PREVENTIVE VETEFUNAFIY
MEDICINE
Preventive Veterinary Medicine 29 (1996) 21-36
An economic study of smallholder dairy farms in Murang’a District, Kenya Gerdien van Schaik ayb7 *, B.D. Perry a, A.W. Mukhebi a, G.K. Gitau a, A.A. Dijkhuizen b b Department
aInternational Livestock Research Institute, Nairobi. Kenya of Farm Management, Wageningen Agricultural lJniversity,Wageningen,
Netherlands
Accepted 11 April 19%
Abstract The h4urang’a District is situated in central Kenya and is a highland area with a high potential for dairy production. This study was carried out to evaluate the performance of smallholder dairy farms by analysing available data and by simulation modelling. Eighteen case farms were selected non-randomly on the basis of differences in reported milk production, based on the amount of milk delivered to a dairy society. Data of these smallholder dairy farms were collected by means of a questionnaire and included all farm resources and enterprises. The Technology Impact Evaluation Simulator (TIES), a Monte Carlo computer simulation model, provides a method for evaluating the financial and economic impacts of technology changes on a whole-farm basis, and explicitly incorporates risk. The simulation of each case farm in the Murang’a District with TIES gave an indication of the variation in economic performance of these Farms. A sensitivity analysis on the economic terms pointed out the key indicators for successful farm performance. The smallholder dairy farms investigated differed in farm performance-itself strongly influenced by the performance of the dairy enterprise of the farm. Milk production and calving interval were the main indicators describing the performance of the dairy enterprise and, through that, overall farm performance. Both milk production and calving interval were influenced by the amount of concentrates fed-suggesting that feeding concentrates is an important indicator of high faml performance. The influence of health-services costs on farm performance was not significant. Off-farm income also influenced overall farm performance; farms heavily dependent on off-farm income were often inefficient in their dairy enterprise. Keywor& Smallholder dairy farms; Economic farm performance; Dairy enterprise; Simulation model
-_ * Corresponding author. 0167-.5877/96/$15.00 Copyright 0 1996 Elsevier Science B.V. All rights reserved. PI/ SO167-5877(96)01062-S
22
G. van Schaik et al./Preventive
Veterinary Medicine 29 (1996) 21-36
1. Introduction In sub-Saharan Africa, per capita food production of all food commodities (including the major livestock products) continues to decline (Mbogoh, 1984). Despite this downward trend, many countries in Africa have demonstrated substantial potential for dairy development (Walshe et al., 1991). Current consensus is that efforts are best targeted at the smallholder dairy sector (Walshe, 1987; Omore et al., 1994). Reasons cited include the large amount of labour and personal commitment required for dairy production, the predominance of smallholdings in the most suitable dairy production zones and the large social benefits from supporting small-scale farmers (Brumby and Sholtens, 1986). In the past, Kenya has been one of the few countries to support smallholder dairy development successfully (Walshe et al., 1991). In Kenya, smallholder dairy farms account for between 75 and 90% of all milk produced (Kenya Government, 1986). However, most of the increased production in the smallholder sector has been due to extended use of land and livestock resources rather than from higher individual-cow productivity (Walshe et al., 1991). A number of pressures (including rapid population growth and limited land resources; Walshe et al., 1991) have already pushed the smallholder to more intensive dairy production. Such intensification will require improved management and increased resources per cow if it is to be sustainable. An important step in evaluating potential development options is to identify the major constraints and opportunities for increased productivity on smallholder dairy farms. An economic evaluation of the farming system may provide farmers with indicators to improve farm performance. In this study such an evaluation was carried out using a computer simulation model. I .I. Research area Murang’a District is a highland area in the Central Province of Kenya with high potential for dairy production. The proximity of Murang’a District to Nairobi provides access to a large market for dairy products. The District is divided into nine agro-ecological zones (AEZ), according to annual rainfall and temperature gradients. Each zone differs in agricultural value and has its own particular farming system with characteristic farm size, crops, grazing management and cattle breeds. The farming system with the highest potential for dairy is situated in the UMl -2 zones, at an altitude of 1500- 1800 m and with annual rainfall of 1600-1800 mm (Jaetzold and Schmidt, 1983). I .2. Smallholder farms Small-scale farms, in Kenya’s context, cover less than 12 ha (Nyangito, 1992). A distinct feature of small-scale farms is that they are family farms. Another characteristic is that the farmers grow temporary and/or permanent crops for sale and food crops (Burger, 1994). Furthermore, farmers raise livestock (Nyangito, 19921, including mainly cattle. Livestock farming is generally integrated with cropping activities in the smallholder farm system, using crop by-products as cattle feed and applying manure to crops.
