| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Dairy Production Department, Teagasc, Moorepark Production Research Centre, Fermoy, Co. Cork, Ireland
2 Department of Agribusiness, Extension and Rural Development, Faculty of Agriculture, University College Dublin, Belfield, Dublin 4, Ireland
3 Department of Animal Science, Faculty of Agriculture, University College Dublin, Belfield, Dublin 4, Ireland
Corresponding author: L. Shalloo; e-mail: lshalloo{at}moorepark.teagasc.ie.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
53,547, and that with S2 was 51,687; the annual EU milk quota was 468,000 kg, and farm size was 40 ha. Monte Carlo simulation showed that the S1 scenario was stochastically dominant over the S2 scenario. Sensitivity analyses showed that farm profit was most sensitive to changes in milk price. The partial coefficients of determination were 99.2, 0.7, and 0.1% for milk price, concentrate cost, and silage quality, respectively, in S1; the corresponding values in S2 were 97.6, 2.3, and 0.1%. Validations of the model showed that it could be used with confidence to study systems of milk production under Irish conditions.
Key Words: whole-farm stochastic budgeting • Monte Carlo simulation • milk production system
Abbreviation key: CDF = cumulative probability distribution function, CSO = Central Statistics Office, DairyMIS = dairy management information system, MDSM = Moorepark Dairy System Model, NE = net energy, S1 = mean calving date February 24, S2 = mean calving date January 27
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Three uses of models have been identified in animal research: component, system, and management research modeling. Modeling factors controlling digestion, metabolism, and production of animals (Neal and Thornley, 1983; Baldwin et al., 1987) are examples of component research modeling. The use of models in system research is aimed at characterizing and understanding the interactions that occur between components at the production system level. Examples of such models found in grazing research include investigating the interaction between the grazing animal and pasture production (Parsons et al., 1983; Korte et al., 1984). Management models are normally used to investigate the effects of short- and long-term management options on output (production, returns, and risk) of the system. Examples of such models are the dairy system simulation model (Rotz et al., 1989), the Texas A&M sheep simulation model (Blackburn and Cartwright, 1987), and the deterministic static linear model (Berentsen and Giesen, 1995).
Mathematical models in agricultural systems use 2 general techniques: simulation or optimization. Simulation models of agricultural systems are developed to accurately describe the evolution of the systems. It can be argued that simulation models lack the creditability of field experiments (McCall, 1993), but that they provide the opportunity to explore difficult relationships that cannot be explored in any other way. They allow examination with a far greater range of variables over a much wider range of conditions than is feasible in practice, e.g., analysis of factors affecting wool production in Victoria, New Zealand (Bowman, 1989). A well-designed model can simulate various outcomes, allowing the study of their probability distribution using stochastic simulation (Swinton and Black, 2000). Optimization models seek to optimize some criterion or set of criteria subject to a set of constraints. Optimization models can support the decision-making objectives of agricultural managers, such as minimizing the cost of animal weight gain, enhancing weed control, or boosting farm profits (King et al., 1993). Optimization models are generally developed for a specific situation and are, therefore, less suited to study the consequences of a wide range of management strategies (Jalvingh et al., 1992). This present model was constructed to allow investigation of the effects of varying biological, technical, and physical processes on farm profitability; therefore, a simulation model was considered most appropriate.
A number of agricultural system models have been developed (Richardson and Nixon, 1986; Rotz et al., 1989; Milham, 1992; Berentsen and Giesen, 1995; McCall et al., 1999). Richardson and Nixon (1986) developed a stochastic whole-farm budgeting model (FLIPSIM), which simulated, under price and yield risk, the annual economic activities of a representative farm. Rotz et al. (1989) developed the dairy forage system model (DAFOSYM) to evaluate technologies and management strategies on representative dairy farms. Milham (1992) developed a stochastic whole-farm budgeting system, RISKFARM, to enable the appraisal of the financial performance and risk effects of alternative farm and non-farm investment of Australian farms. Berentsen and Giesen (1995) developed a deterministic static linear programming model of a dairy farm to determine the effect of institutional, technical, and price changes on a farm plan, economic results, and nutrient losses to the environment. McCall et al. (1999) developed a linear programming model to determine the optimum feeding strategies for dairy systems in both the Northern US and New Zealand.
