Evolution of Milk Production Systems in Tropical Latin America and its interrelationship with Markets: An Analysis of the Colombian Case

Federico Holmann, Libardo Rivas, Juan Carulla, Bernardo Rivera, Luis A Giraldo, Silvio Guzman, Manuel Martinez, Anderson Medina and Andrew Farrow

International Center for Tropical Agriculture (CIAT) and
International Livestock Research Institute (ILRI)
F.Holmann@cgiar.org

Abstract

The objectives were to: (1) identify and quantify the effect of technological change on productivity, profitability, and competitiveness in different milk production systems and regions of the country; (2) analyze the relationship between productivity, technological change, profitability, and competitiveness; (3) analyze the evolution of milk production systems in Colombia; and (4) discuss the market concentration and its impact on the formation of milk price. Data came from a survey to 545 farms during the year 2000 in five regions: Caribbean and Piedmont in the lowlands, Coffee Growing, Antioquia, and the Cundiboyacense altiplanicie in the highlands. The survey was designed to quantify inputs and products in order to determine costs and prices at the farm level in order to calculate (a) variable costs for feed supplementation, labor, health, reproduction, fertilization, and irrigation; (b) gross income from the sale of milk and beef, and (c) to characterize farms according to productivity level and management practices. The statistical analysis of multiple correspondence and general linear models were used to explain the variability observed between productivity and profitability as a function of technological change.

Independent of the production system or the region where farms were located, the increase in competitiveness was in direct relationship with herd size. Thus, as herd size increased, production costs per unit of milk and beef decreased, net incomes per cow increased, and the return to capital investment improved. However, when this increase in competitiveness was associated with increases in productivity, this trend was not observed, which suggested that highly productive farms were not necessarily competitive. The dual-purpose system was the most profitable one in the Piedmont, Caribbean, and Coffee growing regions while in Antioquia and in the Cundiboyacense altiplanicie the most profitable was the specialized dairy system.

With regards to technological change, the adoption of improved pastures and the investment in pasture divisions for a more efficient rotation generated higher productivity and income in all regions and production systems, as well as increased competitiveness through a reduction in production costs per unit of milk and beef. The use of strategic feed supplementation to the basal diet of forage had mixed effects. The best economic response to this supplementation in lowland regions (i.e., Piedmont and Caribbean) was with low quantities (i.e., < 0.5 kg DM/cow/day) of feed supplements while in highland regions (i.e., Coffee Growing area, Antioquia and the Cundiboyacense altiplanicie) was with moderate quantities (i.e., between 0.5 and 2 kg DM/cow/day). The use of fertilization and irrigation increased productivity, but reduced net income and increased production costs, except in the Cundiboyacense altiplanicie. The practice of milking twice a day increased both productivity and profitability and reduced production costs, except in the Caribbeanregion. Farms that de-wormed milking cows with low frequency against internal and external parasites obtained higher incomes and lower production costs in comparison with farms that de-wormed cows with higher frequency although there were no differences in productivity. The amount of years of experience of farmers at producing milk was a key factor to increase profits, although not productivity. Farms located in sites where the commercial value of land was high (>US$6,000/ha) and near market centers had higher productivity that those with commercial value of land medium ($3,000 to $6,000/ha) and low (<$3,000/ha) but were less profitable in all regions.

Box 1: Recommendations to researchers and extensionists

Identify profitable technologies. Technologies that increase productivity are not necessarily profitable, which suggest the need to determine appropriate ways to evaluate them economically. This was the case of fertilization and irrigation. It is necessary to determine the best economic response to various levels of N₂ and H₂O to different species of improved grasses under various soil types and conditions.

The most competitive and profitable breed group in the dual-purpose system was the crossbred with low (24% European-76% Zebu genes) and medium levels of dairy genes (55% European-45% Zebu genes) but had lower productivity than the purebred group (98% European genes). In the specialized dairy system, the purebred group was slightly more profitable, productive and competitive than the crossbred group with medium level of dairy genes, but this difference was not significant.

The Colombian dairy sector has become more productive and competitive, but less profitable. Comparing the evolution of dairy farms with studies 12 years ago, milk production per hectare increased by 44% in dual-purpose herds and 14% in specialized dairies. This increase in productivity reduced the milk production cost by 16% and 10% in dual-purpose and specialized dairies, respectively, due to an increase in stocking rate by 15% and 17% in dual-purpose and specialized dairies as well as to an increase in investment in infrastructure and equipment by 258% and 37% in dual-purpose and specialized dairies, respectively. However, the net income per hectare during this period decreased by 27% and 69% in dual-purpose and specialized dairies due to a reduction in the producer's price of milk and beef of 22% and 20% in dual-purpose systems, and of 41% and 27% in specialized dairies.

Nevertheless, this reduction in price to producers was never translated in lower prices to consumers, but remained in the hands of supermarkets and milk processing plants which expanded and modernized with long-life technology. Development agencies must internalize the fact that policies oriented to markets will be increasingly "oriented to supermarkets". If one adds that in Colombia exists 3 or 4 supermarket chains that control the food retail market, the conclusion is that sectoral policies will need to learn how to deal with a handful of giant companies. This in a huge challenge, and demands an urgent review of ideas and strategies.

Box 2: Recommendations to decision makers

Regionalize research. Due to the fact that the most profitable production systems are region-specific, Colombia should have different strategies for research and technology transfer in order to exploit more efficiently the comparative advantages of each region and production system.

Promote collective action. It is necessary to promote cooperatives and associations to help small producers to adapt to new patterns with higher levels of competition. Otherwise, the new rules of the game could induce a massive exodus of producers in the short term and in a relatively brief period of time.

It is possible, in the short run, to adopt technologies that increase milk productivity and reduce production costs while profits are reduced as a result of falling real prices as occurred in Colombia during the 90's. However, in the long run, this situation is simply unsustainable.

The proposals and challenges presented in this case study have illustrated the problems and opportunities of the dairy sector in Colombia. However, these systems could represent similar situations in other countries of Latin America. Given the phenomenon of globalization and higher degree of efficiency that these systems are being exposed to, the issues of productivity, technological change, competitiveness, and markets, are critical and of enormous relevance for the performance and survival of the livestock sector in the next decades.

Box 3: Recommendations to producers

Without research there is no future. Efficiency goes hand in hand with technology and this depends on research and technology transfer. However, public funds allocated to agricultural research are being reduced. The challenge consists that producers in Colombia take greater control of livestock research by building alliances with local, regional and international organizations leaders in forage and livestock research. For this it is necessary that producers define and fund their own research agenda.

Production-to-consumption participation. In the coming years, producers cannot limit themselves to participate only in the primary phase of production, but to expand their scope of action to other phases of the market chain to have a higher participation in the formation of milk prices and to capture a greater piece of the final price.

Introduction

Cattle production has traditionally been one of the main activities of Latin America's agricultural sector. The region's extensive savannas and profuse forests make it most suitable for livestock production. Latin America and the Caribbean (LAC) currently have a total area of 602 million hectares under permanent pastures and a cattle inventory of 359 million heads, of which 40 million (11%) correspond to milk cows (FAO 2002).

The tropical belt of LAC harbors most of the forage and livestock resources - 72% of grazing land, 82% of total livestock and 88% of milk cows (FAO 2002). In 2001, livestock production in tropical LAC accounted for 13% of world livestock production and 35% of that of developing countries all together.

Despite the region's enormous provision of forage resources, livestock production in tropical LAC faces serious problems in terms of quantity, quality, and productivity of pastures, especially during prolonged dry periods. The problem is widespread, mainly because a high fraction of the available forage base is composed by native pastures that are adapted but present low productivity and by highly degraded introduced species. Multiple production systems coexist in tropical livestock production at different thermal floors, with different degrees of intensification, and located in highly diverse socioeconomic environments.

Furthermore, a great deal of internal discussion exists within LAC countries regarding the viability of these systems in an open economic environment, especially now that the entry into the Free Trade Area of the Americas (ALCA, its Spanish acronym) is being negotiated to compete openly with North America. This study attempts to analyze milk production systems, using Colombia as case study, and determine their importance, limitations, and economic and technical possibilities within the context of small livestock producers and the competitiveness of regional livestock production.

The Case of Colombia

Growth and milk production systems

The dairy activity in Colombia has been very dynamic over the past 30 years. In the 1970s, it grew at an annual rate of 4.7%, then presented an exceptional, sustained growth of 6.5% during the 1980s, and grew at an annual rate of 3.8% during the 1990s, producing approximately 5,877 million liters of fluid milk in 2001 (Balcázar 1992; FEDEGAN 2002).

Two types of milk production systems exist in Colombia: (1) specialized dairy and (2) dual-purpose. Of the estimated 25 million heads composing the national herd, the cattle population found in milk producing farms is estimated at some 6 million, of which 89% are in dual-purpose production systems that account for 55% of the country's milk production (CORPOICA 1998).

The specialized dairy system consists in milking the cow without the calf close by and the male calf is usually sold a few days after birth. Cows are usually purebred or with a high percentage of genes from European Bos taurus (i.e., Holstein) breeds and are supplemented with feed concentrates. As a result, milk productivity is high. On the other hand, the dual-purpose system consists in raising the male calf and selling it after weaning. The cow is milked with the calf close by. Furthermore, these cows have a high percentage of Bos indicus (i.e., Brahman) genes or are crossed with Bos taurus breeds. Their feeding is based on extensive pasture-based systems with low milk and beef productivity (Arias et al 1990).

Compared with specialized dairy production systems, several advantages of the dual-purpose system are: (1) reduced risk because of variations in milk and beef prices; (2) lower incidence of mastitis because of suckling of calves; (3) reduced need for capital investment; and (4) fewer requirements of technical support (Seré 1983).

Most specialized dairy systems are located in the tropical highlands, in regions with cool-to-cold climates located near urban centers. Dual-purpose systems, on the other hand, are usually located in lowland regions with high temperatures and further away from markets. Twelve years ago, Aldana (1990) reported that specialized dairy production systems were more profitable than dual-purpose milk production systems. He also reported that this profitability was directly associated with the level of technology used by the producers. That is, farms presenting higher productivity were more profitable regardless of the production system used.

Contribution of the livestock sector to the Colombian economy

Around 1994, milk and beef production accounted for 25.2% of Colombia's agricultural gross domestic product (GNP) (Lorente 1996), more than doubling the 12.2% attributable to coffee and higher than that of all semi-annual crops put together (24%). By year 2000, livestock production had increased its share to 29.9% of agricultural GNP (DANE 2002). Livestock production as generator of employment has gained importance within the agricultural sector and in the economy as a whole. In 1999, this sub-sector generated 1,400,000 permanent jobs, equivalent to 19.8% of the permanent work force in rural areas (Martínez et al 2002).

Livestock products (beef and milk) play a significant role in the economy's aggregate demand. According to national records, beef, milk, and their derivatives accounted for nearly 13% of total domestic consumption expenditures and for more than 45% of food expenditures in 1989 (Balcázar 1992). The rapid growth of the dairy sector allowed milk consumption by the population to increase from 57 liters per capita in 1970 to 136 liters in 2001, representing an increase of 138% in 30 years (Balcázar 1992; FEDEGAN 2002).