G. van Schaik er al./Preuentive
Veterinury Medicine 29 Cl9961 21-36
23
Livestock and crop production serves the overall objective of providing food and cash income for the farm family. Often the farmer does not rely completely on this on-farm income, but also has a considerable off-farm income (Burger, 1994). Culturally, possession of a farm is very important in Kenyan society; the farm provides income, social security and a dwelling for the entire family. The farm system in UMl-2 of Murang’a District is one in which grade dairy cattle are kept under a zero-grazing management system. The system is characterised by exotic breeds (e.g. Friesians, Ayrshires, Guernseys) permanently housed under a shed in which they are fed and hand-milked. Compared with indigenous Zebu cattle, the exotic breeds have high milk yield but are more susceptible to diseases such as theileriosis (Norval et al., 1992.). Zero-grazed cattle are fed on fodder crops (e.g. Napier grass), crop residues (e.g. ma:lze stover), and manufactured feeds. Cow fertility has a considerable effect on the economy of dairy farming; on a per herd-year basis a shorter calving interval results in higher milk production and more calvings per year. Pregnancy percentage at first insemination, the number of inseminations per conception and the number of days from calving to conception all influence the length of the calving interval. Long calving intervals are often the result of several factors, such as difficulties in heat detection, irregular availability of artificial insemination (AI:) services, and poor condition of the cow due to inadequate feeding (Kiptarus, 1994). Furthermore, farmers may influence calving intervals by not inseminating cows when a prolonged lactation or seasonal calving is desired. 1.3. Simulation
modelling
with TIES
The Technology Impact Evaluation Simulator (TIES) is a model based on Monte Carlo Simulation and offers a flexible method for assessing and predicting the financial and economic impacts of new technologies on farms. TIES was developed by the Agricultural Experiment Station at Texas A&M University in collaboration with ILRAD, specifically for use in developing countries by incorporating farm production and consumption aspects characteristic of these countries. The model uses one year at its time step and simulates 10 years recursively by starting each year with the ending debt, asset, livestock herd, and household information of the previous year. The lo-year planning horizon is repeated for 100 iterations to generate estimates of the parameters for empirical probability distributions of key output variables such as internal rate of return, benefit cost ratio, and net present value. The rnodel incorporates crop production activities such as crop mix, yield, production, household consumption, livestock feeding, and marketing. The annual livestock production activities include breeding, calving, culling cows, raising replacements, bull replacement, household consumption, milking, and marketing. The annual economic activities include calculation of variable and fixed costs, debt repayment, machinery depreciation and replacement, household consumption, off-farm income, marketing and total receipts, net cash farm income, cash flow, and balance sheet values (assets, liabilities, and net worth). TIES uses a pseudo-random number generator to incorporate production and market risk faced by producers. Stochastic prices and production levels for crops and cattle are
24
G. van Schaik et d/Preventive
Veterinary Medicine 29 (1996) 21-36
drawn at random each year from empirical probability distributions for these variables. Using a pseudo-random number generator guarantees that each technology is evaluated with the same sequence of random weather (crop and milk yield), market (price), and mortality conditions (calving rate, rate of gain, and death rates). At the start of each IO-year planning horizon, TIES generates random values using probability distributions based on historical yield information, and uses these values to calculate production by multiplying the harvested hectares and yields for respective crop enterprises. Each crop’s output is available for household use, livestock feed, or sale. The cattle herd is specified by composition of different age/type cohorts and is simulated assuming the specific herd size (number of cows, oxen, and bulls) is to be maintained over time. Maintaining a pre-specified herd size requires the model to account annually for raising male and female calves for both sale and replacement, and purchasing cows if sufficient replacements are not available. The model was tested, refined and validated with data from farms in Kenya before it was applied in this study. A detailed description and an application are given in Nyangito (19921, Nyangito et al. (1994) and Nyangito et al. (1995). In our study, simulation modelling with TIES was used to provide financial and economic indicators for farm performance to obtain a basis for comparison of the farms. 1.4. Objectives This study was done to evaluate the performance of smallholder dairy farms of the upper-midland 1 and 2 agro-ecological zones in Murang’a District through data analysis and simulation modelling, as a basis for evaluating the impact of technical interventions, especially those on the performance of the dairy enterprise. Key indicators for future study of the variation in farm performance of the smallholder dairy farms were identified.
2. Materials and methods 2.1. Study design
Case farms were situated in agro-ecological zones upper-midland 1-2 (UMl-21, and selected from members of the Boyo Dairy Co-operative Society in the Kangema Division of Murang’a District. The case farms were selected non-randomly on the basis of differences in reported milk production, based on the amount of milk delivered to the Society. Eighteen farms were selected, six with a high milk delivery (20-30 kg milk per day), six farms with an average daily milk delivery (IO-20 kg) and six with a low daily delivery (l-10 kg milk). These farms were used to study the influence of farm size (e.g. number of cattle, hectarage), management (e.g. milk production, reproduction) and disease incidence on the economics of smallholder dairy farming. They were considered typical of the production system in UMl-2 of Murang’a District, and were characterised by a limited hectarage, on which food crops for both family consumption (maize and beans) and cash crops (mostly coffee) are grown.
