Previous Page Table of Contents Next Page

Milk recording in developing countries

Sound dairy management involves keeping good records. Once a government realizes that milk recording should complement breeding programmes, farmers must still be convinced that they have something to gain from keeping records. Here are ways in which milk recording systems in developing countries can be adapted to the needs of farmers and governments.

by U.B. Lindström

The first milk recording organizations were formed in Denmark in 1895 (Johansson, 1961). The recording movement spread rapidly and now plays an important role in all countries with an advanced dairy industry. In some countries (e.g. Denmark, Israel, the Netherlands, Norway and Sweden) more than 50 percent of dairy cows are recorded. On the other hand, in most developing countries production figures on individual animals are usually available for only a tiny proportion of the total cow population (McDowell, 1972). For example, in Kenya the number of officially milk recorded cows is about 10 000 out of a total cow population of some 3 million, of which over 800 000 are high-grade European animals (Livestock Recording Centre, 1974). In many developing countries there is no systematic recording at all. What has brought about this situation? Is it possible to develop simple milk recording systems adapted to the needs of the developing nations? This article discusses these questions and pays special attention to aspects of recording related to the selection of animals.


Milk recording has two main objectives:

  1. To help the individual farmer to produce milk more efficiently, i.e. more economically.
  2. To provide data for government administrative, research, breeding and extension purposes.

Both objectives are of course equally important, but it should be recognized that if the individual farmer cannot be convinced that he/she has something to gain from the practice, it is very difficult to develop extensive recording systems. One should therefore avoid emphasizing the over-all national benefits too much, because otherwise many farmers will believe that milk recording serves only the interests of various official bodies.

The starting point in all recording schemes should be the farmer, how to get him/her to see that it really pays to keep some simple records. If one succeeds in this, the collection of data, and cooperation in other matters, will be much easier.

U.B. Lindström is with the Institute of Animal Breeding, Agricultural Research Centre, Box 18, 01301 Vantaa 30, Finland.

Integration with other services

One of the weakest points of many schemes is that milk recording is carried out more or less divorced from other related activities. Figure 1 shows an example of desirable integration of data collection and dissemination of results in the field of milk production. The activities of the central dairy board or a comparable body supervising the dairy industry should be closely coordinated with those of the AI organization, the extension services, the milk recording organization and the research institutions. Only too often the channels of communication between these organizations are practically non-existent. In Figure 1 information flows in two main channels: “down” from the various governmental and other bodies, and “up” from the farmers and advisers working in the field.

In most developing countries milk recording is either non-existent or very rare. Subsidies and economic incentives are needed to promote record keeping

If only the first channel is used (as is often the case), the scheme cannot be fully effective. A system with built-in feedback which gives both small and large farmers and extension workers adequate opportunities to make suggestions has better chances of developing along the right lines. Stigler (1961) says: “One should hardly have to tell academicians that information is a valuable resource: knowledge is power. And yet it occupies a slum dwelling in the town of economics.”

Applied to the field of milk recording we could say: Listen to the man/ woman in the field, and let us make the best use possible of the information he/she can provide us.

Problems and needs

Apart from the general lack of funds, the main problems in setting up and running milk recording schemes in developing countries are:

  1. Low educational level of farmers.
  2. Lack of qualified extension workers and recorders.
  3. Few incentives for farmers to record their cows.
  4. Small average herd size.
  5. Poor communications.
  6. Unrecognized need for progeny testing and other breeding programmes.
  7. Poor data collection and processing facilities.

Although the problems are serious, they are not impossible to cope with. For example, a small average herd size and difficult communications have not prevented the expansion of milk recording in Norway and Finland.

It is difficult to develop extensive milk recording services in a country if the government does not actively promote them. One of the many ways in which the authorities can promote milk recording is through the price paid to the producer.

A fair (high) price automatically increases interest in producing more milk, i.e. in feeding and managing the animals better and in getting cows of higher genetic quality, and this usually also generates more interest in records.

Another general incentive for creating more interest in records is to pay the farmer not only according to the quantity of milk, but also on the basis of the content of fat and/or protein. In most developing countries the farmers are paid only on the basis of quantity, perhaps with a low minimum requirement on the fat percentage (McDowell, 1972). This does not encourage detailed record keeping and discriminates against breeds with a high solids content in the milk, and thus usually against the indigenous breeds. In addition, there is little justification for producing lowprotein milk by favouring quantity at the expense of quality. Payment should be for both quantity and quality, the latter preferably determined by the protein percentage.

