| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Agricultural Economics and Rural Sociology, and
2 Department of Dairy and Animal Science, The Pennsylvania State University, University Park, 16802
Corresponding author: Kenneth W. Bailey; e-mail: baileyk{at}psu.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: milk composition • multiple component pricing
Abbreviation key: cwt = 100 pounds, DRMS = Dairy Records Management System, GR = gross revenue, IOFC = income over feed costs, PPD = producer price differential.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Under the previous regulations, producers were paid based on one price adjusted by milk fat content above or below 3.5%. Today they are paid a separate price for milk fat, true protein, and other dairy solids. These prices reflect the value of milk components in manufactured dairy products. In addition to this value, producers in federal orders also receive a producer price differential (PPD) each month. The PPD reflects the value of the federal order pool in a particular month above the manufacturing value (the statistical uniform price less the Class III price). The PPD is usually positive each month because Class I and II prices in federal orders are usually higher than Class III and IV prices due to mandated formulas that use differentials. Finally, some of the federal orders now pay producers a premium or discount each month according to their SCC level.
The implications of this new pricing system on farm profitability are not clearly understood. For example, is it more profitable to expand overall production by increasing productivity per cow? Or is it more profitable to maintain milk production per cow, but improve component levels for milk fat and protein by changing the feeding system or altering the genetic base of the herd? These are important factors to consider because 1) producers in federal milk marketing orders that use a multiple component pricing plan receive a substantial portion of their gross income from milk components, 2) producers have the ability to alter milk component levels to some degree, and 3) the relative price for milk fat and protein changes frequently with alternative market conditions creating income risk for producers.
A number of studies have analyzed the implications of alternative multiple component pricing schemes on farm income (Gruebele, 1982; Cragle et al., 1986; Kirkland and Mittelhammer, 1986). Other studies have focused more specifically on the economic returns by breed under multiple component pricing plans (Schmidt and Pritchard, 1988; Elbehri et al., 1994). Other studies have taken a broader look at the social welfare implications of switching to a multiple component pricing system (Perrin, 1980; Lenz et al., 1991; Gillmeister et al., 1996). Despite this large body of literature, no studies have been completed that evaluate the economic returns for actual herds from component performance since the most recent federal order reform.
The first objective of this study was to analyze the statistical properties of actual component data from herds participating in a federal milk marketing order in the Mideast and Northeast United States. In particular, we were concerned with the degree of dispersion of component data around the mean and its implications for profitability. The second objective was to develop a monthly simulation to evaluate the economic returns for Holstein and Jersey herds under the current multiple component pricing system.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Our initial criteria for data collection was to identify a federal order that 1) contained a significant volume of milk each month from Pennsylvania producers, 2) used component pricing, and 3) was willing to provide at least 3 yr of data. Using these criteria, milk production, component levels, and price data were obtained from the USDA’s Agricultural Marketing Service for Mideast Federal Order 33. The data supplied by USDA provided a monthly summary of every producer who shipped milk into the Mideast federal order for the period January 2000 through December 2002. This order includes producers located mainly in Ohio, Michigan, Illinois, Indiana, and Western Pennsylvania. More specifically, the USDA data set included monthly records for farm identification code, total pounds of milk shipped, percentage fat, percentage protein, percentage other dairy solids, and SCC. The farm identification code was used to protect the identity of individual herds. During the period of study, producers in the Mideast federal order did not have every load of milk sampled and tested for milk components and SCC. Rather, they had their milk tested an average of 4 to 5 times per month.
The original data set contained 356,339 observations. After analysis of extreme observations using the UNI-VARIATE procedure of SAS (SAS Institute, 1999), data were edited to exclude records as follows: milk fat or protein percentage >6%; other dairy solids percentage >6.5%; and SCC > 3,000,000 cells. After edits, the total number of observations was 356,322. Production and component data for the Mideast order were then summarized by month and year using the MEANS procedure of SAS (SAS Institute, 1999).
To provide information related to breed and production per cow, data were obtained from Dairy Records Management Systems (DRMS), Raleigh, NC. Data were extracted from herd summary reports for Holsteins and Jerseys in PA, OH, IN, MI, WV, and WI. For each herd a single test date for each month of 2000, 2001, and 2002, was used. Data included breed, herd-level averages for milk (milking cows only), milk fat, protein, SCC, and linear SCS. The original data set contained 170,352 observations. Records missing year or month information were dropped from the data set. After analysis of extreme observations using the UNIVARIATE procedure (SAS Institute, 1999), data were edited to exclude records as follows: daily milk per cow < 10 lb or > 200 lb; milk fat percentage > 9%; milk protein percentage > 6%; somatic cell score > 10; and SCC > 3,000,000 cells or < 10,000 cells. The total number of observations after edits was 166,025. Production and components from DRMS data were summarized by breed, month, and year using the MEANS procedure (SAS Institute, 1999).
A final source of data used in this study was feed prices from the Penn State Team Dairy Web site (Ishler, 2001–2002). These data are collected monthly from several sources including the Pennsylvania Agricultural Statistics Service market summaries for hay and grain, Feedstuffs magazine market reports, and local agribusinesses. Prices are calculated and reported monthly via the Internet. This report was accessed each month during 2001 and 2002 to develop a regional price database. All feed costs were provided in dollars per 100 pounds (cwt) of DM, which was converted to dollars per pound of DM for use in this study. Silage values in this report are calculated from values for dry hay expressed on a DM basis. Legume silage values are based on market prices for alfalfa hay, and corn silage values are based on market prices for grass hay. The corresponding value of silage DM is then used to calculate silage values at an average silage DM of 45% for legume silage and 33% for corn silage. For each ration ingredient, the feed cost used for simulation was a 2-yr average price that reflected regional feed costs for Pennsylvania dairy operations.
Simulation Model
An economic simulation model was developed to analyze the economic returns to milk production under alternative component levels for representative herds of Holstein and Jersey cows. The simulation model computed gross revenue (GR) by production and component level, feed costs, and income over feed costs (IOFC) by breed. The model also simulated the impact of producing above and below average component levels by breed.
Averages obtained from the federal order and DRMS datasets were used to develop simulated monthly revenues for milk deliveries for each breed; herd size of 100 cows was assumed. Gross revenues were calculated via the multiple component pricing formulas used in Federal Order 33 as follows:
 | ([1]) |
where GRi = monthly gross revenue, i = month, ßk= percentage of component k for delivered milk (mf = milk fat, pr = protein, and os = other solids), Pk = price of component k per pound, Pscc = SCC adjustment in $/cwt, PPD = producer price differential in $/cwt, and PRD = pounds of delivered milk.
Pounds of milk delivered were assumed to equal pounds of monthly milk production identified as follows:
 | ([2]) |
where Di are the number of days in the month, DAPi is daily average milk production, and COWSi are the number of cows milked each month.
The SCC adjustment is specified by the Mideast federal order as follows:
 | ([3]) |
where SCC = somatic cell count in thousands, and SCCR is the somatic cell count rate in $/cwt.
Finally, IOFC is specified as follows:
 | ([4]) |
where FC is feed costs per cow per day for month i. FC is specified as follows,
 | ([5]) |
where DM is pounds of DM per cow per day for feed component k in month i, and PDM is the price of DM k in month i in dollars per pound.
The data used to simulate monthly representative Holstein and Jersey herds in Pennsylvania were from the USDA federal order and DRMS datasets. The monthly baselines for daily average production, milk fat, protein, and SCC by breed were derived from the DRMS data set using a 3-yr average (2000 to 2002). The DRMS data set did not provide the percentage of other solids by breed. The yield of other solids does not vary as much as other components (Sharma et al., 1983; Rodriguez et al., 1997) and is not as valuable (USDA, 2000–2002). Therefore, for each month, a 3-yr average other solids value from the Mideast federal order data set was used for both Holsteins and Jerseys. Three-year average prices for each component, somatic cell count rate, and PPD were computed based on data obtained from the federal order market administrator’s Web site (USDA, 2000–2002).
Gross revenue by itself does not adequately represent economic returns to each breed for a typical Pennsylvania dairy farm under a multiple component pricing system. For example, milk production and component levels respond in part to the nutrient intake of the cow. Feed costs are usually the largest variable expense incurred by dairy farms and should be considered in any measure of economic returns. There are other variable and fixed costs associated with milk production that also would be prevalent on a commercial Pennsylvania dairy farm. Some of these data are published each year by the Pennsylvania Farm Bureau Member’s Service Corporation (MSC) Business Services (2002) and represent Pennsylvania and Northeast production costs for labor, feed, interest, rent, milk marketing, depreciation, and other costs. However, none of these data can be sorted by breed. In addition, these nonfeed expenses vary considerably from farm to farm. On the other hand, the relationships between milk yield and composition and feed requirements are well established and can be determined for different breeds (NRC, 2001). Therefore, for the purposes of this study, only feed costs were used to compare production expenses by breed.