G. van Schoik et al. /Preventive
Veterinary Medicine 29 (1996) 21-36
25
2.2. Data collection
A structured and pretested questionnaire (Nyangito, 1992) was available to collect data from the farms. Individual farmers and members of their households were interviewed through single half-day visits by a veterinarian. The interviews were carried out in the first 2 weeks of November 1994. The two veterinarians administering the questionnaires were familiar with the farming system. Bias resulting from differences in interpretation of data between farms or between the veterinarians was avoided as much as poss:ible by a briefing early in the study. During the briefing the questionnaire was discussed with the two veterinarians and one farm was visited together and the interpretation of the data obtained were compared. Data collected included all farm resources and enterprises, yields, inputs, output and input prices, and production and management practices of the enterprises on the farm. Data on farm assets, liabilities, off-farm investments and income, crop and livestock sales, and consumption as well as food requirements by the household were also collected. Data relied heavily on memory recall of the farmer and that of household members, and on interpretation by the veterinarian. In cases where records were available (such as title deeds for farms, payments for milk sales and other products), the recorded values were used. Annual crop hectarage and yields were based on 1994 results. Respondents were also asked to provide the value of off-farm employment and off-farm investments and to estimate household expenditures on food items as well as non-food items. The Agricultural and Livestock Annual Reports (Ministry of Agriculture, Livestock Development and Marketing, 19941, Statistical Abstracts (Ministry of Agriculture, Livesto’ck Development and Marketing, 1992) and personal communications from animal production and marketing officers were used to value output prices for crops and livestock products. The records of the National Dairy Development Project (NDDP) of Murang’a District cover the period 1988-1994 and contain much data, such as average milk production and fertility performance for the District. The data were collected as part of their extension programme at the NDDP contact farms. Farms belonging to an extension programme have access to more information to improve their farm performance ~(Solet al., 1984). Farms belonging to the NDDP were likely to perform better than other farms in Murang’a District (Kiptarus, 1994; Odima et al., 1994). The NDDP records were used for time-series data required to calculate the probability distributions needed to run TIES and as a basis for comparison between the case and NDDP farms as validation. The case farms were also compared with data derived from studies in Kiambu District neighbouring Murang’a District. The farm system and climate in Kiambu District are comparable with those in the Murang’a District (Gitau, 1994). Data on interest rates and loans originated from commercial banks and the Government’s Agricultural Lending Institutions for loans on land purchase in the District. 2.3. Data validation In the data validation process, data of all case farms were cross-checked with secondary data derived from consultations with animal production officers, reports from
26
G. uan Schaik et al./Preoentiue
Veierinary Medicine 29 11996) 21-36
the Ministry of Agriculture and NDDP farms. The data were consistent and only few extreme values were found. Only household living expenses seemed to be a hard figure to estimate for the farmers. This figure had to be added or modified for a few farms according to estimates of farms with a similar family size, and earlier data from other areas of Kenya (Nyangito, 1992). The averages and ranges of data derived from the case farms were used to calculate the probability distributions. Other necessary information, such as expectations on increase or decrease in prices and yields, was obtained from consultations with animal production and marketing officers, and veterinarians among others. 2.4. Economic analysis indicators Total annual farm expenses are calculated from variable costs incurred in crop and livestock production plus fixed costs for the farm. Variable costs of production for crops include all production expenses, harvesting, and marketing costs. Livestock variable costs include expenses such as: breeding, purchased forage and purchased feed, disease treatment, immunisation, acaricide, anthelmintics, other health costs, and replacement costs. Livestock variable costs are calculated on a per head basis. Receipts and expenses from all farm sources are used to calculate various measures of financial and economic performance of the farm. Key economic variables are net cash farm income, net farm income, net present value (NPV) (discounted value of the difference between the farm’s total receipts and expenses), benefit cost ratio (BCR), and internal rate of return (IRR). The farm annual cash flow and balance sheet information is calculated at the end of the year. The cash flow consists of adding off-farm income to net cash farm income and subtracting family living expenses and principal payments. The balance sheet reflects annual changes in assets (cash reserves, land, value of livestock, etc.), and liabilities (long- and intermediate-term debt). The value of livestock and stored crops is updated annually based on their respective prices. Land values are adjusted annually based on assumed rates of inflation while machinery values are updated to reflect the effect of depreciation and replacement. Liabilities are updated annually to reflect principal payments for initial loans and loans incurred over the planning horizon to replace machinery and to meet cash flow deficits. At the end of the lo-year planning horizon, TIES calculates and saves selected output variables for further analysis. The model then re-initialises all variables to their original input values and repeats the calculation process for the next iteration, using another set of random yields, prices, and livestock production values. After simulating 100 iterations over the simulation period, the model calculates summary statistics for key output variables. The model calculates values for IRR, BCR, and NPV after each iteration. Thus, the 100 values for IRR, BCR, and NPV represent empirical probability distributions that can be used to compare scenarios (Nyangito et al., 1995). Economic success of the farm was analysed using IRR, BCR, NPV and the ratio of ending net worth to beginning net worth and is calculated as the probability that the estimated value for each of these variables was equal or greater than the cut-off level (1 for BCR). Economic success using the ratio of ending net worth to beginning net worth was estimated as the probability of lowering real equity, which was defined as the