Incentives. General incentives are helpful but are usually not sufficient to get large numbers of farmers to join a milk recording scheme. Therefore, one should seriously consider other ways of encouraging milk producers to keep records. In Finland a popular saying has it that “Money is the best extension worker”. If farmers could actually be paid for keeping records and sending them to the authorities, this would obviously be the most efficient incentive. Even if this system cannot be used generally, it could be tried when starting and building up a scheme in a developing country. And farmers willing to cooperate in a progeny testing programme should certainly be paid a fee for each daughter record.

Other possible incentives are:

Subsidies. Apart from incentives, recording and associated extension activities should be subsidized directly or indirectly by the dairy industry or by the government. In developed countries funds are often accumulated by issuing a small levy (e.g. US$ 0.05 per kg) on all milk delivered to the dairies; milk recording is financed or subsidized with these funds. For example, in Norway the farmer pays about 15 percent of the costs of milk recording, and the rest is financed by the dairy industry or other authorities. In developing countries, subsidies should play an even more important role. At the start of a milk recording scheme, when the number of participating herds is small, farmers should probably not be levied any charge. Even a very low fee per cow seems to put small farmers off joining the recording service. In the beginning it is more important to interest the farmers in recording than to make the scheme self-supporting. Gradually, as the scheme is built up, the services offered are improved and the farmers begin to see the benefits, a small fee can be charged. It should be stressed that cooperation between the milk recording sector on the one hand and AI, extension and research on the other, should be as close as possible. The various subsidies, incentives and services offered should be well coordinated so that they are not in conflict with each other. All field workers should get an all-round traning. AI technicians should, for example, be well acquainted with the milk recording scheme, and similarly the recorders should be familiar with AI operations.

Figure 1. Schematic outline of desired flow of information in milk recording and associated activities

Figure 1

What does the farmer want?

As pointed out earlier, the low educational level of farmers is one obstacle in all improvement programmes in developing countries. How serious is this problem? Are farmers willing to cooperate? Table 1 gives some results of a survey carried out in Kenya at the end of 1972 among small dairy farmers (Lindström and Lindström, 1974). Most of the farmers were already giving extra feed to their cows (about half of the herds were in the high-potential areas of Kenya). Nevertheless, 80 percent of the farmers felt that they needed more advice on keeping dairy cattle. There was no specific need that stood out; most farmers wanted advice on dairy cattle management in general. Only one quarter of the farmers mentioned agricultural and extension officers as sources of information on how to manage their cows. The most common complaint was that the extension officers never or too seldom contacted the farmers. “They are only in towns,” “Never seen them in these parts,” “Officers are only seen in public meetings” were common remarks throughout.

Recording milk yields of Sahiwal cattle in Kenya under field conditions

These findings indicate that farmers would be willing — and in fact would be quite happy — to cooperate with extension personnel. The problem is not only to get enough extension field workers, but also to utilize them effectively. With better planning and organization of the field work, more farmers could be reached and persuaded to participate in milk recording schemes in most developing countries. For example, when Kenya Milk Records employed an extension officer travelling round to farms, there was a substantial increase in the number of participating herds (Livestock Recording Centre, 1974).

Collecting information from farmers

But are small farmers really able to collect useful information? Table 2 shows some of the information obtained from the above survey. The most encouraging result was that most of the farmers surveyed seemed to be quite well aware of how much their cows produce. Of course, the production figures given were not exact, but the farmer was generally able to place his cows in order of merit. From the point of view of enrolling more herds in a milk recording scheme, this is promising. Thus it should be possible to interest farmers in recording the yields in a more systematic and detailed manner once they have started observing their animals. On the other hand, the survey showed clearly that the identity of the sire of the cow in question was hardly ever known, even when the owner reported it to be an AI sire. This was so despite the fact that some of the farmers using AI had been provided with a card and file for keeping track of the matings and calving of their cows. Apparently the farmers did not realize why this should be done and what use there was in knowing the identity of the sire. One can hardly blame them for this, and one should certainly not interpret the result as meaning that it is impossible to get the sires recorded. However, at the beginning it would be desirable to make this task the responsibility of AI technicians, milk recorders and extension officers.

What should be required of a milk recording system in a developing country? The following is by no means a complete list, but the requirements given are probably the major ones:

The routine weighing of the milk from each cow at regular intervals is not in itself the most important aspect of recording. The most important thing is the utilization of the records obtained. Unfortunately, many schemes have given priority to the routine collection and tabulation of figures, and to a large extent have neglected the main task of milk recording. There is little justification for continuously collecting detailed records if they are not properly utilized.

