it will be exceed some critical level for adoption by farmers.
Experiments with the models
Simulations with the models are termed experiments. Four basic experiments are done. Each corresponds to productivity gains which might be achieved with some form of trypanosomiasis control. Experiment 1 involves reductions in calf mortality, accompanied by reductions in calving interval. Experiment 2 involves reductions in calf mortality, accompanied by increases in weaning weights and in subsequent LWs, without changes in calving interval. Experiment 3 involves reductions in calf mortality and in calving interval, accompanied by increases in weaning weights.
Table 3. Results of experiments, NPV of incremental cash flow (000s CFAF).
|
|
Change in calf mortality, 0-1 years |
|
|
-5% |
-10% |
|
|
|
Change in calging rate without changes in LWs |
|
|
5% |
1,513 |
2,608 |
|
10% |
2,007 |
3,145 |
|
15% |
2,516 |
3,699 |
|
|
Cumulative LW gains without changes in calving rates |
|
|
10% |
2,623 |
3,725 |
|
15% |
3,483 |
4,611 |
|
20% |
4,371 |
5,527 |
|
|
10% increase in calving rate and LW gains |
|
|
10% |
3,672 |
4,866 |
|
15% |
4,574 |
5,798 |
|
20% |
5,507 |
6,761 |
A reduction in calf mortality of 5%, accompanied by an equal increase in calving rate, would produce benefits having an NPV of the order of CFAF1.51 million (Table 3). Reducing calf mortality by 10% and increasing calving rate by 10% would produce benefits of about CFAF3.15. At a 10 % decrease in calf mortality to one year, accompanied by a general 20% of animal LWs, but without a change in calving rate, benefits would be roughly CFAF 5.53 million.
Are the mean comparisons stable? Distributions of the benefits are plotted in Figure 1. Obviously, the results from experiment 1 are always inferior, while experiment 3 is slightly superior to experiment 2. The risks in these hypothetical situations, expressed as within-experiment variation of the NPVs, would not change the mean comparisons.
How do the benefits compare to the investment capacity of a herd owner? Assume that the herd owner's cash income and investment capacity are related to liveweight sales, that his subsistence income is related to milk sales and that the entire value of sales is invested in the treatment. The value of offtake in the first year of the model, for the untreated situation, is about CFAF1.39 million, shown as the horizontal line in Figure 1. Investing all offtake {Footnote 3} in experiment 1 would produce an NPV less than the investment cost. Investments would produce positive returns about 90% of the time for experiment 3 and about 80% of the time for experiment 2.
How do these productivity gains compare to those in the Network? The data modelled are loosely based on those from Togo, Gabon and Zaire. Preliminary data analysis (ILCA, 1986) showed that reducing trypanosomiasis (as measured by PCV and by number of infections) would have, at most, an effect of 17% on the calving rate in Gabon, though the effect was not statistically significant. In Togo, the effect was roughly 11% and not significant. Infection had a small, but usually significant, on birth and weaning weights. (Data on mortalities seemed exceptionally low and have been ignored.) Therefore, with the exception of the mortality effects, the assumed effects of control are probably greater than what the Network has estimated so far, especially in view of the fact that many of the estimated effects were not statistically significant.
1. Moderate increases in calf viability and in calving rate would produce small benefits compared to the average value of offtake from an untreated herd. Substantial increases in calf viability, calving rate and in LWs, resulting from trypanosomiasis control would produce large benefits compared to offtake value and would constitute attractive investments.2. Major increases in benefits are, however, improbable, without major increases in LWs. To put it another way, changes in reproductive efficiency and in calf viability alone, as currently indicated by Network results, would not provide large economic returns.
3. Risks of loss on investments, caused by random changes in pasture supply, are not likely to affect the first two conclusions.
4. The temporal pattern of benefits from improvements in reproductive efficiency-raising calving rate, for example-favours treatments with low fixed costs. This is because such improvements pay off more slowly than do improvements in liveweights. However, because such improvements would pay off over many years, their maintenance puts constant pressure on national veterinary services to maintain the quality of their services.
5. Risk in the supply of veterinary inputs was not modelled. However, if the supply of drugs varies, then drug investment becomes less profitable on average. Such risk tends to make investment in trypanotolerant animals relatively more profitable, if they do require fewer such drugs, than investment in drugs for non-tolerant animals.
6. Indications from Network data show, at least at this point, that the physical increases in productivity associated with control (or at any rate those associated with reduction in indicators of trypanosomiasis) have not yet been achieved.
7. LW prices in the models are high by international standards because markets in coastal West and Central Africa are protected from import competition. Using more competitive (de, lower) import prices would reduce the benefits of treatments.
1. It is fair to assume zero purchases for two reasons: lack of information about the purchased animal's true condition and capital. In a normal year, herders do not usually buy animals because they lack information about condition. What they do is to care for breeding stock in return for a share of the offspring. This contract resolves the information problem and allows borrowers to substitute labour for cash.
2. The effects of relative prices of products and inputs, such as veterinary services, are straightforward. If the costs of veterinary or marketing services are expressed as a percentage "h" of LW prices, the net price per kg LW is then p*(1-h) and similarly for milk. Net income as a function of meat and milk prices is then directly related to net income as a function of service costs. If the costs of services are fixed in each year and do not depend on herd size or total LW, then they can be subtracted from the incremental cash flow of the treatment in each year. If such costs are fixed in the first year of the treatment, then they can be subtracted from the NPV of the incremental cash flow. Similarly, the effects of different herd sizes are unimportant as long as there are no economies of scale in marketing; such economies would increase the net product price received by large herders and would increase the benefits of technical changes to such herders.
3. This is an extreme example, because it assumes that nearly all of the herder's cash income in one year is invested, which is unlikely to be the case. As such, it overstates the minimum acceptable return.
ILCA. 1986. The African Trypanotolerant Livestock Network: Indications from results. ILCA, Addis Ababa.
ILCA. 1979. Towards an economic assessment of veterinary inputs in tropical Africa. Working Document No 1. ILCA, Addis Ababa.
Jahnke, H.E. 1976. Tsetse flies and Livestock Development in East Africa. Munich: Weltforum Verlag.
Konandreas, P. and F.M. Anderson. 1982. Cattle herd dynamics: An integer and stochastic model for evaluations production attractives. ILCA Research Report No. 2. ILCA, Addis Ababa.
Rosen, S. 1987. Dynamic Animal Economics. Am. J. Agricult. Econ. 69: 547-557.
Von Kaufmann, R., J. McIntire, P. Itty and Edjigayehu Seyoum. 1987. ILCA Herd Model Users' Manual Addis Ababa, Ethiopia: ILCA.