The rationale for the development of the present farm simulation model was to be able to examine aspects of seasonal grass-based systems of production using minimal concentrate. Alternative management strategies and the use of different breeds or strains under varying feeding regimens have been evaluated (Dillon et al., 1995; Buckley et al., 2000; Kennedy et al., 2002; Dillon et al., 2003). Computer simulation is an approach that can help in the planning, interpretation, and analysis of such studies. There are 3 objectives in this paper. The first is to describe the Moorepark Dairy System Model (MDSM). The unique aspect of this model is the use of budgetary simulation and stochastic modeling of a milk production system. The budgetary simulation incorporates the many biological (milk yield and composition, BW, nutritional requirement, fertilizer), physical (land, labor, buildings), and economic (costs, valuations, profit and loss account, balance sheet) processes on a typical dairy farm. Stochastic budgeting, using Monte Carlo simulation, is used to determine the influence of variation in milk price, concentrate costs, and silage quality on farm profitability. The second objective is to validate the model by comparing both the physical and financial performance of a sample of dairy farms to the outputs from the model. Finally, the third objective is to demonstrate an application of the model using 2 calving patterns.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
shows a schematic diagram of the MDSM. The model was simulated over a 12-mo period. The dairy herd was broken down by month of calving. The worksheet displays the following information for each month of calving over the 12 mo: livestock movement (culling, purchases, and sales) at the start and end of each month, livestock valuations (cows, male and female calves, and male and female yearlings) at the start and end of each month, expected milk production, and feed requirement. Details of the simulation model are described subsequently. The case farm used in this study was 40 ha in size with a 468,000-kg EU annual milk quota, which limits the amount of milk that can be sold from the farm. The soil type was considered a free-draining acid brown earth of sandy loam to loam texture. Ireland enjoys a favorable climate for grassland farming, with winter temperatures high enough to allow some grass growth and relatively high summer rainfall (i.e., serious drought is not common). The basic element of the MDSM is the dairy cow.
|
This milk production system was spring calving grass-based (Dillon et al., 1995; Kennedy et al., 2002), which is representative of the production system on most Irish dairy farms. It is very dependent on the efficient utilization of grazed grass (Stakelum, 1991). Calving date is targeted to commence at the start of the grass-growing season. The objective of the system is to optimize grazed grass as a proportion of the total diet of the lactating dairy cow, allowing high cow performance while, at the same time, minimizing the cost of milk production. Feeding and management were based on current best practices in seasonal spring calving pasture-based systems (O’Donovan, 2000). The proportion of feeds offered (grass, grass silage, and concentrate) was altered to meet the net energy (NE) requirement for milk production, maintenance, and BW change (Jarrige, 1989). A fixed feeding system was used for young stock as per Teagasc guidelines (Teagasc, 2002).
Livestock movement and valuation.
The model displays the number of cows at the start and the end of each month and the number of cows culled (Table 1). It was assumed that cows were dried off at 2 mo before the next calving, and replacements were introduced as in-calf 2-yr-old heifers to replace culled and dead cows. The culling percentages (voluntary plus involuntary) per month of lactation used in the model were 1.46, 1.31, 1.03, 2.34, 0.50, 0.05, 0.92, 0.63, 0.83, 1.16, 5.37, and 3.26% for mo 1 to12, respectively, and similar to that being obtained in Irish spring calving dairy herds (Crosse, 1986). These culling rates resulted in 20% of cows being replaced each year. The model also displays the purchases and sales of livestock in each month. The valuation of cows and other livestock is also calculated at the start and the end of each month.
|
1206 at 23 mo of age. It was assumed that these replacements were purchased for the same price, in fixed proportions, 70, 20, and 10%, in February, March, and April, respectively. This simulates the practice among dairy farmers to consider young stock rearing as a separate enterprise; it also allows for the flexibility at farm level to sell surplus replacements, which, for example, calve at a more suitable time. All male and surplus female calves were sold at 1 mo of age. Therefore, MDSM allows the flexibility to consider rearing replacements as a separate enterprise or part of the dairy enterprise, depending on the sale date of young stock.
Milk supply.