In terms of foreign trade, Colombia is practically self-sufficient in milk production. During the 1990s, the country imported an average of 2% of its production per year (FEDEGAN 2002). Although from time to time the country has been a net exporter of beef over the past 20 years, the relative importance of this activity has visibly diminished since the early 1990s. In 1991, 5% of domestic production was exported. From then on, exports have been decreasing and since 1996 the country exports less than 1% of its total beef production (FEDEGAN 2002).

Market concentration in hands of supermarkets

A trend recently observed, parallel to the concentration of the population in large urban centers, is the growing participation and concentration of supermarkets in the chain of food distribution and sale. In many countries, distribution is handled by a small number of commercial firms, which give them great negotiation power in terms of deciding which products are offered on the shelves as well as pricing and forms of operation (Castro et al 2001). Reardon and Berdegué (2002) reported that during 2000 supermarkets in Latin America billed 60% of retail food sales. In 1990, this percentage barely reached 20%. In the case of Colombia, by the year 2000, supermarkets billed 38% of retail food sales with a forecasted annual increase of 7% for the first decade of the millennium (Hernández 2000).

This structural change has modified the parameters underlying milk marketing. Nowadays, supermarkets have more negotiating power before the milk processing plants, not popular neighborhood stores as in the past. This change has directly influenced the mechanism of fixing the milk price received by producers.

Addressing the problem

Colombia has a proven capacity to increase milk production and pressing socioeconomic reasons to develop the sector. However, there is internal discussion regarding the most suitable technologies to achieve this development, and whether they alone will be sufficient to make livestock production competitive within and outside the region under a scheme of open, unsubsidized economies (Consejo Nacional Lácteo 1999).

Objectives

The objectives of this study were to: (1) identify and quantify the effect of technologies on the increase in milk production, profitability, and competitiveness in both specialized dairy production and dual-purpose systems in different regions of the country; (2) analyze the relationship between productivity, level of technology, profitability, and competitiveness; (3) analyze the evolution of milk production systems in Colombia; and (4) discuss market concentration and its impact on the formation of milk price received by producers.

Methodology

Data came from a survey of 545 farms carried out from February to November 2000. Farms were distributed in five regions as follows: (a) 145 farms in the savanna piedmont (departments of Arauca, Casanare, and Meta); (b) 116 in the Caribbean region (departments of Atlántico, Guajira, Magdalena, César, Bolivar, and Córdoba); (c) 105 in the coffee-growing region (Department of Quindío, northern Valle del Cauca, Caldas, and Risaralda), (d) 97 in Antioquia's altiplanicie, and (e) 82 farms in the altiplanicie of the departments of Cundinamarca and Boyacá. These five regions account for more than 80% of the nation's milk production (FEDEGAN 2002) and were accordingly chosen because they are Colombia's main dairy areas, together with the southern region (Department of Nariño), which was not covered due to lack of funds. The goal was to survey an average of 100 farms per region (for a total of 500 farms) to obtain the variability needed for statistical analyses.

The survey was designed to quantify inputs and products so that farm-level costs, prices and management practices could be determined. Data were then used to calculate not only the variable costs of supplementation, labor, animal health and reproduction, fertilization, and irrigation, but also the gross receipts from sale of milk and beef, as well as to characterize farms according to levels of productivity and technological change. The survey instrument can be found in Holmann et al (2003).

The variable costs of supplementation, labor, animal health and reproduction, as well as gross sales, net income, and milk and beef production costs were estimated using the methodology described by Holmann et al (1990). A total of 55 additional variables were created to consolidate and synthesize the information gathered in the surveys. These variables are described in Holmann et al (2003). In addition, all figures reported in this study were converted to US dollars at the existing average exchange rate for year 2000 (US$1 = Col$2,094).

For this study, competitiveness was defined as the capacity of the producer to remain in the dairy business and was measured by the unitary milk and/or beef production cost. Thus, the lower the production costs, the greater the competitiveness. Profitability was defined as the annual net income per cow or hectare as well as the farm annual net income divided by the total capital invested in the farm, represented by land, livestock, facilities, and equipment. Technological change was measured using the concept of productivity, which was expressed as milk and beef production per cow and per hectare.

Surveys were carried out under the coordination of the animal production faculties of the following universities: Universidad de los Llanos for the Piedmont region, the Fundación San Martín for the Caribbean region, the Universidad de Caldas for the Coffee-growing region, and the Universidad Nacional (Medellín and Bogotá campuses) for Antioquia and the Cundiboyacense altiplanicie. These academic centers contacted the main milk buyers in each of the regions. In the Coffee-growing region, Antioquia, and Cundiboyacense altiplanicie, buyers were mostly processing plants. In the Caribbean and Piedmont regions, milk buyers were intermediaries and/or owners of small-scale cheese factories and, to a lesser extent, processing plants. These milk buyers provided academic centers with a list of their routes with the names of the farms. Routes were chosen based on their accessibility and whether four-wheel drive vehicles could enter. Interviewers were mostly undergraduate students who visited farmers accompanied by those responsible for collecting milk. This ensured a level of trust that would warrant the success of the survey. The students were also trained to execute the survey and were permanently supervised by the co-authors of this study in their respective regions. Of survey respondents, 73% were farm owners.

During November 2002, two managers from the five largest supermarket chains of the country and two milk-processing plants in the city of Cali were interviewed to understand the marketing mechanisms of dairy products in supermarkets. In addition, a telephone poll was carried out in November 2002 to analyze the evolution of the installed capacity of 30 processing plants belonging to the 13 largest dairy companies in Colombia to relate their marketing strategy with the type of technology used to upgrade installed capacity during the 1990s. Information on the poll can be found in Holmann et al (2003).

Statistical analysis

Descriptive statistics were estimated by region and production system to characterize surveyed farms. The general linear model technique was used to prove several hypotheses about the effect that different technologies have on animal productivity. This technique allows the variability observed in different productivity indicators to be expressed as a function of the different technologies, regions, and production systems. No interactions occurred because of lack of data in the different combinations of technology categories. Therefore the model used only estimated the main effects of each technological change.

Furthermore, the averages of each indicator were estimated and compared for each combination of technology*region and/or technology*production system. A multivariate principal component analysis was performed to analyze milk production systems and also to group those farms with similar characteristics on the basis of component values. Ward's clustering methodology (1963) was also used. Multiple correspondence analysis was used to establish relationships between:

Productivity, defined as milk production per cow per day (Milk);

Profitability, defined as net income per cow per year (Income) and annual return to capital invested (Return); and

Technological change, defined by the following technologies or management practices:

Proportion of pasture area with improved pastures (Impfor);
Amount of feed supplements offered to milking cows (Asup), in kg DM per cow per day;
Reproductive system used: artificial insemination (AI); natural service (NS); or both reproductive systems (Both);
Specialized dairy production system (Spec) or dual-purpose (Dpurpse);
Frequency of milking per day (Milking);
Use of fertilizer on pastures (Ferpa);
Use of irrigation on pastures (Irripa);
Mature body weight (BW);
Proportion of cows in milk (Mherd);
Number of grazing paddocks used by milking cows (Paddock);
Market value of land where farm is located (Land);
Years of experience producing milk (Xprod);
Herd size (Herd);
Frequency of internal de-worming (Idew); and
Frequency of external de-worming (Edew).

Results

Farm characteristics

Brief description of regions

The Piedmont and Caribbean regions are located in the lowland tropics of Colombia with an annual mean temperature ranging between 27 ºC and 30 ºC, whereas the Coffee-growing region, Antioquia and the Cundiboyacense altiplanicie are located in the colder highland tropics with mean temperatures that fluctuate between 14 ºC and 22 ºC (Table 1). In the Piedmont, the Coffee-growing region and Antioquia, the annual mean precipitation is higher (2,400 to 2,700 mm) and the rainy season lasts about 8 months, whereas in the Caribbean region and the Cundiboyacense altiplanicie, annual mean precipitation is lower (1,100 to 1,200 mm), with a rainy season that last around 6 months.

Similarly, the topography of the Piedmont and Caribbean regions and to a certain extent the Cundiboyacense altiplanicie is very flat, with few undulated areas, whereas the Coffee-growing region and Antioquia are located in mountainous areas where undulated and broken topography prevails (Table 1).

Table 1. General characteristics, production systems utilized, and existing public infrastructure where surveyed farms were located in each of the five regions during 2000.
Parameter	Region
Parameter	Piedmont (n=145)	Caribbean (n=116)	Coffee (n=105)	Antioquia (n=97)	Cundi-Boyacense (n=82)

Mean annual temperature (^oC)	27.2	30.4	22.0	15.2	13.9
Mean annual precipitation (mm)	2,698	1,203	2,418	2,374	1,105
Length of rainy season (months)	8.0	6.5	7.7	7.8	6.2
Farm topography (%)
- Flat	79.3	74.2	14.7	14.6	51.7
- Ondulated	16.6	22.4	64.2	47.9	34.6
- Steep	4.1	3.2	21.3	37.4	13.1
Farms with following services (%)
- Electricity	90	65	100	96	98
- Telephone	25	57	83	56	70
- Water from municipality	47	18	59	28	63
- Own water resources	68	95	80	87	100
Production system utilized
- Specialized dairy (#)	4	0	37	75	60
- Dual-purpose (#)	141	116	68	22	22

In addition, public infrastructure varied broadly among the regions - from fewer services in the Caribbean region, especially electricity, to better infrastructure in the Coffee-growing region and the Cundiboyacense altiplanicie. The dual-purpose production system was most used in the Piedmont (97% of farms surveyed), Caribbean (100%), and Coffee-growing regions (65%), whereas the specialized dairy system prevailed in Antioquia (77%) and Cundiboyacense altiplanicie (73%).

Land use and pasture management

The scale of operation differed broadly among production systems and regions. The dual-purpose farms, located mainly in the Piedmont, Caribbean, and Coffee-growing regions, were larger than specialized dairies located mainly in Antioquia and Cundiboyacense altiplanicie (Table 2). Around 80% of total farm area was sown to pastures in dual-purpose systems compared with more than 90% in specialized dairies. The area under forest was less than 10% of total farm area in all regions and in all production systems.