Collecting and processing data

The fewer figures that are collected and the simpler the recording sheet is, the better are the chances of making the scheme work. Table 3 gives examples of recording sheets that would be completely adequate for both the farmer and the official sector (Mahadevan, 1966). When the recording sheet is introduced, two things should be stressed:

Table 1. General information provided by 146 small dairy farmers in Kenya

of all farmers interviewed
Extra feeding practised82
Need for more advice80
Complaints on extension51
Complaints on AI39
Information on management of cows obtained from: 
Farmers' training college
Agricultural and extension officers

Table 2. Information on dairy cattle production provided by 146 small dairy farmers in Kenya1

Proportion of cows with known milk yield%87
Reported average milk yield per cow per day  
After calving
Average monthly income per cowK.Sh49
Proportion of cows with AI sire%57
Sire of cow known%<1

1 Data are for 418 cows, of which about three quarters were European purebreds or crossbreds.

Table 3. Simple recording sheets

A. — Permanent information (one sheet per cow)

Cow's ear (tattoo) No.                                                   Cow's breed                                          

Cow's birth date                                                             Sire of cow                                            

Dam of cow                                            

Lact. No.Service datesBulls usedCalving dateCalfLactation length
305-day production
Total production
% fat or proteinDays dry
DayMonthYearSexEar No.

B. — Test day production

Test dayCow No.Milk yield kg
(AM or PM)
% protein
(% fat)
Remarks on calving, services, drying off, illnesses

In the beginning farmers should get detailed advice from an experienced extension worker on how to keep and use the records. If possible, the farmers should be visited at intervals of two to three weeks, and at least three times in the first two months. After that, regular monthly or bimonthly visits are required to maintain contact and ensure good follow-up

How the farmer's records should be collected and processed depends on prevailing conditions. As long as the number of participating herds is small, the necessary calculations can easily be made on simple desk calculators. As more herds and cows are enrolled, the use of a minicomputer, or even a computer, may become justified. This would ensure fewer errors and rapid processing at a reasonable cost with the modern equipment now available. At present, in many developing countries the same figures are transferred several times before final processing, resulting in many errors. It is important that the farmers get the results as soon as possible. These should relate to current and total production for each cow and herd, fat or protein percentage, service and calving dates, etc. Summaries and recommendations should be sent to the farmers at regular intervals, not just once a year. Especially at the beginning, figures on age at first calving, dry periods, calving intervals, and lactation lengths should be carefully explained and attention drawn to any irregularities.

For comparison, each farmer should also receive averages for other herds in the same area and elsewhere in the country.

Accuracy of individual yields. How simple can a scheme be and still give sufficiently reliable results? The first question in this connexion concerns the interval between weighings and how often samples for fat and protein determinations should be taken. A number of studies, made mainly in Europe and North America, have shown that recording milk once a month is accurate enough for individual selection and once every two months for progeny testing (see literature review by McDaniel, 1969). Table 4 gives some results from a study of 10 herds in Kenya (Lindström, 1976). The average error is small: even for bimonthly testing it is only about 1.5 percent of the actual mean. However, due to the large standard deviations, the individual errors for biweekly testing are of the order of -9 to +12 percent of the actual yield, i.e. -251 to +337 kg. Prolonging the recording interval to one month increases the error to ±400 to ±500 kg. Thus it would seem that in Kenya one should preferably not go beyond a 14-day interval for recording if a relatively high accuracy in selecting individual cows is to be maintained. Moreover, for managerial and educational purposes biweekly testing seems desirable, at least at the start of the scheme. These conclusions may be valid in many other countries with similar conditions. On the other land, as far as selection accuracy is concerned, there is no reason to continue with the present Kenyan system of weighing the milk every day. If large farmers wish to keep daily records in order to check if milk is stolen, it is up to them to do so, but one should not make the majority (i.e. the small farmers) follow the same system.

As regards the collection of samples for fat and protein determinations, several studies (McDaniel, 1969) have shown that monthly recordings generally (at least for fat percentage) give a somewhat lower accuracy than monthly recording of milk yields. For selection purposes they are, however, sufficiently reliable, although more frequent testing would be desirable from a managerial point of view. However, the difficulties in collecting and distributing milk samples in developing countries make this unrealistic in most cases.

Figure 2. Correlations between single-test day production and total lactation yields

Figure 2

Table 4. Accuracy of estimated lactation yields of 1 097 cows of four European breeds and the Sahiwal breed on 10 farms in Kenya, 1969–721

RecordingCorrelation with actual yield (within farm and calving season)Average difference: estimated minus actual yield = d95% confidence interval for estimated individual yields - d ± 1.96 × standard deviation
Weekly0.994391.41-201 to +279-7.3 to +10.1
Biweekly0.991431.55-251 to +337-9.1 to +12.2
Monthly0.978371.34-422 to +496-15.2 to +17.9
Bimonthly0.943421.52-689 to +773-24.9 to +27.9

1 Actual yields were based on daily weighings of milk.
2 Percent of actual average yield (2 767 kg)

Accuracy of progeny testing.