Annual milk production averaged 6421 kg per cow with 38.6 g/kg fat, 34.0 g/kg protein, and 46.6 g/kg lactose; the stocking rate was 2.35 cows/ha. The lactation curves for milk yield and milk composition were based on the results of a 3-yr (1998—2000) farm study carried out at Moorepark (Kennedy et al., 2002). The lactation curves for milk yield, as affected by calving date, were obtained from Irish spring calving herds (Crosse, 1986), with no random variation. Milk yield and milk composition were modeled for an average group for each month of calving rather than for individual cows. Similarly, the average BW and BW change for milking cows during lactation were also calculated from the Moorepark farm study.
Feed demand.
The feeding regimen modeled had fixed ratios of grazed grass, grass silage, and concentrate for each month of the year for the 3 mo of calving. These ratios were not influenced by milk yield, but the amount of feed offered was altered to meet the NE system (Jarrige, 1989). The NE content of concentrate was determined using the feed unit for lactation (UFL) content of the ingredients (O’Mara, 1996). The NE values of the silage were related to the in vitro DM digestibility. The NE content of the herbage was related to its chemical composition (Jarrige, 1989). The following equations were used to determine the energy required for maintenance, pregnancy, and energy output in milk.
Energy required for maintenance:
 | ([1]) |
Energy required for pregnancy:
 | ([2]) |
Energy required for milk output:
 | ([3]) |
where AA = activity allowance based on an increase of 10% when cows are indoors and 20% when on pasture; BWC = BW of calf at calving (assumed to be 40 kg); DBC = days before calving; and FC, PC, and LC = milk fat, protein, and lactose content (g/kg), respectively.
A correction factor was added to the energy requirement based on the proportion of concentrate in the dietary DM:
 | ([4]) |
where 
NE = NE requirements corrected for the proportion of concentrate in the diet, PrC = proportion of concentrate in diet, and NEI = NE intake.
Grass supply.
Total yearly grass production was assumed to be 13 tonnes DM/ha based on measurements at Moorepark (1982 to 2001) using the method described by Corral and Fenlon (1978). Grass utilization was assumed to be 85%. Grass growth was dependent on nitrogen and the location of the model farm. Grass cover estimates were similar to recommended guidelines (O’Donovan et al., 2002). Nitrogen, phosphorous, and potassium concentrations used in the model were 337, 12, and 33 kg/ha, respectively. Teagasc nutrient management guidelines were observed; these allowed the recycling of phosphorous and potassium produced on the farm in the form of slurry (Teagasc, 2001).
Land and capital.
Land area was treated as an opportunity cost; additional land was rented in when required or leased out when not required for on-farm feeding of animals. Grazing management, silage harvesting, and grass production were similar to that reported previously (Dillon et al., 1995). The land areas for first and second cut silage (ratio 3:2, respectively) and for grazing were optimized to meet silage and grass requirements. Costs for fertilizer application, reseeding, and silage making (contractor, additives, and polythene) were based on the actual area required for silage and grazing.
Land improvements and buildings were depreciated at 10%/yr using the straight line method, and machinery was depreciated at 20% using the reducing balance method (O’Mahony, 1992). The book values used at the start of January for buildings and static machinery were those for the 6th and 7th yr as building or purchase, respectively. Provision was made for the purchase and depreciation of new fixed assets during the year. A 15-yr bank term loan was used to fund the cost of land improvements and buildings. The interest rate was fixed at 7.3%, and the loan was considered to be in its 7th yr; the interest portion of the repayment was considered as an expense.
Labor.
Labor requirements were based on the results of a comprehensive farm labor efficiency study on 143 Irish dairy farms (O’Donovan et al., 2003). Total farm labor requirement was divided into 8 main categories (milking, maintenance, grassland, management, calf care, cleaning, veterinary, and miscellaneous). In the model, labor requirement was divided between time associated with the cow and other farm tasks. The labor associated with the cow includes tasks such as actual milking, calf care, cow care, and veterinary. Droving, herding, maintenance, and miscellaneous were included as other tasks. The labor cost was calculated assuming 1848 h equates to one labor unit/yr at a cost of 22,855. Full labor costs were included in the MDSM because of its large influence on efficiency and productivity at farm level.
Receipts.