Table 2. Land use, proportion of pasture area under improved forages, stocking rate and general pasture management by milk production system and region of the country in 2000
Parameter	Production system		Region
Parameter	Dual purpose (n=333)	Specialized dairy (n=212)	Piedmont (n=145)	Caribbean (n=116)	Coffee (n=105)	Antioquia (n=97)	Cundi-boyacense (n=82)

Land use (ha / farm)
- Total area	164	47.5	94.6	300	72.0	50.7	47.6
- Area improved pastures ¹	110	30.9	68.7	200	46.0	26.1	34.3
- Area native pastures ²	31.4	8.3	12.7	70.4	15.3	14.0	8.9
- Crops	5.7	1.3	2.2	9.9	5.6	0.7	0.5
- Forest	8.0	3.9	8.8	9.5	4.0	4.7	3.0
- Other	5.3	2.2	2.2	10.5	1.1	5.2	0.9
Improved forages (%)	78.0	78.8	90.8	65.5	77.9	71.3	91.0
Stocking rate (AU / ha)	1.47	2.68	1.22	1.33	2.35	2.66	2.70
Resting period in pastures grazed by milking cows (# days)
- Rainy season	31.3	42.1	27.0	35.4	33.4	43.4	45.1
- Dry season	32.6	42.0	26.7	38.9	33.4	43.1	45.3
Weed control frequency (# / yr)	3.5	2.2	2.5	2.0	6.2	2.9	1.0
Pastures fertilized? (% farms)
- Yes	42.6	85.4	27.6	38.8	78.1	93.8	79.3
- Proportion fertilized (%)	29.3	71.7	35.5	22.5	47.8	81.8	65.4
- Fertilizer (kg N/ha/yr)	103	193	77	47	172	213	155
- Frequency applications (# / yr)	4.2	7.0	1.6	1.5	7.1	7.7	6.5
Are pastures irrigated? (% farms)
- Yes	12.6	32.1	1.4	24.1	15.2	20.6	53.7
- Proportion irrigated (%)	58.8	51.0	27.7	40.0	32.2	36.8	52.7
Grazing paddocks (# / farm)	17.2	26.5	8.8	14.4	32.2	29.6	26.2
- paddocks for milking cows	7.3	14.6	4.2	5.1	16.8	15.4	13.2
- paddocks for dry cows	4.1	7.9	2.7	3.3	7.6	8.9	7.5
- paddocks for rest of herd	5.8	4.0	1.9	6.0	7.8	5.3	5.5
¹ Brachiaria brizantha , B. decumbens, B. humidicola, B. mutica, Cynodon nlemfuensis, Panicum maximum, Hyparrhenia rufa, Echinocloa polystachya, Dichanthium aristatum, Andropogon gayanus, Penisetum clandestinum, and P. purpureum ² Mainly Bothriochloa pertusa found in dual-purpose systems located in lowland areas (Piedmont and Caribbean regions) andLolium multiflorum, Poa trivialis,Dactylis glomerata, and Paspalum notatum found in specialized dairies in highland areas (Coffee, Antioquia, and Cundiboyacense altiplanicie).

Both production systems also presented a similar proportion of grassland established with improved pastures (79%), although this proportion differed among regions, being greater in the Cundiboyacense altiplanicie and Piedmont and lower in the Caribbean region. Brachiaria species were more common in dual-purpose systems, whereas Penisetum clandestinum was more common in specialized dairy systems. For more information on grass species established on farms per region, please refer to Holmann et al (2003).

The stocking rate in specialized systems and in high-altitude regions was twice that reported in dual-purpose systems located mainly in the Caribbean and Piedmont regions. This was attributed to the fact that a greater proportion of specialized dairies not only fertilized and irrigated their pastures, but also applied a higher level of nitrogen more frequently and supplemented animal diets with greater amounts of feed concentrates than dual-purpose farms (Tables 2 and 4). In addition, specialized dairies had a higher number of grazing paddocks (Table 2), in smaller areas, thus allowing better use of the amount of forage biomass produced. This rotational management in specialized dairies, accompanied with pasture fertilization, could explain the lower frequency of weed control (2.2 times/year) compared with dual-purpose farms (3.5 times/year), which were managed more extensively.

Herd structure and genetic composition

Dual-purpose farms had larger herds composed mainly by crossbreds between Bos indicus (i.e., Brahman) and Bos taurus of European breeds (i.e., Holstein), whereas specialized dairies in high-altitude regions had smaller herds of either purebreds (ie., Holstein) or livestock with a very high proportion of European genes (Table 3).

Table 3. Herd structure, genetic make-up, reproduction system utilized, mortality, parturition and culling rates, age and weight of breeding heifers, and mature body weight by milk production system and region of the country in 2000.
Parameter	Production system		Region
Parameter	Dual purpose (n=333)	Specialized (n=212)	Piedmont (n=145)	Caribbean (n=116)	Coffee (n=105)	Antioquia (n=97)	Cundi-boyacense (n=82)

Herd structure (#)
- Milking cows	48.7	36.0	20.9	86.8	33.1	33.9	48.3
- Dry cows	34.2	10.8	14.7	68.3	14.4	10.4	13.3
- Heifers > 2 years	21.9	9.1	10.2	41.5	10.4	9.2	11.4
- Heifers 1-2 years	21.2	9.7	10.7	40.4	9.1	9.4	12.3
- Female calves 0-1 year	24.1	10.0	12.3	44.5	11.4	8.2	15.0
- Male calves 0-1 year	21.2	1.2	10.8	42.2	6.1	1.1	1.0
- Steers 1-2 years	16.2	0.6	6.8	31.0	8.1	0.5	0.4
- Steers > 2 years	13.8	0.3	8.8	18.9	11.0	0.1	0.3
- Bulls	3.3	0.9	1.8	6.4	1.5	1.0	0.8
Total heads	204	78.6	97.1	379	105	73.8	103
Total animal units (UA)	155	66.0	72.6	286	84.5	62.7	85.6
Genetic make-up of adult herd (%) ¹
- Cows 100% Zebu	10.3	0.4	8.6	15.9	3.9	0	0
- Cows 75% Zebu – 25% European	22.0	0.2	16.3	40.4	3.2	0	0
- Cows 50% Zebu – 50% European	37.5	6.1	45.3	33.3	23.6	6.8	2.4
- Cows 25% Zebu – 75% European	21.8	7.3	26.5	10.0	35.8	0.5	0
- Cows 100% European	8.4	86.0	3.4	0.3	33.5	92.7	97.6
Reproduction system (% farms)
- Only natural service	74.5	42.9	79.3	76.7	65.7	47.4	24.4
- Only AI	7.8	38.7	8.3	.9	14.3	35.1	56.1
- Both	17.7	18.4	12.4	22.4	20.0	17.5	19.5
Annual mortality (%)
- Adults	1.8	3.5	1.3	1.9	1.3	3.7	3.2
- Calves	7.4	8.7	6.4	7.7	7.5	12.0	6.3
Breeding heifers
Age (mths)	26.9	22.4	27.7	27.4	24.2	21.2	23.3
Body weight (kg)	317	337	317	303	336	316	364
Mature body weight (kg)	432	477	427	423	444	472	509
Parturition rate (%)	69.2	74.3	71.3	64.1	73.8	72.3	76.3
Culling rate (%)	15.6	13.4	20.4	14.8	14.0	12.8	13.3
¹ More than 90% of cows with Zebu phenotype (Bos indicus) had Brahman genes and more than 95% of cows with European phenotype (Bos taurus) had Holstein genes

Herd management

Compared with dual-purpose farms, specialized dairies used artificial insemination more frequently, the age of breeding heifers was younger, and body weight of both heifers and mature cows was higher. This reflects the genetic composition of the herd and the better nutrition received by the latter group. Most specialized dairies practiced twice-a-day milking without the calf close by, mainly in grazing paddocks (especially in Antioquia and the Cundiboyacense altiplanicie), whereas in dual-purpose farms cows were milked once a day, in corrals, with the calf close by.

The proportion of cows in milk was greater in specialized dairies systems and in the Coffee-growing region, Antioquia and Cundiboyacense altiplanicie than in dual-purpose farms found mostly in the Piedmont and Caribbean region because of longer lactations and higher reproductive rate in specialized dairy systems (Tables 3 and 4). The levels of supplementation with feed concentrates were also 3 times higher (Table 4) in specialized dairy systems compared with dual-purpose farms.

Table 4. Milk production, number of daily milkings, location and milking system utilized, proportion of cows in milk, lactation length, and quantity of feed supplement offered to milking cows by production system and region of the country in 2000
Parameter	Production system		Region
Parameter	Dual purpose (n=333)	Specialized (n=212)	Piedmont (n=145)	Caribbean (n=116)	Coffee (n=105)	Antioquia (n=97)	Cundi-boyacense (n=82)

Milk production (kg)
- ha/yr	1,515	7,605	888	731	4,544	8,045	7,875
- Cow/yr	2,053	4,697	1,904	1,540	3,209	5,064	4,837
- Cow/day	5.63	12.87	5.2	4.2	8.8	13.9	13.3
Number of milkings (% farms)
- Once a day	81.7	8.0	94.5	94.0	30.5	2.0	11.0
- Twice a day	18.3	92.0	5.5	6.0	69.5	98.0	89.0
Milking place (% farms)
- In grazing paddock	3.9	68.4	0.7	0	27.6	80.4	61.0
- Open-air corral	31.5	3.3	14.5	65.5	8.6	0	7.3
- Roofed shed	54.7	13.7	81.4	32.8	39.0	4.1	12.2
- Milking parlor	9.9	14.6	3.4	1.7	24.8	15.5	19.5
Milking system (% farms)
- By hand with calf close by	81.4	7.5	91.7	99.1	27.6	1.0	11.0
- By hand without calf	7.5	67.0	3.5	0	44.8	82.5	42.7
- Mechanical with calf close by	3.6	0.5	4.1	0.9	5.7	0	0
- Mechanical without calf	7.5	25.0	0.7	0	21.9	16.5	46.3
Cows in milk (%)	64.4	77.5	64.2	61.3	71.1	78.4	78.1
Lactation length (days)	261	314	240	267	308	308	308
Amount of feed supplements offered to milking cows (g DM / cow / day)
Salt and minerals	133	120	142	141	100	126	117
Molasses	102	163	243	77	101	125	105
Comercial concentrate	520	2,549	125	54	1,816	3,174	2,073
Rice millings	21	16	18	32	0	0	0
Wheat millings	72	25	27	112	0	0	174
Other supplements¹	112	147	35	264	163	185	115
Total	960	3,020	590	680	2,180	3,610	2,584
¹ Cottonseed and/or soybean cake, cottonseed and/or oilpalm hulls, and corn.

Production parameters

Milk production in specialized systems was almost 5 times higher per hectare and 2.3 times higher per cow than that of dual-purpose systems. Similarly, milk production was higher in Antioquia, followed by the Cundiboyacense altiplanicie, the Coffee-growing region, the Piedmont, and finally the Caribbean region. Adult and calf mortality was higher on specialized dairy farms than in dual-purpose herds although this was compensated by a higher calving rate in specialized dairy herds (Table 3).

Production costs and income

The production cost per kilogram of milk was 14% higher in specialized systems than on dual-purpose farms, including the opportunity cost for family labor, and 21% higher if this opportunity cost was not included. The milk price received in specialized dairies was slightly higher (4%, Table 5). As evidenced further on, the higher milk price obtained in specialized systems in highland areas can be attributed to their being closer to markets and urban centers (more accessible) and possibly to the bonuses given for cold milk, because 27.8% of farms with specialized dairy systems had milk tanks compared with only 7.8% of dual-purpose farms (Holmann et al 2003).