For progeny testing purposes, relatively few observations per individual cow should still give a satisfactory accuracy if enough daughters are sampled. Figure 2 shows average correlations between single-day test production and total lactation yield as calculated by McDaniel (1969) for several studies in the United States, and corresponding results from Kenya (Lindström, 1976). In the latter study the maximum correlation of about 0.5 during the first six months is reached some 10 to 12 weeks after calving. In the former, the highest correlation of 0.87 is reached some 16 to 20 weeks after calving. In Kenya it appears that the wide fluctuations caused by climate and management, in combination with the relatively low absolute level of milk production, result in a lower accuracy of single observations. This may also be true in other developing countries, and it should be taken into account if progeny tests are to be based on singletest day production.

However, even a correlation of only 0.5 is no hindrance for progeny testing bulls. Assuming the heritability of a complete lactation yield, h2c, to be 0.3 (Meyn and Münsterer, 1968), the heritability of a single test day, h2s, can be calculated from:

Thus a progeny test based on 100 to 150 daughters, for example, would still give an accuracy of 0.66-0.74, i.e. sufficient for picking the best bulls.

Getting 100 to 150 daughter records per bull may seem difficult or impossible in a developing country. It obviously depends much on how the recording of yields and the sampling of bulls are organized. By carefully distributing semen from the bulls to a number of selected herds, and preferably by paying the owners for each daughter record, enough bulls can be tested. The testing can be carried out either separately or as part of the ordinary milk recording system. In the former case, the yield of daughters of selected young bulls would be recorded, either by special recorders or by the farmers themselves, for about two to three months at two-week intervals, starting some 10 to 12 weeks after calving. This system would make it possible to test enough bulls per year at a fairly low cost. Generally, it would of course be better if the progeny tests could be calculated as a by-product of ordinary milk recording. But this is often not possible, and one should then not hesitate to set up a separate progeny testing scheme; later on it can easily be incorporated into the routine milk recording system. In Kenya the experiences of concentrating the testing of batches of young bulls in large governmental and private herds have been good (Livestock Recording Centre, 1974).

How to test

In order to extend recording to as many herds as possible, to keep costs down, and to employ recorders gainfully (i.e. in actual extension work), the routine weighing of the milk should be left to the farmer. The recorders should, of course, supervise the weighings periodically, but this is not their main job. Moreover, there is no need to weigh the milk in both the morning and the evening. Many studies (McDaniel, 1969) have shown that alternate morning-afternoon recordings are almost as accurate as morning and afternoon recordings. Also in Kenya, one weighing a day gave only slightly lower correlations with actual yields than two weighings a day (Lindström, 1976). How then should the weighings be done? If the farmers can be provided with scales or measuring cylinders free of charge, the yields can be estimated fairly accurately. The equipment now available is so simple that any farmer can easily learn to master the technique. But if exact measuring equipment cannot be provided, the yields can still be estimated with satisfactory accuracy for most purposes. For example, the buckets, pails, gourds, bottles, etc. ordinarily used by farmers when milking could be marked into 1- to 2-kg sections by use of exact measuring equipment. Even illiterate farmers will then be able to set down the necessary figures satisfactorily. In the survey mentioned earlier, small farmers in Kenya were generally able to state fairly accurately the number of bottles or gourds of milk of a certain volume each cow produced. It would, of course, be preferable if all farmers could be provided with a simple bucket or bottle premarked into easily readable sections. The dairy industry would certainly benefit from the widespread use of some kind of recording instrument, even if these have to be supplied free.

Protein percentage

In the developing countries, many of which suffer from an acute shortage of animal protein, it is often suggested that protein determinations of the milk should be included in the recording scheme. The cheapest equipment (half automatic) for measuring protein content now costs around US$ 2 200 to 2 500. Fully automatic machines with a capacity of 150 to 180 samples per hour (also determining the fat and lactose content) are available for US$ 65 000 to 70 000. With a very limited number of such machines, protein determinations could be made centrally for a whole country. The milk samples can be sent for determinations either by the recorders or by the farmers themselves, or they can be collected in connexion with milk deliveries to the dairies. For example, in Finland, after collection and addition of preservative (potassium bichromate), 50-ml test bottles are sent to a central laboratory by mail. The results are returned by mail to the farmer either directly or via the recorder. There is no reason to assume that similar systems would not work in developing countries. In the beginning, farmers must naturally get detailed instructions and see demonstrations on how to collect the samples. The recorder should supervise the sampling the first two or three times; after that the farmers should be able to do it themselves.