G. uan Schaik et al./Preoentiue
Veterinary Medicine 29 (1996) 21-36
27
probability that ending equity to beginning equity was less or equal to 1 (Nyangito, 1992). An indicator for the financial efficiency of the farm is the average annual ratio (%) of cash expenses made for the farm to the total farm receipts for years simulated. The chance of survival is the probability that the farm will maintain an equity-to-assets ratio equal to the minimum level required for borrowing funds from commercial banks (Z!O%),although smallholder dairy farmers usually do not borrow from commercial baniks(Nyangito, 1992). 2.5. ClassQication by farm peflormance A stochastic whole-farm analysis with TIES including off-farm income and household consumption was used as a basis to classify the farms by high, average and low economic performance. A farm with a high performance had a 100% chance of surviving and economic success. The probability of the BCR exceeding 1 was also 100% and the probability of the IRR exceeding the discount rate of 18% had to be greater than 40%. An average-performing farm had a 100% chance of survival and economic success and also a 100% probability of BCR exceeding 1. On the other hand, the prob’ability to have an IRR exceeding 18% was less than 30%. A farm with a low performance had chances of survival and/or economic success less than 100% (range 3-93%) or, in case of one farm with both these chances equal to lOO%, had a chance (12%) of decreasing real equity. Furthermore, for most low-performance farms, the probability of a BCR exceeding 1 was less than 100% (range O-93%) and for all low-performance farms the probability of an IRR exceeding 18% was 0%. 2.6. Data analysis using computer simulation The TIES model was used to derive descriptive statistics and to carry out a sensitivity analysis. The descriptive statistics showed differences among the three farm performance groups in financial indicators, productivity data, expenses and efficiency parameters. Simple regression formulas were used to determine the association between calving interval, milk production, and hectares of cropland with annual net farm income and the costs of concentrates per cow with milk production and calving interval. The importance of the dairy enterprise for the farm was evaluated by entering one extra cow in the simulation of the farms. For that, cropland had to be converted into land to grow Napier grass to make it possible to keep an extra cow. On average, the farms kept one cow per 0.2 ha of Napier grass. Cropland for maize and beans was converted into land to grow Napier grass, while leaving land for maize and beans. When maize and/or beans failed to satisfy family needs, this was shown by a lower value of the household consumption of the farm products and higher costs incurred for food. We assumed that no costs were involved in converting cropland into pasture. The effect of increasing the milk price from on average 12 Kenya shillings ’ (Ksh) to Ksh 15 Ikg- ’ was evaluated. The influence of the calving interval on farm performance
’ 1 Kenya shilling = US$O.42 (November
1994).
28
G. von Schaik et al./Preventive
Veterinary Medicine 29 (1996) 21-36
was determined by decreasing the calving interval on every farm by 42 days (based on a saving of two oestrous cycles of 21 days). Computations were carried out to test if feeding concentrates and health care according to the average level of the high-performance farms would improve the financial results of the low- and average-performance farms. 2.7. Statistical analysis The statistical analysis was performed in Microsoft Exe1 4.0. A one-tailed Student’s t-test assuming unequal variance was used to test significant differences (P < 0.10) with regard to the high farm performance group. Within farm performance groups, Student’s t-test was used to test if significant differences (P < 0.10) occurred due to changing parameters in the sensitivity analysis. Simple regression formulas were used to determine the associations (at P < 0.05) between milk production, costs of concentrates, calving interval, hectares of cropland, and net farm income.
3. Results 3.1. Descriptive analysis using computer simulation
3.1 .I. Description of the farms Table 1 is a summary of the data derived from case farms compared with data from NDDP farms (NDDP Survey, 1994) and a study done in the nearby Kiambu District (Gitau et al., 1994a). Although significance could not be calculated, the data of the case farms seem to be generally comparable to data derived from the other studies. The NDDP farms had a better performance, a higher milk production and a shorter calving interval. Furthermore, the NDDP farms spent less on concentrates but more on health costs (such as costs of immunisation, acaricides and health care).
Table 1 Summary Kenya)
description
of case farms, NDDP farms and farms in Kiambu District (all dairy farms situated in Case farms (n= 18)
No. of cows per farm Cropland (ha) Napier grass (ha) Milk yield per cow (kg year- ’ ) Calving interval (days) Concentrate costs per cow (Ksh) Health costs per cow (Ksh) ’ Only means were available.
NDDP farms (n= 171ja
Kiambu farms (n=90)
Mean
Range
Mcatl
Mean
Range
2.4 1.0 0.6 2384 519 8000 1185
1-4 0.1-3s 0.1-1.6 800-7200 365-730 25Ot- 17000 400-3640
1.8 1.4 0.5 3413 448 6662 2548
4.3 1.1 0.7 2336 633 _
l-24 0.1-9 0.1-8 365-9125 308- I256 _ _
G. van Schuik et al./Preventive
3.12.
Veterinary Medicine 29 (1996) 21-36
29
Description of farm pelformance groups
Milk yield per cow is an average figure for the whole farm for 1 year (the total amount of milk produced on the farm divided by the number of cows on the farm> (Table 2.). The calving interval (Table 2) is the average interval of the cows on the farm (all farms had intervals of 12 months or greater). Pasture per herd is the hectarage of Napier grass grown for the whole herd, including young stock. Furthermore, Table 2 summarises the costs incurred for the dairy enterprise, which are mainly costs of concentrates and health services, and some efficiency parameters, such as net farm income per hectare, net cattle income per cow, net cattle income per kilogram milk, and cattle costs per kilogram milk. The range of the annual net farm income (in 1000 Ksh) was 93.9-788.0 Ksh for the high-performance farms, 43.5-172.8 Ksh for the average-performance and - 5.5 to 57.9 Ksh for the low-performance farms. The average milk production per cow of high- and low-performance farms, 3009 kg and 1240 kg per cow, respectively, is significantly different (t = 3.40, P = 0.01).