Although differences in productivity among production systems were very pronounced depending on the degree of intensification, in terms of unitary production costs, the dual-purpose systems were highly competitive compared with specialized dairy systems. The latter produced between 3 to 7 times more milk than the former, but when unitary costs were estimated for both systems, these differences were not observed. This was attributed to the fact that the lower productivity of dual-purpose systems was neutralized by a low level of operational expenses.

Production costs were higher in Antioquia and the Coffee-growing regions, followed by the Piedmont and the Cundiboyacense altiplanicie, and were lowest in the Caribbean region. The regions with higher prices received by producers were Antioquia, followed by the Caribbean and Coffee-growing regions, then the Cundiboyacense altiplanicie, and lastly the Piedmont (Table 5).

Table 5. Milk and beef production costs, prices, and income by production system and region of the country in 2000.
Parameter	Production system		Region
Parameter	Dual purpose (n=333)	Specialized (n=212)	Piedmont (n=145)	Caribbean (n=116)	Coffee (n=105)	Antioquia (n=97)	Cundi-boyacense (n=82)

Production costs (US$ / farm / year)
- Feed supplements	9,005	23,639	3,084	8,923	13,493	27,528	27,680
- Permanent hired labor	8,766	7,166	5,125	13,178	7,398	7,133	8,511
- Family labor	1,912	2,074	2,724	1,314	1,246	2,498	1,898
- Temporarily hired labor	392	161	249	554	324	177	160
- Irrigation	3,654	1,688	67	9,174	818	1,192	2,885
- Reproduction	1,316	715	642	2,430	698	680	923
- Animal health	1,144	469	675	1,839	731	408	645
- Fertilization	581	1,538	297	283	1,197	1,942	1,579
- Herbicides	343	47	15	672	361	41	29
Total	26,599	37,498	12,879	38,907	26,265	41,599	44,311
Annual production (kg / farm)
- Milk	92,772	184,547	39,880	125,931	106,432	179,640	256,416
- Beef	15,230	4,475	8,954	27,445	6,389	4,197	5,615
Cost of milk production ($ / kg)
- Including opportunity cost of family labor	0.194	0.221	0.200	0.176	0.222	0.242	0.187
- Excluding opportunity cost of family labor	0.159	0.193	0.142	0.160	0.200	0.213	0.159
Product prices ($ / kg)
- Milk	0.207	0.215	0.189	0.222	0.213	0.228	0.207
- Beef	0.818	1.24	0.771	0.760	0.847	1.056	1.455
Gross income ($ / farm / year)
- Milk	19,204	39,678	7,537	27,957	22,670	40,958	53,078
- Beef	12,464	5,528	6,906	20,849	5,409	4,433	8,168
Total	31,668	45,206	14,443	48,806	28,079	45,391	61,246
Net income ($ / cow / year) ¹
- Including opportunity cost of family labor	45.8	6.84	9.9	90.4	-20.6	31.2	122.0
- Excluding opportunity cost of family labor	113.	104	127	111	37	77	211

In terms of net income, dual-purpose farms obtained the highest annual net income per cow, regardless of whether the opportunity cost for family labor was considered or not. Region-wise, the highest income was obtained in the Cundiboyacense altiplanicie, followed by the Piedmont, Caribbean and Antioquia regions. The Coffee-growing region reported the lowest level of income.

Capital investment and returns

About 88% of the investment in dual-purpose farms was represented in land and livestock compared with 80% in specialized dairies. This difference occurred because 20% of the investment in specialized dairies was in infrastructure and equipment whereas this percentage was lower (12%) in dual-purpose farms (Table 6). The market value of land was usually lower in the Piedmont and Caribbean regions where most farms were dual-purpose and higher in high-altitude regions where farms were specialized dairies because of their proximity to urban centers and better public infrastructure. The market value of livestock was higher in high-altitude regions compared with the Piedmont and Caribbean regions.

Table 6. Capital invested in land, livestock, facilities and equipment by milk production system and region of the country in 2000.
Parameter	Production system		Region
Parameter	Dual purpose (n=333)	Specialized (n=212)	Piedmont (n=145)	Caribbean (n=116)	Coffee (n=105)	Antioquia (n=97)	Cundi-boyacense (n=82)

Average prices
Value of land where farm is located (US$/ha)	3,600	5,873	3,541	1,938	3,923	3,656	6,869
Labor wage rate (US$/day)	4.57	5.01	5.13	3.65	4.83	5.30	4.84
Value of a milking cow (US$)	511	634	479	413	696	686	579
Value of a culled cow (US$)	225	193	237	219	193	179	225
Value of a pregnant heifer (US$)	355	497	339	274	473	445	608
Capital investment (US$ / farm)
Land	351,790	203,864	288,205	472,174	240,474	146,612	296,732
Livestock	65,430	40,847	30,979	105,876	52,365	40,305	52,031
Infrastructure	39,715	26,682	36,549	47,694	35,839	14,267	35,398
Equipment	19,742	33,831	7,196	34,834	16,442	16,340	65,250
Total	476,677	305,224	362,929	660,579	345,120	217,524	449,410
Total capital invested ($ / ha)	3,359	7,786	4,459	2,437	5,631	5,432	10,403
Annual return to capital invested (%)
- Including opportunity cost of family labor	1.63	0.32	0.47	3.29	0.23	0.31	2.06
- Excluding opportunity cost of family labor	2.72	2.83	2.21	3.80	1.05	3.15	4.00

Regarding total investment per unit area, use of capital was more intensive in specialized systems, being 2.3 times higher than in dual-purpose farms. This difference was also observed at the regional level, where the highest investment per unit area was in the Cundiboyacense altiplanicie, followed by Antioquia and the Coffee-growing regions, and lastly the Piedmont and Caribbean regions. For more information on the type of investments on farms, please refer to Holmann et al (2003).

Return to capital investment was higher in dual-purpose systems when the opportunity cost for family labor was considered as compared with specialized dairy systems, but was similar when this variable was not considered. Returns to capital investment varied broadly depending on the region, being higher in the Caribbean region and in the Cundiboyacense altiplanicie, lower in Antioquia and in the Piedmont, and lowest in the Coffee-growing region.

Relationship between productivity, technological change, and profitability

Table 7 presents the variables utilized to link productivity and profitability with technological change (with their respective categories and acronyms used), and the diagram in Figure 1 is a perceptual map of the links between these variables.

Table 7. Variables related to productivity, technological change, and profitability utilized in the multiple correspondence analysis to construct Figure 1
Variable	Category	Code name	Best dimension ²

Production system	Dual purpose	DPURPSE	1
	Specialized dairy	SPEC	1
Región ¹	Piedmont	PIEDMT	1
	Caribbean	CARIBBEAN	1
	Coffee	COFFEE	1
	Antioquia	ANTIOQ	1
	Cundiboyacense	CUNDIB	2
Amount of feed supplements offered to milking cows (kg DM /cow / day)	< 0.5	ASUP-L	1
	0.5 to 2.0	ASUP-M	1
	> 2.0	ASUP-H	1
Proportion of cows in milk (%)	< 60	MHERD-L	1
	60 to 80	MHERD-M	2
	> 80	MHERD-H	1
Proportion of pasture area with improved forages (%)	< 33	IMPFOR-L	2
	33 to 67	IMPFOR-M	2
	> 67	IMPFOR-H	1
Frequency of milking (# daily)	Once	MILKING1	1
	Twice	MILKING2	1
Fertilize pastures	No	NFERPA	1
	Yes	YFERPA	1
Irrigate pastures	No	NIRRIPA	1
	Yes	YIRRIPA	1
Reproduction system utilized	Natural Service	NS	1
	Artificial insemination	AI	1
	Both	BOTH	1
Number of grazing paddocks used by milking cows	< 10	PADDOCK-L	1
	10 to 20	PADDOCK-M	2
	> 20	PADDOCK-H	1
Mature body weight (Kg)	< 400	BW-L	1
	400 to 500	BW-M	2
	> 500	BW-H	1
Experience producing milk (# years)	< 5	XPROD-L	2
	5 to 15	XPROD-M	2
	> 15	XPROD-H	2
Milk price (US$/Kg)	< 0.18	PRICE-L	1
	0.18 to 0.24	PRICE-M	2
	> 0.24	PRICE-H	2
Internal deworming frequency ( # / yr)	< 2	IDEW-L	2
	2 to 3	IDEW-M	2
	> 3	IDEW-H	2
External deworming frequency (# / yr)	< 6	EDEW-L	2
	6 to 12	EDEW-M	2
	> 12	EDEW-H	2
Net income per cow ($/yr)	< 50	INCOME-L	2
	50 to 100	INCOME-M	1
	> 100	INCOME-H	2
Annual return to investment (%)	< 1	RETURN-L	2
	1 to 3	RETURN-M	2
	> 3	RETURN-H	2
Milk production per cow per day (Kg)	< 5	MILK-L	1
	5 to 10	MILK-M	2
	> 10	MILK-H	1
Cost per kilo of milk ($/Kg)	< 0.13	MCOST-L	1
	0.13 to 0.20	MCOST-M	2
	> 0.20	MCOST-H	2
Cost per kilo of beef ($/Kg)	< 0.5	BCOST-L	2
	0.5 to 0.7	BCOST-M	2
	> 0.7	BCOST-H	2
Herd size (mature cows/farm)	< 30	HERD-S	2
	30 to 100	HERD-M	2
	> 100	HERD-L	2
Commercial value of land (US$ / ha)	< 3,000	LAND-L	1
	3,000 to 6,000	LAND-M	1
	> 6,000	LAND-H	1
1 This variable was used in the analysis as supplementary. Thus, its different categories were not used to estimate the dimensions. 2 The best dimension for any particular variable was that in which the weight (coefficient) of the variable was greater than in the other dimension.

Figure 1. Diagram showing the contribution of variables related to productivity, technological change, and profitability that explain both
dimensions (DIM 1 y DIM 2) of the perceptual map. For more information on the meaning of each acronym refer to Table 7.

Variables associated with productivity

The contribution of variables to each of the two dimensions accounted for approximately 25% of total variation. The best dimension for a particular variable is that in which its weight (coefficient) is higher than that of the other dimension. The variables that defined the first dimension were clearly those corresponding to milk productivity. High values in this dimension were associated with farms with high milk yield per cow per day (Milk-H), and low values in this dimension were related to farms presenting low milk yields per cow per day (Milk-L).

Other important variables in this dimension (# 1A) were those related to the different farm management practices, which given their coordinates, explained the differences in levels of productivity. Therefore, farms associated with high levels of productivity were those on which a high level of feed supplements was offered to milking cows (Asup-H), artificial insemination (AI) was used, a specialized dairy production system (Spec) was in place, cows were milked twice a day (Milking2), and pastures were irrigated and fertilized (Yirripa, Yferpa), the mature body weight was high (BW-H), and the proportion of cows in milk and the number of grazing paddocks on the farm for better rotation were both high (Mherd-H, Paddock-H). In addition, the most productive farms had a high market value of land (Land-H).