The gross milk price was 0.295/kg, based on a reference of 36.0 g fat/kg and 33.0 g protein/kg used by most Irish dairy manufacturers in 2001. A relative price ratio of 1:2 for fat and protein was assumed based on world market prices and international trends (Veerkamp et al., 2002). Hence, milk payment was based on kilograms of fat (3.04/kg) plus kilograms of protein (6.09/kg) delivered minus a cost of 0.017/kg made up of 0.004 for levies and 0.013 for transportation and cooling (Keane et al., 1998). Value-added tax was added to the gross milk price at a rate of 4.3% in accordance with current tax legislation in Ireland (Teagasc, 2002).
The calf and cull cow prices were based on average prices between 1996 and 1998 (Teagasc, 2002). Cull cow valuation depended on the month of disposal and lactation stage. Prices ranged between 476 in the 2nd mo after calving to 373 at the end of lactation. Male and female calves were sold at 1 mo of age and were valued at 207 and 336, respectively.
Costs.
The worksheet holds information month-by-month for each of the operating costs, such as fertilizer application, silage making, veterinary medicine etc. All variable costs (concentrates, fertilizer, reseeding, machinery hire, silage making, veterinary medicine, and AI) were based on current prices (Teagasc, 2002). Reseeding costs were based on 5% of the farm being reseeded each year with a reseeding cost of 577/ha. Veterinary costs included routine animal treatments as well as compulsory annual tuberculosis and brucellosis testing of animals and drugs involved in correction of infertility problems in cows. Artificial insemination costs were based on 1.7 inseminations per conception with each insemination costing 25.40 plus a service charge of 11.43. As for variable costs, all fixed costs (car, electricity, machinery operation and repair, telephone, insurance) were based on current prices (Teagasc, 2002). Two-thirds of the total costs associated with maintenance and operating a car were charged to the farm (O’Sullivan and O’Neill, 2002). Costs associated with operation and servicing the milking parlor were included in the machinery operation and repair costs.
Output indicators.
The outputs from the model include financial indicators (operating cash flow, profit and loss account, and balance sheet) and physical outputs such as feed budget, nutrient balance sheet, and physical ratios. Operating cash flows are summarized for each month of the year and indicate the cash surplus or deficit in each month. The estimated farm profit and loss account is presented on a total farm basis as well as per hectare, per cow calving, and per kilogram of milk produced. Farm net profit included total receipts less total costs, including a notional charge for full labor costs. The fixed cost distinguishes interest costs on an overdraft account from term liabilities. Interest earned on the cash flow in the current account is distinguished from other farm receipts. The farm balance sheet summarizes the assets and the liabilities of the business, and it estimates net worth.
The physical outputs from the model include physical ratios and a nutrient balance report. Physical outputs also contain a tornado feed-budget graph of the requirements of grass, grass silage, and concentrate for each month of the year on an individual cow basis.
Stochastic Budgeting
A key and innovative part of the development of MDSM was the inclusion of stochastic simulation modeling. Stochastic features were introduced into the budget. Stochastic simulation was carried out in the MDSM using the computer software @Risk (Palisade, 2000), which works by a process, of "Monte Carlo Sampling." Monte Carlo risk assessment (also called Monte Carlo uncertainty assessment) specifies a probability distribution for each sensitivity parameter, draws a set of those parameters, and repeats the conventional analysis for multiple draws (Phillips and Maldonado, 1999; Phillips, 2000). A sufficiently large number (10,000) of simulations were run with the same input distributions so that the probability distribution functions of the outputs were adequately described (Isukapalli et al., 1998).