Table 2 Averages
(md standard deviations
of the farm performance
Financial indicators BCR IRR Total costs /receipts (%) Annual net farm income ( 1000 Ksh) Off-farm income ( 1000 Ksh) Cattle receipts/total farm receipts (%) Productivity data Milk per caw (kg) Calving inlterval (days) Pasture per cow (ha) Expenses per cow (1000 Ksh) Concentrates Health Total
by an asterisk
Average performance (n=8)
Low performance (n= 5)
Mean
Mean
Mean
SD
25.8 130.4 16 293.3 89.6 18
58 ’ 81.2 158.8 51 *
0.8 5.2 18 40.2 252.7 23
3009 462 0.11
1074 92 0.06
2708 514 0.13
1861 73 0.03
9.7
1.3 1.3 2.2
712.3 293.9 9.2 3.6
136.4 16.4 3.7 1.6
are significantly
different
1.4
’)
SD
15.3 78.5 31 272.2 111.3 30
7.9
Eficiency parameters Annual net farm income ( 1000 Ksh haNet cattle income (1000 Ksh per cow) Net cattle income (Ksh kg- ’ milk) Cattle costs (Ksh kg- ’ milk) Values followed (P 5 0.10).
groups
High performance (n=5)
2.6
10.0
9.3 1.2 11.1
41.5 11.0 * 4.0 * 4.6 from
those
SD
0.9 -5.1 98 * 22.2 * 0 39
1240 * 585 ’ 0.12
0.7 12.0 31 29.8 0 21
451 135 0.04
4.9 5.6 4.8
6.0 * 1.0 8.8
2.7 0.5 3.4
21.5 10.0 3.0 1.7
13.3 4.4 * 1.7 * 7.7 *
21.2 6.5 7.5 4.3
of the high-performance
farms
30
G. van Schaik et al./Preventive
Veterinary Medicine 29 (1996) 21-36
Another significant difference is found with respect to the calving interval, which was 123 days longer for the low-performance farms (t = 1.69, P = 0.07) compared with the high-performance farms. The number of hectares of pasture available per cow is not significantly different (t = - 0.49, P = 0.32) for the low-performance farms. Compared with high-performance farms, the low-performance farms incurred lower costs for concentrates (t = 1.44, P = 0.10) and health services (r = 0.70, P = 0.261, while those incurred by the average-performance farms were not significantly different. Farms of low and average performance appear less efficient than the high-performance farms, which have a higher income per production unit (e.g. cow: t = 3.18, P = 0.01 and r = 2.26, P = 0.03, respectively; kg milk: t = 2.0, P = 0.03 and f = 2.6, P = 0.02). The low-performance farms also incurred more expenses for 1 kg of milk (r= -2.42, P = 0.03). The standard deviation of the net farm income per hectare of high-performance farms is extremely high as a result of the influence of one farm of a low hectarage, with a high income from poultry. Leaving this farm out of the calculations results in a significantly lower annual net farm income per hectare for the low- and average-performance farms (t = 7.49, P = 0.00007 and I = 5.83, P = 0.0001, respectively). 3.1.3. Regression analysis Simple regression analysis (annual net farm income = 527 - 0.80 X CI (in 1000 Ksh); R2 = 0.21, F = 4.29, P = 0.05) showed that the calving interval was significantly associated with the average annual net income of a farm, the shorter the interval the higher the net farm income. Other parameters such as milk production (R2 = 0.34, F = 2.15, P = 0.16) and hectares of cropland ( R2 = 0.02, F = 0.34, P = 0.56) showed no significant influence on net farm income. Some other significant relations were the positive relation of costs of concentrates per cow with milk production (Eq. (1)) and the negative relation of costs of concentrates with calving interval (0 = 635 - 0.15 X kg concentrates; R2 = 0.28, F = 6.16, P = 0.02). Feeding concentrates appears to be an important factor influencing farm performance by increasing milk production and decreasing the calving interval. The cows of the farms investigated often had a negative energy balance for most of the year. The regression analysis for all farms gave the following formula for the influence of concentrates on milk production, assuming that the average price of concentrates is Ksh 10 kg-‘. Y = 527 +2.3X
(1)
where Y is milk production per cow (kg year- ’ > and X is the amount of concentrates fed per cow (kg day-‘) CR2= 0.33, F = 7.97, P = 0.01). Upon a basic milk production of 527 kg, each kilogram of concentrates will result in 2.3 kg more milk. This is slightly higher than figures provided by Unga Feeds Limited (Kenya; producer of concentrates) and by the NDDP, who found increases in production of 2 kg and 1.4 kg milk per day, respectively, for each kilogram of concentrates per day. In further computations, a guarded figure of 1.5 kg milk kg-’ concentrates was used, while the amount of concentrates in the regression equation is based on the costs incurred for concentrates divided by the average price.