This situation contrasted with those variables that presented negative values and that were associated with low productivities (dimension # 1B). For example, farms with a dual-purpose production system (Dpurpse) were associated with management practices were pastures were not irrigated nor fertilized (Nirripa, Nferpa)), the reproductive system used was natural service (NS), cows were milked once a day (Milking1), the amount of feed supplements offered to milking cows was low (Asup-L), the proportion of cows in milk and the number of grazing paddocks on the farm was low (Mherd-L, Paddock-L), the mature body weight was low (BW-L), and the price per kilogram milk received was also low (Price-L). This information confirms the well-known fact that high technological levels effectively correlate with high levels of productivity (Aldana 1990).

Farms with low productivity were also associated with a low market value of land (Land-L). In terms of productivity, Antioquia and the Cundiboyacense altiplanicie (Cundib, Antioq) were the most productive, contrasting with the Caribbean and Piedmont regions (Caribbean, Piedmt) that presented the lowest productivities. The Coffee-growing region (Coffee) ranked intermediate in terms of productivity.

Variables associated with profitability

The second dimension was basically defined by income and profitability. High (positive) values of this dimension were determined by a high annual net income per cow (Income-H) and by a high annual return to capital investment (Return-H). Low (negative) values of this dimension were associated with low incomes low returns to capital investment (Income-L, Return-L). According to this dimension (# 2A), the most profitable farms (Return-H, Income-H) were characterized by having large herds (Herd-L), which indicated the existence of economies of scale. In addition, profitable farms were also associated to farmers with many years of experience in producing milk (Xprod-H).

The most profitable farms also seemed to be associated with a low frequency of de-worming milking cows against external parasites (Edew-L). This could be attributed to the low prevalence of ticks in high-altitude areas, where most specialized dairy systems are found, thus explaining the lower frequency of treatment.

In Dimension # 2B, those farms with low incomes per cow (Income-L) and low returns to capital invested (Return-L) were associated with farms having small herds (Herd-L), with a low proportion of pasture area with improved forages (Impfor-L), and high production costs per kilogram milk and beef (Mcost-H, Bcost-H). In addition, low profit farms were related to farmers with little experience in milk production (Xprod-L) and with high frequency of treating milking cows against both internal and external parasites (Idew-H, Edew-H).

In terms of profitability among regions, the second dimension shows that the Cundiboyacense altiplanicie (Cundib) was the most profitable region, followed by the Caribbean, Antioquia, and Piedmont (Caribbean, Antioq, Piedmt) with average profitability, and lastly the Coffee-growing region (Coffee) with the lowest income.

The reasons why the Coffee-growing region was least profitable could be the following:

Many former coffee growers shifted to dairy activities during the 1990s because of the prevailing coffee crisis, and are still in the "learning curve". As a result, production costs are still high. The market value of land in this region is as high as in Antioquia (Table 6). However, because of the altitude where most farms in the coffee-growing region are located (i.e., <1,500 m), the most common grass is star grass (Cynodon nlemfuensis), which has a lower nutritional quality than kikuyo grass (Pennisetum clandestinum). Although kikuyo performs well in Antioquia and the Cundiboyacense altiplanicie, it only shows good adaptation in very high-altitude areas within the Coffee-growing region. This difference in the nutritional quality of star grass vs kikuyo leads to a lower productivity per cow in the Coffee-growing region (8.8 kg milk/cow per day) compared with the Antioquia and the Cundiboyacense altiplanicie (13.9 and 13.3 kg, respectively, Table 4) and, as a result, lower return to capital investment.
Because many Coffee growers shifted to dairy during the 1990s, they purchased cows of different genetic composition from various regions of the country. The milking herd of farms surveyed in this region consisted of 30% cows with ½ Zebu genes (i.e., Brahman) and ½ European genes (i.e., Holstein). About 36% of cows were ¾ Zebu - ¼ European, and the remaining 33% of cows were European purebreds. This situation contrasts with the genetic composition of cows in regions with similar climates such as Antioquia and the Cundiboyacense altiplanicie, where more than 95% of cows were Holstein purebreds. Thus, another reason why milk production/cow was lower in the Coffee-growing region might have been the lower genetic potential of the milking herd.

Effect of technological change on productivity, profitability, and competitiveness

The general linear model technique was used to estimate the effect of each one of the technologies evaluated in the multiple correspondence analysis on productivity (expressed as milk production per hectare per year), profitability (expressed as net income per cow per year), and competitiveness (expressed as production cost per kilogram of milk). The variability observed in productivity, profitability, and competitiveness was accordingly as a function of technological change (Tables 8-10) for the different regions of the country.

Adoption of improved pastures

Regardless of the region or production systems used, the adoption of improved pastures increased productivity, profitability, and competitiveness. When farms with a high proportion of pasture area with improved forages were compared with those with a low proportion, an increase was observed in productivity and in net income in the Piedmont (147% and 31%), in the Caribbean (126% and 350%), in the Coffee-growing region (309% and 69%), and in Antioquia (235% and 181%), while the production cost of milk decreased by 13% in the Piedmont, 15% in the Caribbean, 9% in the Coffee-growing region, and 8% in Antioquia. In the Cundiboyacense altiplanicie, milk productivity increased by 92% and income by 570%, while unitary milk production cost decreased by 41% (Tables 8 to 10).

Table 8. Comparison of milk productivity per hectare between different technology levels by region of the country in 2000.
Technological change¹	Region
	Piedmont	Caribbean	Coffee	Antioquia	Cundiboyacense
	(kg milk / ha / yr)

Production system
Dual purpose	861 b	731	4.138 a	5.237 b	5.686 b
Specialized dairy	1.827 a	ND	5.290 a	8.867 a	8.677 a
Amount of feed supplements offered to milking cows (kg DM / cow / d)
< 0.5	840 a	583 b	1.162 c	6.080 a	3.445 c
0.5 to 2.0	961 a	1.118 ab	3.722 b	6.037 a	7.221 b
> 2.0	1.307 a	1.235 a	7.967 a	8.471 a	11.774 a
Proportion of cows in milk (%)
< 60	592 c	569 b	1.782 c	7.414 a	6.000 a
60 to 80	983 a	768 b	4.319 b	8.503 a	6.547 a
> 80	1.487 a	1.420 a	7.838 a	8.930 a	9.269 a
Proportion of pasture area under improved forages (%)
< 33	471 a	368 a	1.320 a	2.961 a	6.247 a
33 to 67	905 a	678 ab	2.280 ab	6.514 ab	7.250 b
> 67	1161 b	832 b	5.395 b	9.924 b	11.998 b
Frequency of milking (#/d)
Once	844 b	708 a	984 b	4.438 a	4.378 b
Twice	1.649 a	732 a	6.104 a	8.120 a	8.306 a
Fertilize pastures
No	794 b	663 a	1.029 b	2.082 b	4.013 b
Yes	1.137 a	838 a	5.529 a	8.438 a	8.885 a
Irrigate pastures
No	888 a	707 a	4.052 b	7.351 b	5.666 b
Yes	934 a	806 a	7.278 a	10.716 a	9.783 a
Reproduction management
Natural Service	867 a	699 a	3.171 b	5.495 b	3.554 b
Artificial insemination	1.127 a	727 a	9.773 a	11.499 a	9.647 a
Both	864 a	745 a	5.317 b	8.036 ab	8.181 a
Number of grazing paddocks used by milking cows
< 10	504 a	668 a	4.442 a	7.742 a	7.012 b
10 to 20	985 ab	619 a	4.464 a	7.903 a	7.868 a
> 20	1.212 b	842 a	5.027 a	8.672 a	9.687 ab
Experience producing milk (years)
< 5	651 b	608 a	7.093 a	6.441 a	7.282 a
5 to 15	857 ab	738 a	3.243 b	7.889 a	8.300 a
>15	1.059 a	744 a	3.762 b	8.336 a	7.796 a
Internal deworming frequency (# / yr)
< 2	713 b	714 a	4.030 b	7.120 a	8.074 a
2 to 3	1.180 a	1.014 a	3.860 b	12.545 a	6.521 a
> 3	1.170 a	496 a	7.652 a	10.454 a	8.830 a
External deworming frequency (# / yr)
< 6	830 a	753 a	5.504 a	8.796 a	8.372 a
6 to 12	1.015 a	734 a	3.479 a	9.148 a	7.702 a
> 12	764 a	651 a	4.767 a	6.430 a	6.880 a
Herd size (# cows / farm)
< 30	997 a	434 a	5.626 a	7.037 a	6.195 b
30 to 100	775 ab	839 a	3.918 ab	9.496 a	8.102 b
> 100	207 b	675 a	2.511 b	8.320 a	10.871 a
< 3,000	633 b	707 b	2.401 b	5.295 b	2.330 c
3,000 to 6,000	1.071	703 b	4.344 b	7.781 b	6.484 b
> 6,000	1.153 a	1.634 a	9.095 a	14.901 a	10.767 a
¹ Means with the same letter do not differ signifficantly (p=0.05) according to Tukey’s multiple comparison test (Steel y Torrie 1980)

Rotational management of pastures

Another technological change associated with adoption of improved pastures was the investment in grazing paddocks to increase the efficiency in the quantity and quality of forage produced through rotational management. Thus, farms with a high number of grazing paddocks for the milking herd were 140% more productive in the Piedmont, 13% more productive in the Coffee-growing region, 12% more productive in Antioquia, and 38% more productive in the Cundiboyacense altiplanicie, as compared with farms with few grazing paddocks. Likewise they were 54% more profitable in the Piedmont, 133% in the Coffee-growing region, 100% in Antioquia, and 61% in the Cundiboyacense altiplanicie (Tables 8 to 10). On these farms, milk production costs were also 23% lower in the Piedmont; 22% lower in the Coffee-growing region; 18% lower in Antioquia, and 27% lower in the Cundiboyacense altiplanicie. In the Caribbean region, this technological change increased productivity by 26% but there were no differences in net income. Unitary production cost increased by 11%, indicating that existing grasses may not be the most adapted to this region and, as a result, this investment did not improve the nutritional quality and/or quantity as occurred in other regions.