Income over feed cost was calculated monthly for representative Holstein and Jersey herds. Feed costs were determined by estimated feed requirements and 2-yr regional average prices for ration ingredients (Ishler, 2001–2002). For each month, DMI was estimated for each breed based on the 3-yr average milk yield and composition derived from the DRMS data set. Intake requirements were estimated for a ration containing corn silage at 35% DM (Int. Feed No. 3-28-248), legume silage at 43% DM (Int. Feed No. 3-07-797), dry ground corn (Int. Feed No. 4-02-854), soybean meal (Int. Feed No. 5-20-638), and a mineral-vitamin mix (NRC, 2001). Silage was constrained to 60% of ration DM (30% from each source) and the mineral-vitamin mix was limited to 2% of DM. The amount of corn and soybean meal required was determined using NRC (2001) ration evaluation software. Protein was the first-limiting nutrient in this diet, so corn and soybean meal amounts were adjusted until the entered milk production was within 1 lb of the yield allowed by metabolizable protein (as predicted by the NRC model). Feed requirements were estimated individually for each month and breed.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
. The results indicate that milk fat and protein component levels vary seasonally, as expected. In addition, the standard deviation for milk fat (0.32%) is roughly twice that of either protein (0.19%) or other dairy solids (0.12%). The standard deviation for SCC is 55.5% of the mean, indicating a large dispersion around the mean. These initial results imply that there are large differences in the farm value of milk components among producers in the Mideast federal order in any given month and that there is a strong seasonal pattern for component production.
|
indicate that milk fat and protein percentages were highest in the fall, winter, and spring months and lowest in the summer months. These seasonal changes in milk composition are consistent with previous reports of milk component variation in the Northeast (Allore et al., 1997). In addition, variation in milk composition between herds in any given month was large, indicating that many herds have an opportunity for improving component production.
An initial statistical analysis of the data produced 12 separate means over 3 yr for each component (milk fat, protein, and other dairy solids). This method of analysis could obscure a simple observation of dispersion around the mean. Thus, the monthly data in Table 1 were adjusted to remove seasonal effects, which allowed the monthly data to be combined to isolate the deviation of individual observations around the mean. The seasonal adjustment was calculated by dividing the 3-yr average for each month by the overall average. This adjustment value was then applied to each observation to produce a new data set.
Variation in this new data set was not decreased by the adjustment for seasonal effects (Table 2). In the original data set, milk fat percentage was 3.76 ± 0.32% and protein was 3.05 ± 0.19%. After seasonal adjustment, milk fat percentage was 3.76 ± 0.30% and protein was 3.05 ± 0.17%. This minimal change indicates that many factors other than season contribute to the wide variation in milk component production. The distributions of milk fat and protein percentage after adjusting for the effects of season are shown in Figures 1 and 2, respectively. The distribution of SCC after adjustment for season is shown in Figure 3. Before adjustment, SCC averaged 350,000 ± 182,000 cells/mL; after seasonal adjustment, SCC averaged 350,000 ± 181,000 cells/mL. Although SCC is not normally distributed (evidenced by skewness of 1.22 and kurtosis of 2.89), these results suggest that many factors other than season contribute to the wide variation in SCC.
|
|
|
|
 | ([6]) |
The USDA data set for Federal Order 33 provided data for the entire population of producers who sold milk into the Mideast order over a 3-yr period. To evaluate the effects of herd size on milk components and SCC, each observation was assigned to one of the following categories, based on amount of milk shipped per month: < 50,000 pounds, 50,001 to 100,000 pounds, 100,001 to 500,000 pounds, or > 500,000 pounds. The data were then summarized by these categories of milk shipped (Table 3). The results indicate that smaller producers had higher levels of milk fat and protein, and thus a higher Class III value ($/cwt), than larger producers. On the other hand, SCC levels were much higher for smaller producers than for larger producers. In addition, the standard deviation was greater for milk fat, protein, and Class III value for smaller producers than for larger producers.
|
One weakness with this approach, however, is that the population includes both Holsteins and Jersey herds. Jersey herds would likely raise the population mean for components. Unfortunately, this data set cannot be sorted by breed. Thus, any Jersey herds that realized a decrease of more than one standard deviation from the mean for the Jersey group would not be measured by this analysis because the mean and standard deviation was computed for the entire population.
The concept of a herd-year was used in this analysis. A herd-year was defined as one observation that consisted of a herd code, year, and 12 mo of milk component data. Each herd could have up to 3 yr represented in the data set. For each herd-year, monthly production of milk fat, protein, and SCC were compared with the 3-yr monthly average of milk fat, protein, and SCC, and a rate of attainment (number of months out of that year) was calculated for each of the following: low milk fat = milk fat percentage > one standard deviation below the mean; low protein = protein percentage > one standard deviation below the mean; low milk fat and protein = both milk fat and protein percentages > one standard deviation below their respective means; high SCC = SCC > one standard deviation above the mean. The frequency of these rates of attainment was then calculated using the FREQ procedure of SAS (SAS Institute, 1999). These frequencies did not provide the actual number of herds meeting the above criteria because each herd had a potential of 3 observations (1 per year). Therefore, frequency of each herd code was determined to provide the actual number of herds in each category.
The performance analysis identified 12,445 unique herds that participated in the Mideast federal order over a 3-yr period (Table 4). Over that time, 4.8 and 4.4% of the herds consistently had low milk fat or protein levels for 9 or more months of the year. Low component production for most of a year could result from herd genetics or management factors including nutrition or disease. In addition, 9.7% of all herds had a SCC > one standard deviation above the mean for 9 or more months of the year. Although the exact cause is not known, consistently high SCC most likely resulted from mastitis infection in the herd. Regardless of their origin, these consistent problems cost producers money and present an opportunity for improved profitability.
|
A final question addressed using the USDA federal order data was quantification of the loss in Class III value for a typical producer with milk fat and protein levels one standard deviation below the mean. An average Class III value was produced using equation 6 and average component levels and prices computed over the 3-yr period from the federal order data set (Table 1). This equation was recalculated after reducing milk fat and protein levels by one standard deviation. The results indicated that during this 3-yr period, producing milk with milk fat one standard deviation below average reduced the Class III value by $0.46/cwt, or 4.02%. Milk with low protein was worth $0.36/cwt or 3.07% less. Finally, milk with both milk fat and protein one standard deviation below the mean reduced the Class III value by $0.82/cwt or 7.09%.
Simulated IOFC
The next step in this study was to complete the monthly simulation for representative 100-cow Holstein and Jersey herds utilizing relevant pricing information for milk components and feed ingredients to compute both GR and IOFC.
Breed is a known factor that affects milk component production (Sharma et al., 1983). To quantify these differences, we summarized test day milk production and composition data provided by DRMS by breed. Differences between Holsteins and Jerseys are clearly demonstrated in Tables 5 and 6. Holsteins produced about 20 lb/d more milk than Jerseys. In addition, test-day protein and milk fat were 0.5 and 1.0 percentage points greater for Jerseys than Holsteins. These disparities are in the same range as previous reports of breed differences (Sharma et al., 1983; Rodriguez et al., 1997).
|
|
. The pricing information in Table 7 was combined with equations 1 through 5 and breed information in Tables 5 and 6 to produce the baseline results in Table 8.
|
|
Alternative Scenarios
The results for IOFC described above are for a 100-cow Holstein or Jersey herd producing an average volume of milk with average component levels. Of particular interest is how IOFC would change under alternative scenarios. For example, how much would results change if the Holstein or Jersey herds were to improve milk production or component levels? This would require additional feed, but could improve IOFC. More importantly, what are the IOFC implications for producing components below the baseline averages?
A number of scenarios were developed for both the Holstein and Jersey herds to simulate these alternative impacts on IOFC. These scenarios are described as follows:
Milk revenues and feed requirements were estimated for each scenario (Table 8) as described previously. The results of this simulation suggest that increasing milk fat and protein percentages for the Holstein herd by one standard deviation increased IOFC by $16,096, or 7.7% of the baseline IOFC. On the other hand, a reduction of milk fat and protein by one standard deviation decreased IOFC for the Holstein herd by $15,988, or 7.7% of the baseline IOFC. Increasing milk yield by one standard deviation increased IOFC by $40,887, or 19.6% of the baseline IOFC. Decreasing milk yield by the same amount reduced IOFC by $42,583, or 20.4% of the baseline IOFC. Results for the Jersey herd were similar. An increase in milk fat and protein percentages by one standard deviation resulted in $16,229 more IOFC (9.2% of baseline IOFC); a one standard deviation decrease reduced IOFC by $15,808 (–8.9% of baseline IOFC). Increasing milk production by one standard deviation resulted in $42,266 more IOFC (23.9% of baseline IOFC), while reducing production caused a $41,425 drop in IOFC (–23.4% of baseline IOFC).