Stochastic budgeting was used to model the influence of variation in milk price, concentrate cost, and silage quality, because these 3 factors were considered to have the largest effect on whether S1 was stochastically dominant to the second calving pattern (S2; mean calving date January 27). In S2, a greater proportion of milk was produced from grass silage and concentrate supplementation than from grazed grass; milk price was included, as 77% of total output came from sale of milk in the model. The values for milk price were obtained from projections from the Food and Agricultural Policy Research Institute—Ireland Outlook 2002 (Binfield et al., 2002). The report projected that milk price would decline by 6 and 9% in the EU average and Ireland, respectively, by 2010. The base milk price of 0.281/kg, which corresponded to the year 2000, was reduced by 9% to correspond to the year 2010. Variation in milk price over the period from 1995 to 2002, obtained from Central Statistics Office (CSO) (2003), was used to determine the spread. A computer program called Bestfit (Palisade, 2000) was used to estimate a distribution of milk price from the CSO data. The same approach was used to estimate a distribution of concentrate costs. The Bestfit program used the Chi square test to rank 1 to 38 distributions that best suited the data. The program picked the Beta General Distribution as the one that best suited the data. For silage quality, the stochastic variables were included as triangular distributions, where the minimum, most likely, and maximum values were included (Hardaker et al., 1997). The most likely, minimum, and maximum values were 680, 620, and 740 g/kg DM digestibility, respectively; these were based on results of analysis of silage samples submitted to Teagasc, Grange Laboratories over the period 1993 to 1996 (Keating and O’Kiely, 1997). The stochastic variables were simulated simultaneously, and it was assumed that there was no correlation between variables. Binfield et al. (2002) projected that in Ireland milk price would reduce by 30% between 2002 and 2010; at the same time, concentrate costs would remain unchanged. Similarly data from the CSO in Ireland over the period from 1995 to 2002 indicated no significant correlation between milk price and concentrate cost, with <20% of the variation accounted for. It was assumed that silage quality was independent of concentrate cost and milk price.
Milk Production Scenarios
The milk quota system, introduced by the EU in 1984, currently limits the level of herd milk production in Ireland. Milk production on farms is restricted to available milk quota on the holding. This quota is on milk yield adjusted for fat content using the following formula:
 |
where BF% is butter fat percentage of the milk and the reference year was 1984. It was assumed that farmers were constrained by the EU milk quota, and the consequence of higher milk yield per cow would be a reduction in cow numbers (i.e., evaluation was based on a fixed output). This is representative of the present quota position.
Two milk production scenarios were investigated that corresponded to two different calving patterns. In S1 (default scenario), 70, 20, and 10% of the cows in the herd calved in the months of February, March, and April, respectively; this corresponded to a mean calving date of February 24. In the S2 scenario, 70, 20, and 10% of the cows in the herd calved in the months of January, February, and March, respectively; this corresponded to a mean calving date of January 27 and a calving spread of 89 d. The physical and financial outputs from the 2 calving patterns are presented.
Model Validation
Model validation is a test of the ability of the model to accurately assess the intended application (Milham, 1998; Sørensen, 1990). In the present study, actual data from 21 spring calving dairy herds and simulated data for the S1 scenario were compared for the year 2001 to determine the reliability of key model outputs. The actual data were obtained from herds that were part of a Dairy Management Information System (DairyMIS) (Crosse, 1986) attached to Moorepark. The DairyMIS is a recorder-based computer system for collecting detailed farm inputs and production information on a monthly basis. Financial and physical data for these farms for the past 20 yr were available; however, only 2001 data were used as the costs in the model were based on 2001–2002 prices. Replacement heifer rearing was considered to be part of the dairy enterprise. In farms where there was some beef production (3 farms in total), all production costs on the farm were reduced on a livestock-equivalent basis (O’Sullivan and O’Neill, 2002). The model output data were compared with the mean, minimum, and maximum of the comparable variables from the selected farms.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
shows the number of animals present, animal sales and purchases, milk production and composition, feed requirement, land use, and labor requirement for each of the 12 mo of the year for S1 scenario. Cow numbers averaged 73.0 over the year, and 15.7 replacement heifers were required to replace the 1.2 cow deaths and 14.5 cows culled. The corresponding values for S2 were 71.5 cows, 15.4 replacement heifers, 1.2 cow deaths, and 14.2 cows culled. Number of male and female calves sold were 38.4 and 21.1 for S1 and 37.6 and 20.7 for S2, respectively. The farm produced a total of 462,621 and 462,308 kg of milk in the S1 and S2 scenarios, respectively, and 14,584 and 14,295 kg of this were fed to calves. Peak milk production was in May and lowest milk production was in January in S1; the corresponding months were April and December in S2. Total labor requirement for the year in S1 was 2978 hs (1.61 labor units); in S2, total labor was 2922 h (1.58 labor units).