Table 9. Comparison of net income per cow per year between different technology levels by region of the country in 2000.
Technological change¹	Region
	Piedmont	Caribbean	Coffee	Antioquia	Cundiboyacense
	(US$ / cow / yr)

Production system
Dual purpose	135 a	111	11 b	53 b	145 b
Specialized dairy	-140 b	ND	86 a	158 a	236 a
Amount of feed supplements offered to milking cows (kg DM / cow / d)
< 0.5	164 a	118 a	117 a	-457 c	163 a
0.5 to 2.0	32 b	102 a	58 a	300 a	263 b
> 2.0	-49 c	47 b	-46 b	39 b	238 c
Proportion of cows in milk (%)
< 60	118 a	72 c	-39 a	-27 b	212 a
60 to 80	123 a	126 b	43 b	83 a	177 a
> 80	178 a	234 a	67 b	86 a	242 a
Proportion of pasture area under improved forages (%)
< 33	127 a	27 b	29 a	48 a	-358 a
33 to 67	118 a	116 a	37 a	62 a	203 b
> 67	167 a	121 a	49 a	135b	316 c
Frequency of milking (#/d)
Once	126 a	156 a	17 a	-68 b	173 a
Twice	157 a	108 a	83 b	80 a	216 a
Fertilize pastures
No	148 a	140 a	39 a	130 a	266 a
Yes	72 b	65 b	36 a	74 b	197 b
Irrigate pastures
No	131 a	121 a	42 a	119 a	189 a
Yes	-107 b	79 b	7 b	-85 b	231 a
Reproduction management
Natural Service	127 a	128 a	53 a	68 a	177 a
Artificial insemination	131 a	51 b	1 b	131 a	213 a
Both	123 a	195 a	16 b	62 a	226 a
Number of grazing paddocks used by milking cows
< 10	126 a	102 a	-75 b	-4 c	158 b
10 to 20	148 a	139 a	60 a	112 a	180 b
> 20	194 b	83 a	58 a	102 b	255 a
Experience producing milk (years)
< 5	92 a	88 b	-42 b	66 a	144 a
5 to 15	138 b	73 b	50 a	73 a	244 b
>15	162 b	136 a	78 a	91 a	236 b
Internal deworming frequency (# / yr)
< 2	151 a	112 b	55 a	302 a	399 a
2 to 3	111 a	179 a	30 a	59 b	173 b
> 3	66 b	-14 c	-27 b	53 b	210 b
External deworming frequency (# / yr)
< 6	249 a	139 a	40 b	140 a	201 a
6 to 12	87 b	59 b	76 a	23 b	223 a
> 12	135 b	98 b	-2 c	59 b	222 a
Herd size (# cows / farm)
< 30	101 b	48 b	-10 b	-11 b	88 c
30 to 100	168 a	123 a	57 a	166 a	234 b
> 100	227 c	109 a	87 a	217 a	422 a
Commercial land value (US$ / ha)
< 3,000	101 b	48 b	-10 b	-11 b	88 c
3,000 to 6,000	145 b	114 a	37 a	114 a	143 b
> 6,000	182 a	126 a	43 a	102 b	246 a
¹ Means with the same letter do not differ (p=0.05) according to Tukey’s multiple comparison test (Steel y Torrie 1980)

These results suggest that the adoption of improved pastures with an appropriate paddock rotation system was a low-risk investment to increase productivity, profitability, and competitiveness. Furthermore, it was the basis of technological change that generated synergisms and on which other management practices, such as feed supplementation, were key changes in further increasing productivity and competitiveness.

Use of fertilizer

The use of this technology increased farm productivity but reduced profits. Unitary milk production cost increased in all regions. Farms that fertilized their pastures increased milk productivity by 43% in the Piedmont but net income was reduced by 51% and unitary costs increased by 21%. In the Caribbean region, productivity increased by 26% but income decreased 54% and unitary costs rose 47%. In the Coffee-growing region, productivity increased 437% but income decreased 83% and unitary costs rose 5%. In Antioquia, production increased by 305%, income decreased by 43%, and unitary costs increased by 9%. In the Cundiboyacense altiplanicie, productivity increased 121%, income decreased 26%, and unitary costs went up 19% (Tables 8-10).

The above data indicate that fertilization increases marginal costs more than marginal income, suggesting that livestock producers who used this technology have not learned to manage it successfully and may be applying more N₂ than needed.

Table 10. Comparison of the cost of producing milk between different technology levels by region of the country in 2000.
Technological change¹	Region
	Piedmont	Caribbean	Coffee	Antioquia	Cundiboyacense
	(US$ / kg milk)

Production system
Dual purpose	0.20 b	0.17	0.23 a	0.21 b	0.21 a
Specialized dairy	0.29 a	ND	0.21 a	0.25 a	0.17 a
Amount of feed supplements offered to milking cows (kg DM / cow / d)
< 0.5	0.19 a	0.16 a	0.19 a	0.26 a	0.17 a
0.5 to 2.0	0.23 ab	0.21 b	0.22 ab	0.17 a	0.17 a
> 2.0	0.26 b	0.24 b	0.25 b	0.31 b	0.20 a
Proportion of cows in milk (%)
< 60	0.22 a	0.19 a	0.24 a	0.24 a	0.20 a
60 to 80	0.20 a	0.16 a	0.23 a	0.24 a	0.19 a
> 80	0.20 a	0.15 a	0.21 a	0.24 a	0.17 a
Proportion of pasture area under improved forages (%)
< 33	0.23 a	0.20 a	0.23 a	0.25 a	0.29 a
33 to 67	0.20 b	0.17 b	0.21 a	0.24 a	0.19 b
> 67	0.20 b	0.17 b	0.21 a	0.23 a	0.17 b
Frequency of milking (#/d)
Once	0.20 a	0.17 a	0.23 a	0.33 a	0.21 a
Twice	0.17 a	0.22 a	0.20 a	0.24 b	0.18 a
Fertilize pastures
No	0.19 b	0.15 b	0.21 a	0.22 a	0.16 a
Yes	0.23 a	0.22 a	0.22 a	0.24 a	0.19 a
Irrigate pastures
No	0.20 a	0.16 b	0.22 a	0.23 b	0.17 a
Yes	0.29 a	0.22 a	0.23 a	0.29 a	0.20 a
Reproduction management
Natural Service	0.20 a	0.16 a	0.21 a	0.24 a	0.19 a
Artificial insemination	0.20 a	0.19 a	0.24 a	0.24 a	0.19 a
Both	0.18 a	0.21 a	0.23 a	0.24 a	0.17 a
Number of grazing paddocks used by milking cows
< 10	0.22 a	0.18 a	0.27 a	0.27 a	0.22 a
10 to 20	0.20 a	0.16 a	0.21 b	0.24 ab	0.18 b
> 20	0.17 b	0.20 b	0.21 b	0.22 b	0.16 b
Experience producing milk (years)
< 5	0.21 a	0.18 a	0.25 a	0.24 a	0.18 a
5 to 15	0.20 a	0.19 a	0.21 ab	0.24 a	0.18 a
>15	0.19 a	0.16 a	0.21 b	0.24 a	0.18 a
Internal deworming frequency (# / yr)
< 2	0.18 a	0.15 a	0.22 a	0.19 a	0.14 a
2 to 3	0.22 b	0.17 a	0.22 a	0.24 b	0.17 a
> 3	0.25 b	0.23 b	0.24 a	0.25 b	0.19 a
External deworming frequency (# / yr)
< 6	0.18 a	0.16 a	0.21 a	0.21 a	0.17 a
6 to 12	0.19 a	0.16 a	0.23 a	0.26 b	0.17 a
> 12	0.21 a	0.21 b	0.23 a	0.26 b	0.19 a
Herd size (# cows / farm)
< 30	0.23 a	0.29 a	0.25 a	0.27 a	0.21 a
30 to 100	0.19 a	0.17 b	0.20 b	0.22 b	0.17 ab
> 100	0.16 a	0.17 b	0.20 b	0.19 b	0.14 b
Commercial land value (US$ / ha)
< 3,000	0.18 b	0.15 a	0.22 a	0.23 a	0.18 a
3,000 to 6,000	0.19 b	0.17 a	0.22 a	0.25 a	0.19 a
>6,000	0.24 a	0.27 b	0.23 a	0.25 a	0.19 a
¹ Means with the same letter do not differ (p=0.05) according to Tukey’s multiple comparison test (Steel y Torrie 1980)

Use of irrigation

Just as in the case of fertilization, irrigation technology increased both milk productivity and unitary production costs in all regions, suggesting that additional research is required to determine the level of water needed to produce high-quality forage economically. In addition, irrigation reduced net income in all regions, except for the Cundiboyacense altiplanicie, where the use of this technology increased income by 22% although production costs increased by 17%.

Use of feed supplementation

In all regions, milk productivity and unitary production costs increased on those farms that offered feed supplements to milking cows. However, the best economic yield in terms of net income occurred on farms that provided the milking herd with low levels of supplementation (i.e., < 0.5 kg DM/cow per day) in the Piedmont, Caribbean, and Coffee-growing regions, and intermediate levels of supplementation (i.e., between 0.5 and 2 kg DM/cow per day) in Antioquia and the Cundiboyacense altiplanicie (Tables 8 to 10). Feed supplementation at higher levels increased milk productivity in all regions, but at the expense of reducing net income/cow per year.

During 2000, the milk price received by producers was US$0.21/kg and the average cost of commercial concentrate was US$0.22/kg. Thus, the milk-to-concentrate price ratio was lower than 1 (0.96:1), indicating that the price of commercial concentrate in Colombia was very expensive compared with that of milk. Alternatively, it could be presumed that the reduction of tariffs on imported grains would make the use of supplements more viable and Colombia could rapidly increase its milk productivity.

Production system

In lowland regions (Piedmont and Caribbean), the dual-purpose production system generated higher net income, whereas in highland regions (Coffee, Antioquia and Cundiboyacense altiplanicie), most profits were obtained with the specialized dairy system. In addition, the dual-purpose system produced milk at the lowest unitary cost in all regions, except the Coffee-growing region, suggesting that Colombia should have different research and technology transfer strategies to tap more efficiently the comparative advantages of each region.

Frequency of milking

The twice-a-day milking management practice proved to be 83% to 520% more productive and 25% to 148% more profitable, and unitary production cost decreased between 15% and 27% in all study areas, except for the Caribbean region where twice-a-day milking did not increase productivity but instead adversely affected net income and increased production costs. One reason why this practice did not trigger increases in productivity in the Caribbean region could be the low genetic potential of animals, because more than 56% of the herd were cows with ≥ 75% Zebu genes (i.e., Brahman). In addition, this management practice require farms to have electricity and own a milk cooling system to store the afternoon milk until the following day. As a result, it is not appropriate for livestock producers with limited resources, located in marginal livestock areas and/or owning herds of low genetic potential.

Percentage of cows in milk

This indicator is the result of the interaction between reproductive efficiency (ie., calving interval) and breed (ie., lactation length). Farms with a high proportion of cows in milk (i.e., > 80%) were 20% to 340% more productive and 15% to 225% more profitable in all regions than those farms with a low proportion of cows in milk (i.e., < 60%), meaning that the fixed costs of dry cows were shared by a larger number of cows in production. In addition, milk production cost was 9% to 21% lower with increasing proportion of cows in milk, except for Antioquia where no differences were observed.

Frequency of de-worming

The effect of the frequency of de-worming milking cows against both internal and external parasites did not have a significant impact on productivity but did affect profitability and competitiveness. Farms that seldom de-wormed cows for internal parasites obtained between 77% to 128% more income per cow and unitary production cost was 8% to 35% lower, depending on the region, than those farms that frequently de-wormed cows. Similarly, farms that seldom de-wormed cows for external parasites obtained between 42% to 137% more income, and production costs were 9% to 24% lower, except for the Cundiboyacense altiplanicie where no significant differences in income were observed.