G. van Schaik et al./Preventive
Veterinary Medicine 29 (1996) 21-36
31
3.2. Sensitiviry analysis using computer simulation 3.2.1. OjFfam income The exclusion of off-farm income in the stochastic simulation makes the analysis unrealistic, since off-farm income is used to maintain the family and is also used to provide necessary investments for the farm. However, a simulation of the case farms without off-farm income was carried out to estimate the influence of off-farm income on farm performance groups. The chances of survival and economic success of the high-performance farms remained 100% implying that their farm performance did not depend upon off-farm income. The chances of survival and/or economic success of average-performance farms decreased when off-farm income was not included in the simulation. Some of the average-performance farms seem to rely heavily on the off-farm income. 3.2.2. Dairy enterprise The financial performance of the dairy enterprise of the farms was evaluated by entering one extra cow in the simulation of the farms. For that, 0.2 ha cropland had to be converted into land to grow Napier grass to make it possible to keep an extra cow. The financial indicators in Table 3 reflect the effect of increasing the dairy enterprise for every performance group and, on average, for all farms. The effect of an increased milk price on the dairy enterprise and thus on the financial farm petformance was determined by increasing the milk price from, on average, Ksh 12 to Ksh 15 kg- ’ milk (Table 3).
Table 3 Increase of fmancial indicators for farm performance after obtaining from 12 to 15 Ksh, and decreasing the calving interval by 42 days
an extra cow ‘, increasing
the milk price
High performance (n=4)
Average performance (n=S)
Low performance (n=S)
Mean
SD
Mean
SD
Mean
Extru cow BCR IRR Annual net farm income ( 1000 Ksh)
0.25 3.6 26.6
0.4 2.8 21.6
- 0.03 0.8 16.4
0.2 1.5 17.1
0.28 5.6 15.6
0.5 9.2 9.6
Higher mil,kprice BCR IRR Annual net farm income ( 1000Ksh)
2.2 4.9 45.5
3.6 3.6 47.4
0.3 1.4 15.8
0.2 1.0 14.3
0.2 3.7 12.1
0.2 3.5 15.0
Shorter calving interval BCR IRR Annual net farm income (loo0 Ksh)
0.02 0.14 0.7
0.04 0.11 0.3
0.06 * 0.16 1.3 *
0.05 0.13 0.8
SD
0.0 1.14 0.1
a One high.performance farm could not be used because it did not grow maize or beans. Values fohawed by an asterisk are significantly different from those of the high-pcrfonnance
0.0 1.61 1.3
farms.
32
G. uan Schaik et al./Preuentiue
Veterinary Medicine 29 11996121-36
A shorter calving interval is associated with an increase in yearly milk production per cow (Brumby and Sholtens, 1986). However, while an estimate for this figure was not available under Kenyan circumstances an increased milk production was not included in the simulation and only extra income from cattle sales as a result of more calvings per cow was included in the model. The effects of a decreased calving interval on this farm performance are summarised in Table 3. Simulating a conversion of cropland currently used for maize and beans into land for Napier grass and obtaining an extra cow did not significantly increase the annual net farm income. However, high-performance farms benefited most (by Ksh 26644, a 10% increase), while the increase in net farm income of average- and low-performance farms was Ksh 16 408 (20% increase) and Ksh 15 569 (70% increase), respectively. Again the BCR, IRR, and annual net farm income did not significantly increase from an increased milk price. The high-performance farms seem to benefit most. Low- and average-performance farms experience the same benefit from an increase in milk price, although the low-performance farms had a much lower average milk production than the average-performance farms, 1240 kg and 2708 kg milk, respectively. Overall there is no significant benefit for the case farms from a decrease in calving interval of 42 days. However, the average-performance farms benefit more from an improvement of the calving interval than the high-performance farms, their BCR (t = - 1.57, P = 0.07) and annual net farm income (t = -2.06, P = 0.03) significantly increased. This reflects the on average higher calving intervals of the average-performance farms (see Table 2). 3.2.3. Farm improvements In our study, the only factor which significantly influenced milk production was the costs incurred for concentrates (P = 0.01). Although not significant, the total health costs per cow (P = 0.11) also had a positive effect on milk production. In order to check these key factors some computations were carried out on the low- and average-performance farms. The performance of the high-performance farms was used as the standard for the lowand average-performance farms. The high-performance farms had an average expendi-
Table 4 Increase in measures of farm performance those of hizh-oerformance farms
of increasing
Measure of farm performance
Concentrate costs (Ksh per cow) Health costs (Ksh per cow) Milk (kg per cow) BCR IRR Average annual net farm income ( 1000Ksh)
concentrates
and health services to the equivalent
Low-performance (n=5)
Average-performance (n=7)
Mean
SD
Mean
SD
3000 645 390 0.26 3.2 9.3
2041 279 297 0.45 8.8 11.6
2580 674 484 0.05 -0.8 8.0
1045 164 417 0.2 4.0 13.0
of
G. van Schaik et al./Preuentiue
Veterinary Medicine 29 (1996) 21-36
33
ture of Ksh 8000 for concentrates and Ksh 1400 for health care, with an average milk production of 3000 kg per cow and a calving interval of 462 days (Table 2). An improvement of the low- and average-performance farms was simulated by adding costs of concentrates and health until they amounted to Ksh 8000 and Ksh 1400, respectively. Feeding more concentrates and improving health should increase milk production and fertility. We assumed that every extra kilogram of concentrates fed to a cow, at a price of Ksh 10 kg-‘, would increase milk production by 1.5 kg. The results of the analysis are summarised in Table 4. Some farmers’ costs already exceeded the average costs of concentrates and/or health of the high-performance farms. In these cases, the higher costs of concentrates were not changed and the extra amount was translated into an increased milk production of 1.5 kg milk per extra kilogram of concentrates fed. One average-performance farm was not used in the simulation, it already exceeded the good performance farms in costs incurred for concentrates and health considerably, but also had an iaccordingly high milk production. To reach the level of the high-performance farms (800 kg) the low-performance farms increased their costs of concentrates more, their costs of health services less and the milk; production per cow also increased less than for the average-performance farms. However, BCR, IRR and net farm income of the low- and average-performance farms did not significantly increase compared with the base situation of the farms. 4. Discussion 4.1. Dara