When introducing new practices, it is important to remember the woman in the house. Usually it is the men who are approached, talked to and sent to attend courses, whereas in practice it is the women who often do the actual job, and are therefore the ones more in need of advice. In fact, in the developing countries women would probably be as good or even better extension workers/milk recorders than men.

The importance of breeding programmes

Had the need for sound progeny testing and other breeding programmes been generally recognized in developing countries, milk recording would certainly have got more support. The argument against investing in breeding programmes is often: “The genetic capacity of the animals is not the limiting factor; we must first improve the veterinary, feeding and management practices, and then we can set up breeding schemes.” Such statements completely ignore the fact that good records are the very basis for sound management of animals. Moreover, who can really say when the genetic capacity is the limiting factor and when it is not? Perhaps it seldom is in the high-potential areas of the developing countries where foreign (European) stock can be used, although one can not be too sure about that either. But in medium- and lowpotential areas the genetic capacity certainly is a limiting factor, and it is difficult to see how that can be improved without good breeding programmes based on adequate recording systems. Furthermore, if a strain much better suited to a poor environment than existing ones can be developed by breeding, it becomes easier to spread its genes — much easier in fact than it is to tell farmers and nomads how to improve the environment of their animals.


The starting point in all recording schemes should be the farmer. The recording scheme should be intimately integrated with the activities of the extension service, the AI organization, the research institutes and other bodies working in the same field. Farmers, recorders and extension workers should be given ample opportunities to make suggestions about the scheme.

The authorities should actively promote milk recording through subsidies and incentives. Possible incentives are:

In Kenya a survey showed that most small farmers, many of whom kept no records, knew the approximate production of their cows. When taking into account the fact that the farmers were also very eager to get more advice on dairy cattle management, it should be possible to set up simple recording schemes based on farmers' weighings.

The recording must be followed up by regular extension visits. Advice should be given not only to men, but also to women, as the latter often carry out most of the practical work. In order to encourage production of animal protein, it would be advantageous to include determinations of the protein percentage of the milk in the recording scheme.

Under Kenyan conditions, biweekly testing for milk yield gave errors of -9to + 12 percent of the actual yield, whereas monthly testing increased the error to -15 to +18 percent. For managerial and educational purposes, biweekly testing seems advisable. The testing could well be done on an alternate morning-afternoon basis. For progeny testing purposes, 100 to 150 daughters per bull, recorded for two to three months at biweekly intervals starting 10 to 12 weeks after calving, would give an accuracy of 60 to 70 percent.


Ashton, E.D. 1956. Milk and butterfat recording. A world survey. Farnham Royal, Bucks, Commonwealth Agricultural Bureaux. Commonwealth Bureau of Dairy Science, Technical Communication No. 3.

Johansson, I. 1961. Genetic aspects of dairy cattle breeding. Urbana, University of Illinois Press. 259 p.

Lindström, U.B. 1976. Studies on milk records from Kenya. 1. Accuracy of milk recording with various intervals. Z. Tierziicht. ZüchtBiol., 93. (In press)

Lindström, U.B. & Lindström, K.E. 1974. Survey of dairy cattle breeding on small-scale farms in Kenya. Department of Animal Production, University of Nairobi. Technical Paper No. 1/ 1974.

Livestock Recording Centre, Kenya. 1974. Annual report. Naivasha, Kenya, Ministry of Agriculture.

Mahadevan, P. 1966. Breeding for milk production in tropical cattle. Farnham Royal, Bucks, Commonwealth Agricultural Bureaux. Commonwealth Bureau of Animal Breeding and Genetics, Technical Communication No. 17. 150 p.

Mcdaniel, B.T. 1969. Accuracy of sampling procedures for estimating lactation yields: a review. J. Dairy Sci., 52: 1742-1761.

Mcdowell, R.E. 1972. Improvement of livestock production in warm climates. San Francisco, W.H. Freeman. 711 p.

Meyn, K. & Münsterer, F. 1968. Pflanzliche und Tierische Produktion in den Tropen und Subtropen. In Handbuch der Landwirtschaft und Ernährung in den Entwicklungsländern. Band 2, Rinder, p. 729–775.

Stigler, G.J. 1961. The economics of information. J. pol Econ., 69: 213–225.

Previous Page Top of Page Next Page