Reproductive management

Farms that used artificial insemination were more productive that those using natural service in highland regions (Coffee, Antioquia, and Cundiboyacense altiplanicie). No differences were observed in production costs or levels of profitability, except in the Coffee-growing and Caribbeanregions where farms using only natural service were more profitable (Tables 8 -10). This suggests that the use of artificial insemination, whether as sole reproduction strategy or as complement to natural service to produce breeding animals of superior genetic quality on the farm, may have helped improve the genetic potential of herds and, as a result, farm productivity.

Herd size

Farms with larger herds were always more profitable and more competitive than farms with small herds across all regions and production systems. Farms with larger herds were also more productive in the Cundiboyacense altiplanicie; in contrast, farms with small herds were more productive in the Caribbean and Coffee-growing regions. As evidenced later on in this document, the scale of operation was the factor that most affected profitability and competitiveness.

Farmer experience in producing milk

Farms where the producer had lots of experience in dairying (i.e. >15 years) obtained higher income in all regions, ranging from 38% in Antioquia to 120% in the Coffee-growing region, compared with farms where producers had little experience (i.e., < 5 years). The amount of experience in dairying only had a positive effect on milk productivity in the Coffee and Piedmont regions. Years of experience in dairying only reduced unitary production costs in the Coffee-growing region. This fact suggests that practical training events could help increase the income of those producers with little experience, especially small-scale producers, because it is more difficult for them to access information on new technologies and/or successful management practices.

Market value of land

Farms located in areas where the market value of land was high (> $6,000/hectare) were always more productive than those with intermediate or low market values, but less profitable in all regions. There were no significant differences in production costs, except in the Piedmont and Caribbean regions, where farms with high market value also produced milk at a higher cost. The most profitable category contained farms with an intermediate market value of land (i.e., from $3,000 to $6,000/hectare). Thus, this range might be used as reference when deciding whether intensification of dairying would prove profitable. Above this range, intensification is not justifiable from the financial viewpoint. Furthermore, intensification of livestock production on lands with a high market value should no longer be based on grazing, but on other systems such as confinement because the proximity to urban centers boosts the cost of land.

Based on the results of this analysis, it was concluded that the performance of diverse technologies affect productivity and profitability in different ways, depending on the region. It also showed that, at a regional level, technologies that increase milk productivity are not necessarily profitable. In general, these results show a trend similar to those associations observed in the multiple correspondence analysis presented in the perceptual map (Figure 1), except that the difference in productivity and profitability among levels of each technology are quantified for each region in the latter analysis.

Effect of breed on profitability, competitiveness, and productivity

The variables that determined the profitability, productivity, and competitiveness of three cattle breed groups, categorized according to the proportion of European genes (i.e., Holstein), were identified to define genetic improvement strategies for different production systems and regions of Colombia. The averages of the seven most relevant variables are presented in Table 11. Farms were grouped as follows:

Group A Farms where the milking herd had on average between 75%-100% European genes. The overall average of this group was 98% European genes and 2% Zebu genes.
Group B Farms where the milking herd had on average between 50%-74% European genes. The overall average of this group was 55% European genes and 45% Zebu genes.
Group C Farms where the milking herd had on average less than 50% European genes. The overall average of this group was 24% European genes and 76% Zebu genes.

Table 11. Effect of breed group on production costs of milk and beef, net income, milk and beef productivity, and return to capital investment by production system and region of the country in 2000.
Production system or region	Cost of production ⁴		Milk production		Beef production (kg/ha/yr)	Net income ($/cow/yr)	Annual return to investment (%)
Production system or region	Milk ($/kg)	Beef ($/kg)	Cow (kg/d)	Hectare (kg/yr)	Beef production (kg/ha/yr)	Net income ($/cow/yr)	Annual return to investment (%)

Dual purpose
- Group A ¹ (n=104)	0.215 a	0.77 a	9.5 a	4,421 a	195 a	- 10 a	- 0.09 a
- Group B ² (n=124)	0.195 ab	0.72 ab	5.8 b	1,313 b	156 ab	31 ab	1.49 a
- Group C ³ (n=141)	0.188 b	0.70 b	4.2 c	747 c	138 b	59 b	2.01 b
Specialized dairy
- Group A (n=160)	0.217 a	0.64 ab	13.4 a	8,145 a	191 a	39 a	1.01 a
- Group B (n=18)	0.236 a	0.72 a	9.8 a	5,587 a	183 a	- 71 a	- 1.28 a
- Group C (n=0)	NA	NA	NA	NA	NA	NA	NA
Piedmont region
- Group A (n=27)	0.196 a	0.77 a	6.0 a	1,020 a	114 a	14 a	0.50 a
- Group B (n=66)	0.198 a	0.77 a	5.4 a	913 a	160 a	13 a	0.46 a
- Group C (n=52)	0.205 a	0.83 a	4.5 b	789 a	160 a	3 a	0.46 a
Caribbean region
- Group A (n=0)	NA	NA	NA	NA	NA	NA	NA
- Group B (n=33)	0.169 a	0.59 a	4.7 ab	786 a	126 a	89 a	3.88 a
- Group C (n=83)	0.180 a	0.62 a	4.0 b	716 a	121 a	84 a	3.03 a
Coffee region
- Group A (n=60)	0.231 a	0.76 a	9.5 a	5,326 a	212 a	-56 b	- 0.27 b
- Group B (n=37)	0.220 a	0.73 a	8.1 a	3,859 ab	181 a	- 10 ab	0.53 ab
- Group C (n=8)	0.165 b	0.56 b	6.3 b	1,842 b	156 a	196 a	2.54 a
Antioquia region
- Group A (n=97)	0.200	0.71	14.0	8,260	200	36	0.48
- Group B (n=0)	NA	NA	NA	NA	NA	NA	NA
- Group C (n=0)	NA	NA	NA	NA	NA	NA	NA
Cundinoboyacense altiplanicie region
- Group A (n=80)	0.183 b	0.60b	13.3 a	7,974 a	197 a	133 a	2.31 a
- Group B (n=2)	0.311 a	1.07 a	10.4 a	3,934 a	85 a	-350 b	-8.15 b
- Group C (n=0)	NA	NA	NA	NA	NA	NA	NA
¹ Farms where the milking herd had on average between 75%-100% European genes. The overall average of this group was 98% European genes and 2% Zebu genes ² Farms where the milking herd had on average between 50%-74% European genes. The overall average of this group was 55% European genes and 45% Zebu genes ³ Farms where the milking herd had on average less than 50% European genes. The overall average of this group was 24% European genes and 76% Zebu genes ⁴ Means with the same letter do not differ signifficantly (p=0.05) according to Tukey’s multiple comparison test (Steel y Torrie 1980)

Effect of breed on production system

Farms with dual-purpose production systems were more profitable (i.e., higher net income and improved profitability) and more competitive (i.e., lower production costs of milk and beef), but less productive when the proportion of European genes was low (group C) and intermediate (group B). In addition, farms with specialized dairy systems were slightly more profitable, competitive, and productive when high levels of European genes (group A) were found, although these differences were not significant compared with the intermediate group (group B).

Effect of breed on the region

In lowland regions (ie., Piedmont and Caribbean), no significant differences between breed groups were observed for any of the seven variables analyzed. In the Coffee-growing region, the use of cows of the intermediate or low groups (groups B and C) proved more profitable and competitive but less productive than the use of cows of the high group (group A). In Antioquia and the Cundiboyacense altiplanicie, the only breed group found on farms surveyed belonged to the high level (group A). Thus, it was not possible to compare results.

However, the results reported herein refute those of McDowell et al (1996), who stated that: "for milk farms to be profitable, they must produce over 4,400 kg/lactation (> 14.4 kg/cow per day) using purebreds or cows with a high percentage of Bos taurus genes". The case of Colombia has demonstrated that the profitability and competitiveness of milk production systems are much more complex and depend not only on technological change but also on the interaction of technology change with the effect of the environment and of management practices.

Economies of scale

Based on multivariate principal components analysis, farms were grouped within each production system and region to identify those changes in productivity associated with increased farm competitiveness and profitability. Six groups of farms were formed, and these accounted for 80% of total variability using Ward's minimum variance method (1963).

Results show that improved farm competitiveness and profitability was directly related to herd size (Table 12). In other words, unitary production costs of milk and beef decreased, net income per cow increased, and return on capital investment improved significantly with increasing herd size, suggesting the existence of economies of scale. The logic seems to indicate that fixed costs are distributed among a larger number of cows, meaning that unitary production costs are reduced with increasing herd size.

Table 12. Multivariate analysis of farms grouped by herd size which contains the cost of cost of milk production, net income, annual return on investment, and productivity of milk and beef by production system and region of the country in 2000.
Clusters of farms grouped by production system and region	R²	Number of farms per cluster	Herd size (# cows)	Cost of milk production ($/kg)	Net income ($/cow/yr)	Annual return to investment (%)	Productivity of milk (kg/ha/yr)	Productivity of beef (kg/ha/yr)	Retribution to family labor ($/day)

Dual purpose	82.3
1		108	20	0.24	- 66	-0.7	894	140	3.5
2		21	35	0.21	58	1.3	2193	247	15.3
3		136	83	0.16	106	2.8	734	134	27.7
4		17	78	0.20	87	2.6	5472	173	36.4
5		13	337	0.13	164	6.0	636	140	135.1
6		5	730	0.13	82	6.1	226	78	162.8
Specialized dairy	76.6
1		54	17	0.25	- 152	-2.9	9100	360	1.7
2		52	24	0.26	- 163	-3.7	2976	128	1.1
3		35	37	0.20	180	4.6	15760	262	30.3
4		24	62	0.18	227	6.0	7970	130	40.7
5		31	105	0.20	57	1.7	3090	79	19.9
6		13	159	0.16	413	6.2	14358	245	183.7
Piedmont	77.4
1		59	19	0.19	12	0.2	1099	178	8.9
2		30	23	0.30	-184	-1.9	742	75	1.4
3		9	45	0.08	463	8.5	662	392	70.6
4		29	56	0.16	61	1.0	728	109	17.9
5		5	56	0.28	-182	-1.7	1463	84	2.7
6		8	108	0.17	23	0.3	326	109	12.5
Caribbean	84.8
1		9	48	0.32	-130	-1.8	377	56	1.9
2		27	73	0.19	25	0.4	750	112	8.1
3		35	111	0.14	140	4.8	1028	151	41.8
4		17	175	0.11	253	8.8	758	152	116.6
5		10	528	0.15	84	2.9	410	116	127.8
6		1	926	0.10	280	9.0	108	80	710.4
Coffee	79.1
1		13	8	0.30	-341	-3.6	9300	378	1.1
2		28	19	0.24	-55	-0.8	1460	186	1.9
3		18	28	0.24	-70	-0.8	10100	291	2.1
4		13	76	0.15	115	2.8	600	157	26.4
5		29	85	0.19	179	2.8	3800	99	30.2
6		1	265	0.15	210	3.1	6400	114	152.5
Antioquia	81.1
1		11	13	0.29	-361	-9.6	8500	428	1.2
2		14	18	0.27	-195	-4.8	2370	105	2.6
3		12	26	0.25	48	2.7	20200	385	14.5
4		36	34	0.23	21	1.5	6090	153	14.0
5		10	113	0.20	90	1.7	2800	80	35.7
6		10	117	0.20	255	5.6	14600	197	75.2
Cundi-boyacense altiplanicie	82.0
1		10	10	0.25	-178	-4.7	4900	197	4.5
2		14	21	0.22	-86	-0.6	10600	263	4.1
3		18	38	0.19	25	0.5	2100	126	10.8
4		25	72	0.16	278	5.4	9400	183	56.4
5		7	170	0.15	567	7.9	15800	279	260.7
6		1	330	0.15	591	8.1	12,700	41	534.1

This trend was not observed when increased competitiveness was associated with productivity, suggesting that farms with high productivity are not necessarily competitive nor profitable. Therefore, herd (farm) size is the factor that most influences profitability and competitiveness, regardless of the production system, the level of productivity, or the region where farms were located.