The standard deviation in the data was high, sometimes higher than the averages. This wals caused by the small numbers in the farm performance groups and the wide range of their farm performance estimates. However, the resulting farm data were comparable with earlier applications of TIES (Nyangito, 1992; Nyangito et al., 1994; Nyangito et al., 19951, NDDP data (NDDP Survey, 19941, and data found in earlier studies in Kiambu District, neighbouring Murang’a District (Gitau et al., 1994a). Therefore, the data of the case farms were assumed valid and used as a basis for further analysis and simulation. 4.2. Classification In the financial analysis for the classification, a discount rate of 18% was used, equal to the interest rate for saving deposits. However, inflation was assumed zero and the discount rate should be corrected by taking the real interest rate minus inflation. While the inflation rate is very variable and difficult to define, the discount rate is kept at 18%. The high discount rate meant that some form of risk was explicitly included in the simulatilons. 4.3. Sensitivity analysis Off-farm income is an important feature of smallholder farms (Nyangito, 1992; Burger, 1994), the farmer considers the farm not only as a source of income, but also
34
G. vun Schaik et al./Preventive
Veterinary Medicine 29 (1996) 21-36
provides social security and a dwelling for the entire family. The farmer is culturally attached to the place of birth. Hence, farm survival, efficiency and productivity may not be the first objective of farmers with an off-farm income as they do not completely rely on their on-farm income (Burger, 1994). For improvements on these farms, labour constraints might occur, the off-farm income will also demand labour from the farmer or the family members (Burger, 1994). The decreased calving interval had only a limited influence on the farm performance, since only the effect of extra cattle sales were considered. However, the objective of this computation was to ignore the increase in milk production which will occur through the negative relation with calving interval (R = -0.44, P = 0.07). The importance of milk production on farm performance was already estimated by simulating an increased milk price. All farms benefited from a lower calving interval, but the impact was low. Feeding is the basis for production (Brumby and Sholtens, 1986; Omore et al., 1994) and the proper amount of concentrates fed increases milk production and decreases calving interval, assuming an improved condition of the cow. Careful interpretation of the absolute amounts of concentrates in this study is necessary; the same price of Ksh 10 kg-’ of concentrates was assumed for every farmer. Although the average-performance farms incurred more costs of concentrates than the high-performance farms (Ksh 9273 and Ksh 7948, respectively), they did not produce more milk (2708 kg and 3009 kg). This might be a result of an inefficient feeding strategy (not meeting the needs of the cow>, for example, by not feeding more concentrates at the start of the lactation and less when a cow is lower in milk production. Milk production, as well as fertility, profit from a good feeding strategy (Brumby and Sholtens, 1986; Kiptarus, 1994, Omore et al., 1994). The costs incurred for health services were quite low compared with other studies (Nyangito, 1992; NDDP Survey, 1994). Whether this was a result of a good health status, reluctance of the farmer to make the expenses or the low availability of health care, is unknown. However, from consultations with agricultural extension officers and veterinarians, it appeared that farmers were often deprived of veterinary services as a result of the lack of resources of the veterinary department (which is funded by the government). 4.4. Farm performance Since only the amount of concentrates fed had a direct and significant influence on milk production this was the only variable included in the regression formula. The simplified formula will not be realistic since it should contain a lot more factors influencing milk production (e.g. forage yield).
5. Conclusion
Selected smallholder dairy farms in Kangema division of Murang’a District, situated in agro-ecological zone upper-midland 1-2, differed in farm performance as shown by financial indicators. The indicators themselves were strongly influenced by the perfor-
G. uan Schaik et al./Preuentiue
Veterinary Medicine 29 (1996) 21-36
35
mance of the dairy enterprise of the farm (e.g. milk yield per cow, calving interval, costs per cow). Milk production and calving interval were the main indicators describing the performance of the dairy enterprise and, through that, the farm performance. Both milk production and calving interval were influenced by the amount of concentrates fed suggesting that feeding concentrates is an important indicator of high income per farm. The infuence of costs of health services on farm performance was not significant between the farm performance groups. Another factor influencing farm performance was off-farm income; most farms heavily dependent on off-farm income were inefficient in their dairy enterprise and thus performed worse. Future research should further explore the validity and relevance of these key factors, but should also investigate other descriptions of the way the farm is managed. In particular, further investigation of the forage input will be necessary.