Figure 2 shows the average feed intake per cow for each month of the year for the S1 scenario. The average stocking rate over the total year was 2.35 cows/ha. Total DMI (kg per cow) were 3590, 1201, and 350 of grazed grass, grass silage, and concentrate, respectively, and these correspond to 70, 23, and 7% of the total diet, respectively.
|
shows the average feed intake per cow for each month of the year for the S2 scenario. In S2, the average stocking rate over the total year was 2.44 cows/ha. Total DMI (kg per cow) were 3409, 1183, and 709 of grazed grass, grass silage, and concentrate, respectively, and these correspond to 64, 22, and 13% of the total diet, respectively.
|
shows the profit and loss account for the MDSM for S1 using the assumptions already outlined. In S1, total farm receipts and farm net profit were 177,987 and 53,547, respectively. These correspond to 0.385, 0.114, 0.122, and 0.116/kg, for receipts, variable costs, fixed costs, and farm net profit, respectively. Milk accounted for 77% of total sales in S1; variable, fixed, and depreciation costs accounted for 42, 45, and 13% of total costs, respectively.
|
shows the profit and loss account for the MDSM for S2. Total farm receipts and farm net profit were 177,152 and 51,687, respectively. These correspond to 38.3, 12.0, 12.0, and 11.2 cents (euro)/kg for receipts, variable costs, fixed costs, and farm net profit, respectively. As in S1, milk accounted for 77% of total sales in S2; variable, fixed, and depreciation costs accounted for and 44, 44, and 12% of total costs, respectively. Labor costs accounted for 65% of total fixed costs in both S1 and S2.
|
shows a cumulative density function (CDF) of the influence of variation in milk price, concentrate costs, and silage quality on the spread of farm profit in both S1 and S2. The cumulative probability distribution shows that S1 is stochastically dominant over S2. First-degree stochastic dominance means that the CDF further to the right are preferred. For any given level of risk, there is a higher level of profit, or, conversely, for a given level of profit, there is lower risk when the CDF is further to the right. The mean farm profit was 33,309 and 30,895 for S1 and S2, respectively. The range in farm profit with a 90% confidence interval (5 to 95%) was 26,865 to 38,924 in S1; the corresponding values in S2 were 24,123 to 36,905.
|
Model Validation
Table 4 shows the physical performance on the 21 DairyMIS farms and that modeled in the MDSM using the S1 scenario. The results indicate that the MDSM provided a reliable representation of the observed and average performance in the DairyMIS farms. On average, the culling rates, milk price, milk composition, proportion of farm harvested for first and second cut silage, phosphorus, and potash application rates were very similar in the DairyMIS farms to those obtained in the MDSM. There were modest differences between data obtained in the MDSM and the average data for the 21 DairyMIS farms for calf mortality (MDSM was 6.1 percentage points lower), milk production per cow (MDSM was 251 kg per cow higher), stocking rate (MDSM was 0.12 cows/ha lower), nitrogen application rates (MDSM was 48 kg N/ha higher), and concentrate fed per cow (MDSM was 209 kg lower). Average cull cow and calf prices were 36 and 106 lower and higher, respectively, in the DairyMIS farms than in the MDSM. However, all of these parameters modeled in the MDSM were within the range obtained in the 21 DairyMIS farms.
|
and 6 show a comparison of the variable and fixed costs for the DairyMIS farms (mean, maximum, and minimum and standard deviation) and in the MDSM on a per kilograms of milk basis and a per cow basis, respectively. All of the financial parameters per kilogram of milk and per cow in the MDSM were within the range obtained on the DairyMIS farms. Total receipts, variable costs, fixed costs, and net profit per cow were 78 greater, 199 lower, 161 greater, and 2 lower, respectively, in the MDSM than the average for the DairyMIS farms. Total receipts, variable costs, fixed costs, and net profit (euro cents/kg milk) were 2.1 greater, 3.8 lower, 3.0 greater, and 0.5 greater, respectively, in the MDSM than the average for the DairyMIS farms.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Financial Performance
The good financial performance in the MDSM for S1 relative to 1999, 2000, and 2002 is due mainly to higher milk price in 2001. Fingleton (2002) using National Farm Survey data showed that net margins per liter of milk produced on specialist dairy farms were 1¢ higher in 2001 than the average of 1999, 2000, and 2002. The net profit per cow achieved in the MDSM (675) is similar to that reported by Fingleton (2002) from the National Farm Survey of Irish specialist dairy farms (678). Comparisons of total milk production costs between Ireland and other EU countries have shown that Ireland maintains a consistent competitive advantage over all other EU countries with the exception of Belgium (Boyle et al., 1992). However, costs of milk production in Ireland are 2.5 and 1.6 times those in New Zealand and Australia, respectively (Anonymous, 1993). The better financial performance of S1 relative to S2 is similar to that reported previously (Crosse and Dillon, 1993) and is a result of the lower cost of milk production because more of the cow’s diet is made up of grazed grass, which is a cheaper feed than either grass silage or purchased concentrate.