This foregoing has significant implications for the rural/livestock sector in Colombia because approximately 70% of milk producers produce less than 100 kg/day (Consejo Nacional Lácteo,1999). Table 12 shows negative income per cow per year in the case of farms with small herds. This means that the returns to family labor are lower than the official minimum wage rate and the persistence of these systems can be attributed to the opportunity cost of family labor or the limited opportunities of employment in rural areas. Table 12 also indicates the daily return to family labor, expressed in US dollars. As can be observed, returns to family labor on small farms were lower than the official minimum wage in force during the year 2000 of US$ 4.75/day.

Likewise, the objective function of smallholders may not necessarily be to maximize income, especially in the case of small farms. Mixed systems, where other crops (i.e., coffee) are also planted in addition to livestock activities, were usually found in the small farms surveyed in this study, a common strategy to reduce risk and maximize family labor use. It is highly probable that these small farms are involved in livestock activity to generate cash flow, as a source of savings, as a way to diversify risk, for food security, or a combination of the above.

Evolution of milk production systems in Colombia

Table 13 shows the parameters of productivity and profitability reported by Aldana (1990) in 1988 and those found by this study during 2000. Aldana (1990) utilized a methodology similar to that used in this study to calculate milk and beef production costs, gross sales, and net income. Thus, we believe this study can be used as reference to analyze the evolution of milk production systems in Colombia during this period. To be able to compare the economic variables of both studies, the figures of Aldana were expressed in Colombian pesos of year 2000 and then in US currency, at the 2000 market exchange rate (US$1 = Col$2,094) (Estadísticas Financieras Internacionales 2002).

Table 13. Evolution of productivity, cost of production, capital investment, profitability, and farm-gate prices of milk and beef in dual-purpose and specialized dairy systems in Colombia between 1988 y 2000.
Parameter	Milk production system
	Dual purpose		Specialized dairy
	1988^a	2000	1988 ^a	2000

Productivity
- Milk production (kg/ha/yr)	453	654	4,132	4,708
- Beef production (kg/ha/yr)	115	107	212	114
- Stocking rate (AU/ha)	1.3	1.5	2.3	2.7
Cost of production
- Milk (US$/kg)	0.19	0.16	0.21	0.19
- Beef (US$/kg)	0.73	0.57	0.98	0.60
- Both (US$/ha)	172	174	1,098	903
Capital investment (US$/ha)
- Land	1,828	2,479	7,120	5,201
- Livestock	688	461	2,868	1,042
- Equipment and infrastructure	117	419	1,126	1,544
- Total	2,632	3,359	11,114	7,786
Income (US$ / ha / yr)
- Gross income	239	223	1,906	1,153
- Net income	67	49	806	250
Farm-gate prices (US$/kg)
- Milk	0.27	0.21	0.37	0.22
- Beef	1.02	0.82	1.71	1.24
Annual return to capital investment (%)	4.2	2.7	6.8	2.8
^a Adapted from Aldana (1990). Values in Colombian pesos of 1988 were inflated to constant pesos of the year 2000 and then expressed in US dollars using the average exchange rate during 2000 of Col$ 2,084 pesos to the dolar (Estadísticas Financieras Internacionales 2002). Productivity parameters were estimated based on the weighted average of the production systems’ categories “improved” and “intensive” calculated by Aldana.

The first major difference is that milk productivity per hectare increased by 44% in dual-purpose systems and by 14% in specialized dairy systems. This increase was partly due to a 15% increase in stocking rate in dual-purpose systems and a 17% increase in specialized dairy systems. This trend was also observed in other regions of Colombia not covered by this study, as was the case of the Colombian Amazon region between 1986 and 1997 (Rivas and Holmann 2000). In addition, beef productivity remained stable in dual-purpose systems but decreased by 46% in specialized dairy systems. Therefore, efforts to increase productivity were exclusively focused on milk production, not on beef.

This increase in productivity reduced, in real terms, the cost of producing milk in dual-purpose systems by 16% and in specialized systems by 10%. Likewise, beef production costs decreased by 22% in dual-purpose systems and by 39% in specialized dairy systems. As a result, Colombia's dairy sector has become more competitive, due partly to a higher stocking rate, but also from an increased investment in infrastructure and equipment (eg., adoption of improved forages, rotational grazing management, forage choppers, irrigation equipment, and facilities). This increase in investment was 258% in dual-purpose systems and 37% in specialized dairy systems.

However, even though milk producers have made technological changes and investments to become more competitive and productive, the net income per hectare during this period decreased by 27% in dual-purpose systems and by 69% in specialized dairy systems. This drastic reduction in income was attributed to the respective decline in real milk and beef prices to the farmer of 22% and 20% in dual-purpose systems and of 41% and 27% in specialized dairy systems.

Figure 2 shows the percentage of the price paid per liter milk by consumers that is retained by milk producers. As can be observed, in 1989 milk producers received 70% of the final price; however, during the 1990s, this percentage was systematically reduced until it reached only 37% in 2001.

Figure 2. Proportion of milk price paid by consumers that is retained by producers in Colombia compared to Costa Rica,
Nicaragua, and Venezuela during the period 1989 to 2001 (Source: CEGA 2002; DANE 2002; Cámara Nacional
de Productores de Leche 2002; CORECA 200; ASOLEP 2000; Gaceta Ganadera 2002)

This dramatic loss occurred because the fall of milk and beef prices to the producer was never translated into a fall of prices to the consumer and, as a result, an important segment of the society (the consumers) did not benefit (Figures 3 and 4). As can be observed, the increases in milk and beef prices to producers were always below the inflation rate, whereas adjustments of prices to the consumer usually surpassed the level of inflation. During this period, the average inflation rate was 20.7% per year, but the increase in milk price to the producer was of the order of 15.4% per year (Figure 3) and that of beef price, 18.2% per year (Figure 4). As a result, the prices of milk and beef received by producers in 2001 were 41.7% and 34.7% lower, in real terms, than those received in 1989 (Aldana 1990). This explains why net income and returns on capital investment decreased significantly when the two studies were compared (Table 13).

Figure 3. Inflation rate and adjustments in the price of milk paid by consumers and received by producers
in Colombia during the period 1989 to 2001 ( Based on the weighted average of farm-gate price
received by producers for one kg of milk and the consumer price of one liter of milk
in plastic bag containing 2% butterfat) (Source: CEGA 2002; DANE 2002)

Figure 4. Inflation rate and adjustments in the price of beef paid by consumers and received
by producers in Colombia during the period 1989 to 2001 (Source: CEGA 2002; DANE 2002)

Livestock economist. International Center for Tropical Agriculture (CIAT) and International Livestock Research Institute (ILRI). F.Holmann@cgiar.org

Agricultural economist, research associate. CIAT. Cali, Colombia. L.Rivas@cgiar.org

Animal nutritionist, Professor, School of veterinary medicine, Universidad Nacional de Colombia. Bogotá. Djcarull@veterinaria.unal.edu.co

Animal scientist, Professor, School of Agricultural Sciences, Universidad de Caldas. Manizales. Brivera@cumanday.ucaldas.edu.co

Animal Scientist, Professor, School of Agricultural Sciences, Universidad Nacional de Colombia. Medellín. Conisilvo@epm.net.co

Veterinary and Animal Scientist, Dean, School of veterinary medicine and Animal Production, Fundación Universitaria San Martin. Barranquilla. Sigupe@latinmail.com

Microbiologist, Professor, School of Animal Science, Universidad de los Llanos, Villavicencio. Telephone (57-0986) 69-8662 and Fax (57-0986) 69-8535.

Evolution of Milk Production Systems in Tropical Latin America and its interrelationship with Markets: An Analysis of the Colombian Case

Federico Holmann, Libardo Rivas, Juan Carulla, Bernardo Rivera, Luis A Giraldo, Silvio Guzman, Manuel Martinez, Anderson Medina and Andrew Farrow

International Center for Tropical Agriculture (CIAT) and
International Livestock Research Institute (ILRI)
F.Holmann@cgiar.org

Abstract

Introduction

The Case of Colombia

Growth and milk production systems

Contribution of the livestock sector to the Colombian economy

Market concentration in hands of supermarkets

Addressing the problem

Objectives

Methodology

Statistical analysis

Results

Farm characteristics

Brief description of regions

Land use and pasture management

Herd structure and genetic composition

Herd management

Production parameters

Production costs and income

Capital investment and returns

Relationship between productivity, technological change, and profitability

Variables associated with productivity

Variables associated with profitability

Effect of technological change on productivity, profitability, and competitiveness

Effect of breed on profitability, competitiveness, and productivity

Economies of scale

Evolution of milk production systems in Colombia

Market concentration

Conclusions

The challenge

Acknowledgements

References

Evolution of Milk Production Systems in Tropical Latin America and its interrelationship with Markets: An Analysis of the Colombian Case

Federico Holmann, Libardo Rivas, Juan Carulla, Bernardo Rivera, Luis A Giraldo, Silvio Guzman, Manuel Martinez, Anderson Medina and Andrew Farrow

International Center for Tropical Agriculture (CIAT) and International Livestock Research Institute (ILRI) F.Holmann@cgiar.org

Abstract

Introduction

The Case of Colombia

Growth and milk production systems

Contribution of the livestock sector to the Colombian economy

Market concentration in hands of supermarkets

Addressing the problem

Objectives

Methodology

Statistical analysis

Results

Farm characteristics

Brief description of regions

Land use and pasture management

Herd structure and genetic composition

Herd management

Production parameters

Production costs and income

Capital investment and returns

Relationship between productivity, technological change, and profitability

Variables associated with productivity

Variables associated with profitability

Effect of technological change on productivity, profitability, and competitiveness

Effect of breed on profitability, competitiveness, and productivity

Economies of scale

Evolution of milk production systems in Colombia

Market concentration

Conclusions

The challenge

Acknowledgements

References

International Center for Tropical Agriculture (CIAT) and
International Livestock Research Institute (ILRI)
F.Holmann@cgiar.org