Acknowledgements
The authors thank the following for their assistance, comments and contributions: Dr H.O. Nyangito, The Boyo Dairy Cooperative Society, Murang’a, Dr T. Metz and colleagues, National Dairy Development Project, Dr Njuhya and Dr J.O. Donge.
References Brumby, P.J. and Sholtens, R.G., 1986. Management and health constraints of small scale dairy production in Africa. ILCA Bull. No. 25, International Livestock Centre for Africa, Addis Ababa, Ethiopia, pp. 9-12. Burger, C.P.J., 1994. Farm Households, Cash Income and Food Production; The Case of Kenyan Smallholdings. VU University Press, Amsterdam, 215 pp. Gitau, G.K., 1994. Tick-borne diseases project in Murang’a District, Kenya. Preliminary results. International Livesmck Research Institute, Nairobi, Kenya. Gitau, G.K., O’Callaghan, C.J., McDermott, J.J., Omore, A.O., Odima, P.A., Mulei, C.M. and Kilungo, J.K., 1994a. Description of smalIholder dairy farms in Kiambu District, Kenya. Prev. Vet. Med., 21: 155-166. Jaetzold, R. and Schmidt, H.. 1983. Farm Management Handbook of Kenya, Vol. IIB, Central Kenya. Ministry of Agriculture, Nairobi, Kenya. Kenya Government, 1986. Sessional Paper No. 1. Publication on Economic Management for Renewed Growth. Government Printer, Nairobi, Kenya, 60 pp. Kiptatus, J., 1994. Fertility Report 1993. Ministry of Agriculture, Livestock Development and Marketing, National Dairy Development Project, Hill Plaza, Nairobi, 39 pp. Mbogoh, S.G., 1984. Dairy development and internal marketing in sub-Sabaran Africa: Performance, policies and options. LPU Working Paper No. 5, International Livestock Centre for Africa, Addis Ababa, Ethiopia. Ministry of Agriculture, Livestock Development and Marketing, 1992. Statistical Abstract 1991. Ministry of Agriculture, Livestock Development and Marketing, Nairobi, Kenya. Ministry of Agriculture, Livestock Development and Marketing, 1994. Economic Survey 1994. Ministry of Agriculture, Livestock Development and Marketing, Nairobi, Kenya. NDDP Survey, 1994. Results of the Farm Survey in Murang’a District, 1993. Ministry of Agriculture, Livestock Development and Marketing, National Dairy Development Project, Hill Plaza, Nairobi. Norval, R.A.I., Perry, B.D. and Young, A.S., 1992. The Epidemiology of Theileriosis in Africa. Academic Press, London, 481 pp.
36
G. van Schaik et al. / Preuentioe Veterinary Medicine 29 (1996121-36
Nyangito, H.O., 1992. Economic evaluation of alternative livestock disease control methods in Kenya. Ph.D. Thesis, The University of Tennessee, Knoxville, TN, 259 pp. Nyangito, H.O., Richardson, J.W., Mukhebi, A.W., Mundy, D.S., Zimmel, P., Namken, J. and Perry, B.D., 1994. Whole farm economic analysis of East Coast fever immunisation strategies in Kiliti District, Kenya. Prev. Vet. Med., 21: 215-235. Nyangito, H.O., Richardson, J.W., Mukhebi, A.W., Mundy, D.S., Zimmel, P. and Nan&en, J., 1995. Whole farm evaluation of East Coast fever immunization strategies on farms in the Uasin Gishu District of Kenya. Comput. Electronics Agric., 12: 19-33. Odima, P.A., McDermott, J.J. and Mutiga, E.R., 1994. Reproductive performance of dairy cows on smallholder dairy farms in Kiambu District, Kenya.: design, methodology and development considerations. In: G.J. Rowlands, M.N. Kyule and B.D. Perry (Editors), Proc. of the 7th Int. Symp. on Veterinary Epidemiology and Economics, 15- 19 August, The Kenya Veterinarian, Nairobi, Kenya, pp. 366-368. Omore, A.O., McDermott, J.J. and Carles, A.B., 1994. Comparison of productivity of cattle grazing systems in small-holder dairy farms in Kiambu District, Kenya. In: G.J. Rowlands, M.N. Kyule and B.D. Perry (Editors), Proc. of the 7th Int. Symp. on Veterinary Epidemiology and Economics, 15-19 August, The Kenya Veterinarian, Nairobi, Kenya, pp. 121- 123. Sol, J., Renkema, J.A., Stelwagen, J., Dijkhuizen, A.A. and Brand, A., 1984. A three year health and management program on thirty Dutch dairy farms. Vet. Q., 6: 141-169. Walshe, M.J., 1987. Criteria for success or failure of dairy development. In: Milk the Vital Force: Proc. of the 12th Int. Dairy Congress, Reidel, The Hague, pp. 329-340. Walshe, M.J., Grindle, J., Nell, A. and Bachmann, M., 1991. Dairy development in sub-Saharan Africa: A study of issues and options. Tech. Pap. No. 135, World Bank, Washington DC.