Stochastic Budgeting
Sensitivity analysis, although a valid and useful technique for determining the range of feasible outcomes from a model, does not give any indication of the likelihood of particular results being achieved (Milham and Hardaker, 1990). Risk assessment is a process for identifying adverse consequences and their associated probabilities (McKone, 1996). Whole farm stochastic budgeting involves developing a model that mimics the operation of a business and provides projections of financial performance while taking account of the uncertainty inherent in many aspects of the decisions (Milham, 1998). Beneficial Monte Carlo simulations can only be achieved when meaningful estimates of the input stochastic variables are used in the model. Hardaker et al. (1997) stated that the selected variables to be added stochastically are usually those that will have the largest effect on the level of risk of a certain outcome. The results from a Monte Carlo simulation are only as valid as the input parameter data and the assumptions used in the analysis (Petersen, 2000). There are many sources of data and a variety of methods that can be used to determine the input distributions using Monte Carlo analysis procedures (Petersen and Barraj, 1996). Lien and Hardaker (2001) noted that, in the case where historical data are not relevant, the values may be obtained from expert opinion. In the present study, the factors that were considered to have an effect on farm profit were milk price, concentrate costs, and silage quality.
A powerful risk efficiency criterion is stochastic dominance. The stochastically efficient, or dominant, set is found by comparing the CDF of risky prospects (Thomas et al., 1997). A CDF contains all of the information on the output distribution of the risky prospects and, therefore, provides a useful decision-making criterion by assessing stochastic dominance. When the S1 and S2 were presented on the CDF, the S1 scenario was further to the right than S2. Thus, S1 was stochastically dominant over S2 and would be preferred.
In the sensitivity analysis, the partial coefficients of determination calculated for each input variable measure the percentage variation in the output variable explained by that particular input distribution (Palisade, 2000). Milk price had, by far, the largest effect on the overall output distribution, which means that farm profit is very sensitive to any changes in milk price. Silage quality had similar influences in both S1 and S2, as similar amounts of silage were used in both scenarios. Concentrate cost had a much larger influence in S2 than in S1 because much larger amounts of concentrate were used in S2.
Model Validation
Model validation is dependent upon quantitative and qualitative analyses of differences between simulated and actual data. An essential requirement when building a model is that the model adequately represents the system (Damborg, 1985). The validation determines the extent to which simulated results match actual data and, therefore, test the assumptions inherent in the model structure and function. In the present study, there was good agreement between the actual farm data and the simulated results. Also, the net profit of 677 per cow obtained in the DairyMIS farms is similar to the 678 per cow in specialist Irish dairy herds (Fingleton, 2002). There were differences in land rental and interest repayment on term loans. It was assumed that there was no milk quota leasing in the simulated model, but there was milk quota leasing in the group of farms. It was not possible to obtain the level of borrowing on the DairyMIS farms. However, the model had interest charges and no lease charges while the farms had lease charges and no interest charges; both costs were similar and therefore, cancelled each other.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication July 23, 2003. Accepted for publication November 26, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. Kristensen, S. Ostergaard, M. A. Krogh, and C. Enevoldsen Technical Indicators of Financial Performance in the Dairy Herd J Dairy Sci, February 1, 2008; 91(2): 620 - 631. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. L. M. Schils, M. H. A. de Haan, J. G. A. Hemmer, A. van den Pol-van Dasselaar, J. A. de Boer, A. G. Evers, G. Holshof, J. C. van Middelkoop, and R. L. G. Zom DairyWise, A Whole-Farm Dairy Model J Dairy Sci, November 1, 2007; 90(11): 5334 - 5346. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. McCarthy, B. Horan, P. Dillon, P. O'Connor, M. Rath, and L. Shalloo Economic Comparison of Divergent Strains of Holstein-Friesian Cows in Various Pasture-Based Production Systems J Dairy Sci, March 1, 2007; 90(3): 1493 - 1505. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | < |
