SciELO - Scientific Electronic Library Online

vol.41 issue2Hedging with futures contracts in the Brazilian soybean complex: BM&F vs. CBOT author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Revista de Economia e Sociologia Rural

Print version ISSN 0103-2003

Rev. Econ. Sociol. Rural vol.41 no.2 Brasília Apr./June 2003 

Risk analysis of investments in-farm milk cooling tanks



Danielle D. Sant´AnnaI; Carlos Arthur B. da SilvaII; Sebastião T. GomesIII

I M.S. in Science and Food Technology
II Full Professor, Universidade Federal de Viçosa. E-mail:
III Full Professor, Universidade Federal de Viçosa




A risk analysis for the installation of milk cooling tanks (250, 500 and 1,000 L) on Brazilian rural properties was conducted in this study. The results showed that all investments had a return higher than the annual 12% minimum rate of attractiveness. There was a direct relationship between tank size and investment profitability and an inverse relation between size and risk. The probability of achieving returns lower than the opportunity cost was highest for the smallest tank (42%). In order to make the investment in small cooling tanks more attractive, the dairy industry incentives offered to farmers for supplying cooled milk could be increased. However, this approach might make investments in bulk milk collection by dairy companies infeasible. Thus, a recommendable strategy for a successful modernization of the Brazilian dairy sector’s inbound logistics would be to promote an increase in the volume of the milk produced per farm.

Key words: milk, bulk milk collection, risk analysis.



1. Introduction

Dairy products are currently one of the main sources of income for a significant share of the Brazilian population. In 1990, Brazil’s annual milk production was 14.84 billion liters. The estimated production for 2001 was 20.825 billion liters (CNA, 2000).

Changes in the country’s economic environment over the last decade (deregulation, price liberalization, market openings, MERCOSUR, economic stability etc.) have made it crucial that its dairy products sector modernize and become more competitive. Operating costs need to be reduced and raw material quality needs to be improved (Jank & Galan, 1998). To attain these goals, bulk milk collection systems are being put into operation. Using these systems, milk is cooled milk at producing farms and then collected on alternate days by temperature controlled tank trucks.

One of the probable consequences of the bulk milk collecting system is that it excludes small dairy farmers from the dairy business, since their output is considered insufficient to justify, or afford, the purchase of cooling tanks. As a contribution to the analysis of this issue, the present work assesses the risks involved in the installation and operation of three sizes of cooling tanks: 250 L, 500 L and 1,000 L. The financial analysis of these investments was performed by Sant’anna et al (2000).


2. Materials and methods

Data collection and identification of the equipment needed for on-farm milk storage and cooling were performed through direct contact with suppliers and producers who had already begun their milk collection system’s modernization. The choice of tank sizes considered in this study was based on results of a comprehensive diagnostic of the dairy chain published by SEBRAE–MG (1996). That study concluded that dairy activity in Minas Gerais was best characterized in terms of the number of small producing farms (up to 50 l/day) and in terms of yield and volume of the middle sized producing farms (from 51 to 250 L/day).

In order to carry out a financial analysis, cash flows were built for the investment expense and operation cost for each tank size. The farmer’s investments were for tank purchase and the construction of a structure to house the tank. The cost components were consumption of electric energy, consumption of detergents for tank sanitation, maintenance costs, and investment depreciation. The benefits offered by dairies to farmers for cooling milk on the farm came in the form of two incentives: an additional 5% added to the price paid for warm milk for supplying cooled milk, and a reduction of 50% in the typical charge for using the traditional milk collection and transportation system for using the much less costly bulk transport system.

Based on the results from financial analysis, a sensitivity analysis of the investment was carried out, in which some parameters considered in the cash-flow estimation were varied. For each variation, a new internal rate of return (IRR) was calculated. This produced a set of graphs known as "spider plots," which allow the assessment of the degree of uncertainty associated to the cooling tank investments (Eschenbach, 1992). The analysis identified the revenue and expense items that had the most significant impacts on the financial indicators (Sant´anna et al., 2000). The potential risks associated to these items are evaluated in detail in this study.

Since this study is characterized as a case of replacement analysis, whereby product collection in milk cans is replaced by the bulk collection system, only new investments and added values associated with the new system were taken into account (White et al., 1998).

2.1. Risk Analysis

The financial analysis performed by Sant´anna et al (2000) considered that the project would be implemented with perfect control of its variables. However, due to uncertainties in the near future, this approach only approximates reality. In practice, company risks should be investigated, defined, and then controlled by decision-makers (Lopes, 1992).

Risk is defined as the possibility of future loss in predicted return over a certain period of time (Palisade Corporation, 1995; Martins & Assaf NETO, 1992). Current data is imperfect, future data only more so. Every current business decision is associated to a series of hypotheses regarding future events (increase or decrease in raw materials prices, new competition, climatic conditions, social and political events, etc). Even if prediction techniques were improved, they would not eliminate the uncertainty brought by unforeseen change in various factors affecting profitability (OCDE, 1977).

Szekres (1986) states that the decision to take risks depends on the utility function that each individual subjectively attributes to a particular situation. People react to risk with three general types of behavior, risk-avoidance, risk-indifference, or risk-orientation; and the decision-making process is directly related with one of these personal tendencies.

Although sensitivity analysis can point out the need for further study of uncertain variables, by itself it is not sufficient for the analysis of an investment project’s risk. Firstly because sensitivity analysis is only a partial analysis, with only one variable considered at a time, it does not permit the negative effect caused by one variable to be counterbalanced by the positive effect from another. Secondly, the mere indication that a project is or is not sensitive to certain variables, though useful, does not satisfy the decision-maker. It is also important to have an idea of the probability of an adverse situation occurring, as well as this situation’s effect on project results (Noronha, 1987).

@RISK (Palisade Corporation, 1995) software was applied for risk analysis in this study, and Monte Carlo simulation methodology was adopted. This methodology consists of simulating variables previously selected for the cash flow estimation and then calculating financial indicator values (IRR, net present value, etc). After a number of random interactions, a frequency distribution of the financial indicator considered is generated (Neves, 1996).

According to Noronha (1987) the Monte Carlo method simulation calculation sequence proposed by Hertz (1964) consists of 4 phases: a) identifying the probability distribution of each of the relevant cash flow variables; b) randomly selecting a value for each variable, based on its probability distribution; c) calculating the value of the chosen indicator (IRR and/or net present value) for every random selection indicated in item b; d) repeating the process until the chosen indicator’s frequency distribution is adequately confirmed.

Applications of this method’s application for the analysis of agricultural and agroindustrial projects may be found in the works of Lopes (1992), Adib (1996) and Neves (1996).


3. Results and Discussion

The study carried out by Sant´anna et al. (2000), based on estimated cash flows, showed that investment in the 250 liter tank would not be recommendable, since the 8% internal rate of return from that investment would be below the adopted 12% minimum rate of investment attractiveness. Investment in both the 500 and 1,000 liter tanks would be feasible and attractive, since their annual returns of 31% and 47% would be significantly above the opportunity cost of capital. Sensitivity analysis, in turn, revealed that an increase of 25% in the incentive paid to farmers for supplying cooled milk (increasing the 5% incentive to 6.25%) or an increase of 1% in the volume of the milk produced after introducing the bulk milk collection system (increasing production from 112.5 to 114 liters of milk per day) would be sufficient to make the 250 l/tank investments attractive, with a return above 12% per year.

To sum up, the analysis revealed that, although technologically and financially feasible, cooling tank investments are subject to uncertainties the effects of which can jeopardize efforts to modernize the Brazilian milk collection system. These issues are discussed in depth in the following analysis.

3.1. Risk Analysis

Variation of the price paid to the producer for milk and volume of milk produced on the farm were adopted for risk analysis: Sensitivity analysis found that these were the most unstable parameters.

Price paid to the farmer for milk is a risk-dependent item since it varies according to market conditions and milk composition. Most dairy companies offer farmers bonuses and other awards linked to the quality and/or volume of milk supplied.

Volume of milk produced is another risk-dependent item since output depends on several factors, such as climate, costs, and investments. Milk production in Brazil typically increases during the summer (rainy season) and decreases during the winter (dry season). This seasonal change is due to change in available pasture: more pasture in the summer, less in the winter. Farm investments to expand the herd, improve breeding stock, control diseases, and install milk cooling systems are economic factors that can contribute to increase production.

This study is based in the cash flow estimated to carry out Sant´anna et al’s. (2000) financial analysis, where the items "incentive for supplying cooled milk" and "incentive for freight reduction", both seen as benefits from the use of cooling tanks, represent price and volume variation. These two items were selected in such a way that when each value is altered, all other values are altered, generating a new estimate for the IRR (return internal rate). A total of 1,000 simulations were performed for each of the three alternative tank sizes, and the computations were processed by the @RISK program using the technique known as Monte Carlos simulation.

Random selection of the parameters to be analyzed requires fitting the respective input data into a probability distribution. These distributions were selected by using the software "BestFit" (Palisade Corporation, 1994) that can test up to 28 types of different distributions using the minimum squares method and find the curves that better fit the supplied data.

Milk price values were obtained from the Brazilian web site Milk Point (2000). The monthly average milk prices paid to farmers in 1999 (Table 1) were evaluated by means of the BestFit software, which recommended the uniform distribution as the most adequate, i. e, presenting the lower qui-square.



Daily milk production volume variations were estimated as a function of the amount of raw milk acquired monthly in Brazil in 1999 (Table 2). These values were evaluated with the aid of BestFit, which indicated that the triangular distribution best fit the input data.



The year 1999 was chosen as the only basis for analysis, as these values are the most representative of recent changes in milk production.

In the @RISK software’s RISKUNIFORM function, minimum and maximum prices are the only input parameters. For the same software’s RISKTRIANG function, the input data are the minimum, median, and maximum production volume values. Table 1 shows that in 1999 the minimum milk price was R$0.2608 and the maximum price was R$0.309.

Monthly volume of milk produced, shown in Table 2, varied 12% between the dry season (from April to July) and rainy season (November to February). This difference was utilized to calculate the minimum milk value, with the correspondence of 90% and 100% of the capacity of each tank being considered as the medium and maximum values respectively. For volume of milk produced, the following daily volume values were attributed by tank size (125, 250, and 500 liters): minimum values were 99, 198, and 396 liters (88% of the medium value, corresponding to a variation of 12% between the amounts of milk acquired during the wet and dry periods); medium values were 112.5, 225 and 450 liters (90% of the tank’s daily capacity), and maximum values were 125, 250 and 500 liters (100% of the tank’s daily capacity).

The values resulting from each alteration of the "price" and "volume" parameters are shown in Tables 3, 4, and 5, and reflect the statistical results generated after @RISK program simulation. Considering that milk supply is inversely related to price paid to the farmer, a correlation coefficient of 90% between these two variables was adopted.







Table 3 shows that the most frequently simulated IRR value for the 250 liter tank is 17.3%. This means that there is a 50% probability that profitability will be either lower or higher than this index value. For the500 l and 1000 l tanks, Tables 4 and 5 show that the most probable IRR values are 38.0% and 57.9% respectively.

Figures 2 and 3 show that for the 500 and 1,000 l tanks, the probability of IRR lower than 12% per year, which is the minimal rate considered attractiveness in this work, is 23% and 13% respectively.Figure 1 shows that, despite an 83% chance of positive results, there was a 42% chance of annual profitability below 12% for the 250 l tanks. Therefore, adoption of 500 and 1,000 l tanks is less risky and more desirable than the adoption of 250 l tanks, although risk remains. It should be noted that the financial analysis of the acquisition of 250 l tanks showed that their profitability was lower than the opportunity cost of capital used for their purchase.


Figure 1 = Click to enlarge



Figure 2 = Click to enlarge



Figure 3 = Click to enlarge


4. Conclusions

Risk analysis for the acquisition of 500 and 1,000 L cooling tanks shows 23% and 13% probabilities, respectively, of obtaining RIR values below 12% annually. For the 250 l tank, the probability of returns below the opportunity cost of capital (12%) is 42%. Using this study’s volume parameters, results show that the acquisition of cooling tanks that will have a mean yield of 112.5 liters/day or less is the least attractive investment.

The results also indicate that the risk of cooling tank investment failure (an increase in profitability less than the opportunity cost of invested capital) is reduced by the potential for increased milk production. Conversely, capital requirements will restrain small farmers’ cooling tank investments. This restraint and difficulties faced by small farmers when trying to increase production will probably lead to the exclusion of a significant number of small farmers from the formal dairy product market as Brazil’s dairy business modernizes, the demand for cooled milk delivery by dairy industries grows, and short-term Ministry of Agriculture regulations are implemented. The exclusion of small dairy farmers cannot be overlooked when public policies for the dairy sector are discussed. In this context, it is important that the implementation of new milk quality regulations occurs simultaneously with the implementation of non-exclusionary compensatory policies that smooth the transition from the current logistical system to a more modern one, thereby, minimizing social costs.

Within this context, an alternative has been considered: the implementation of cooperative cooling tanks to be utilized by a group of farmers. However, since this system has disadvantages and organizational and quality-control related problems, the use of cooperative cooling tanks should be seen as a shorter term mitigation measure.

Current social concerns and the fast pace of Brazilian dairy business modernization suggest that the installation of cooling tanks will become essential to the survival of Brazil’s small dairy farmers. It appears that Brazil’s dairy industries will increasingly demand cooled milk transported in isothermal tanks. Therefore, the sustainability strategy recommended for small dairy farmers in Brazil is to increase milk production by increasing yields through the adoption of herd nutrition and breeding management technologies.



ADIB, A. R. Análise de renda e risco na integração entre pequenos agricultores e agroindústria: o caso do Projeto Ilha Grande - Bahia. Viçosa, MG: UFV, 1996. 100 p. Dissertação (Mestrado em Economia Rural) - Universidade Federal de Viçosa, 1996.        [ Links ]

CONFEDERAÇÃO NACIONAL DE AGRICULTURA - CNA. Departamento Econômico. Informação Pessoal. Brasília, 2002.        [ Links ]

ESCHENBACH, T. G. Spiderplots versus tornado diagrams for sensitivity analysis. Interfaces, v.22, n.6, p.40-46, 1992.        [ Links ]

INSTITUTO BRASILEIRO DE GEOGRAFIA E ESTATÍSTICA - IBGE. Quantidade de leite cru ou resfriado adquirido, segundo os meses - Brasil, 1999. [21/05/2000]. ( agropecuaria/producaoagropecuaria/default.shtm         [ Links ])JANK, M. S., GALAN, V. B. Mercado atual e competitividade do sistema agroindustrial do leite. In: CONGRESSO LEITE - CADEIA PRODUTIVA: DESAFIOS E OPORTUNIDADES, 1998, Belo Horizonte, MG. Congresso leite..., Belo Horizonte, MG: FAEMG, 1998. p.17-33.        [ Links ]

LOPES, J. E. P. Análise econômica de contratos de integração usados no complexo agroindustrial avícola brasileiro. Viçosa, MG: UFV, 1992. 105 p. Dissertação (Mestrado em Ciência e Tecnologia de Alimentos) - Universidade Federal de Viçosa, 1992.        [ Links ]

MARTINS, E., ASSAF NETO, A. Administração financeira: as finanças da empresa sob condições inflacionárias. São Paulo: Atlas, 1992. 559 p.        [ Links ]

MILKPOINT. Mercado de leite: cotações com base no mês de produção. [21/05/2000].(        [ Links ]

NEVES, A. L. R. A. Viabilidade técnico-econômica e análise de risco da implantação de microcervejarias no Brasil. Viçosa, MG: UFV, 1996. 82 p. Dissertação (Mestrado em Ciência e Tecnologia de Alimentos) - Universidade Federal de Viçosa, 1996.        [ Links ]

NORONHA, J. F. Projetos agropecuários: administração financeira, orçamento e viabilidade econômica. 2. ed. São Paulo: Atlas, 1987. 269 p.        [ Links ]

ORGANIZAÇÃO DE COOPERAÇÃO E DESENVOLVIMENTO ECONÔMICO - OCDE. Manual de análise de projetos industriais nos países em desenvolvimento. Tradução Artur Lenza e outros. São Paulo: Atlas, 1977. 1 v.        [ Links ]

PALISADE CORPORATION. Fit distributions to data - bestftit. New York, USA: 1994. Não paginado.        [ Links ]

PALISADE CORPORATION. Risk analysis and simulation add-in for Microsoft Excel or Lotus 1-2-3. New York, USA: 1995. Não paginado.        [ Links ]

SANT´ANNA, D. D., SILVA, C. A. B., GOMES, S. T. Análise de viabilidade financeira da implantação da coleta de leite a granel nas propriedades rurais. Revista Leite & Derivados, ano X, n.56, p.52-56, Dez. 2000.        [ Links ]

SERVIÇO BRASILEIRO DE APOIO ÀS MICRO E PEQUENAS EMPRESAS - SEBRAE-MG. Diagnóstico da pecuária leiteira do Estado de Minas Gerais. Belo Horizonte: 1996. 102p.        [ Links ]

SZEKERES, S. Considering uncertainty in project appraisal. Washington, USA: International Bank for Reconstruction and Development, 1986. 60 p. (EDI Training Materials, 030/088).        [ Links ]

WHITE, J. A., CASE, K. E., PRATT, D. B., AGEE, M. H. Principles of engineering economic analysis. 4. ed. New York, USA: John Wiley & Sons, 1998. 491 p.        [ Links ]