Feed costs in this simulation were calculated with limited ingredients for specific combinations of milk yield and composition and ration changes accounted for alterations in milk fat and protein simultaneously. When attempting to change herd production of milk fat and protein, ration changes to increase milk protein are likely to be more expensive than changes to increase milk fat.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
The simulation developed in this study for representative Holstein and Jersey herds in Pennsylvania quantified the amount of money lost by producing milk with below average components. Producing milk at one standard deviation below the mean for both milk fat and protein resulted in a loss in IOFC of 7.7% for the Holstein herd and 8.9% for the Jersey herd.
Nutritional changes can be expected to alter milk fat by about 1 percentage point, whereas changes in milk protein are limited to 0.1 to 0.3 percentage points. Thus, for producers at or below average, modifications to rations and feeding management could be used to improve production of fat and protein by one standard deviation. Ensuring that rations provide adequate energy and protein and balanced amounts of rapidly fermentable carbohydrate and effective fiber is the first step. Additionally, avoiding high levels of supplemental unsaturated fats and slug feeding of concentrates can help to increase both fat and protein in milk.
Management factors contribute about 45% of the variation in milk composition, and genetics explain 55% (Van Tassell et al., 1999). However, effects of changes in management practices are seen shortly after their implementation; genetic selection is a long-term approach to improving milk composition.
Milk yield is positively correlated to yield of milk fat and protein; however, milk yield is negatively correlated to milk fat and protein percentage (Welper and Freeman, 1992). Heritabilities for milk yield range from 0.32 to 0.40 for Holsteins and 0.38 to 0.48 for Jerseys (Van Tassell et al., 1999). Heritabilities for milk fat and protein yield range from 0.33 to 0.36 and 0.25 to 0.36 for Holsteins and 0.31 to 0.37 and 0.32 to 0.42 for Jerseys (Van Tassell et al., 1999). Given these relatively high heritability rates, producers could benefit from considering component production in their selection criteria, particularly if fat and protein production are below average.
On a herd level, more rapid genetic change could be achieved by purchasing cows with higher component production, including adding Jersey cows to a Holstein herd, adding Holsteins to a Jersey herd, or by cross-breeding. This study shows there is an economic balance between greater gains in milk production per cow and higher component levels in improving IOFC. With any of the above strategies, it is important to remember that total volume of components resulted in the highest IOFC. In the case of adding Jerseys to a Holstein herd, the addition of too many cows with milk production lower than the current herd average could negate any gains in higher component percentages. The reverse would be true for adding Holsteins to a Jersey herd to increase total milk volume; adding too many cows with low component percentages might offset the benefit of greater milk production. Similarly, a crossbreeding scheme that reduces milk production could negatively influence revenues. A more detailed analysis of the factors affecting an individual herd would be required to determine the levels of milk, milk fat, and protein that would result in the greatest increase in IOFC. Changes in the breed structure of a herd could also generate additional costs related to modifications of facilities and introduce new challenges for herd management that are not considered in a simple calculation of IOFC.
In all of the scenarios presented above, the most important factor affecting IOFC was the total amount of milk fat and protein produced. Although milk price is determined in part by the percentage of each component, it is the volume of milk components sold each month that results in higher levels of IOFC. Regardless of breed, increasing milk volume improved IOFC more than increasing component levels. Thus, a logical management strategy would be to set a target level for milk fat, protein, and daily milk production for each month of the year based on herd genetics, past performance, seasonality, and goals. If component levels or daily milk production drop below the target level, corrective action should be implemented quickly to minimize IOFC losses.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication September 20, 2004. Accepted for publication February 2, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Cho, T. R. Overton, C. G. Schwab, and L. W. Tauer Determining the Amount of Rumen-Protected Methionine Supplement That Corresponds to the Optimal Levels of Methionine in Metabolizable Protein for Maximizing Milk Protein Production and Profit on Dairy Farms J Dairy Sci, October 1, 2007; 90(10): 4908 - 4916. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
