# Documentation

ERS compares farm and retail prices for specific food items as well as for commodity groups. For commodities, individual foods are grouped into market baskets which contain a collection of foods representing what an American household may buy at retail during 1 year. The costs of market baskets at retail are compared with the prices received by farmers for a corresponding basket of agricultural commodities.

Whether calculating statistics for individual foods or for market baskets, ERS relies on conversion factors to specify the amount of farm commodities in a retail product and to account for coproducts made from these commodities where necessary.

The methodology behind this data series consists of three parts:

- How ERS Calculates Market Basket Statistics
- How ERS Calculates Farm Share for Individual Foods
- Conversion Factors Used in Calculating Farm Share Statistics

Additionally, ERS uses data from the Agricultural Baseline Database in the data visualization: Price Spreads and Food Markets. See the section on Data Visualization below.

## How ERS Calculates Market Basket Statistics

The process of generating market basket statistics has two steps:

- A market basket of foods is constructed to represent what U.S. households buy for at-home consumption over a 1-year period.
- Estimates of farm share and the farm-to-retail price spread (or difference between costs of the food basket at retail and the farm basket) are calculated using annual consumer and farm price data.

### Data

Key sources of data are the Consumer Expenditure Survey (CE), which includes a diary for recording purchases, and Nielsen Homescan data. The CE is produced by the Bureau of Labor Statistics (BLS), and the data are used by BLS to calculate expenditure weights for the Consumer Price Index (CPI).

Households participating in Nielsen's "Homescan" panel keep a record of their food purchases at retail stores using a scanner installed in their homes. Upon returning from a shopping trip, panelists re-scan purchased items or manually enter information for products lacking a bar code.

### Constructing Baskets

Retail food baskets are specified, and then ERS determines the farm products required to produce the contents of the retail food baskets. For example, the consumer basket for dairy products contains certain quantities of milk, cheese, and other foods. The corresponding agricultural basket contains enough farm milk to produce all of these foods. Conversion factors are used to calculate farm quantities as well as coproduct amounts. For a discussion of conversion factors, see Conversion Factors Used in Calculating Farm Share Statistics.

### Farm Share

A formal equation for the farm share of a retail basket at time t, FS_{t}, is: FS_{t} = Q'_{f}P_{ft} / Q'_{r}P_{rt}

where Q_{r} is a vector of food quantities bought by households and P_{rt} is a vector of unit retail prices for these same foods at time t. The arguments in the numerator include Q_{f} which is also a vector of quantities. It includes the agricultural goods used to make the foods in Q_{r} as well as any coproducts. Agricultural inputs enter Q_{f} as a positive number. Coproducts enter as a negative number. P_{ft} is a vector of unit prices for the agricultural goods and coproducts in Q_{f}.

### Calculating Annual Data Series

To facilitate the calculation of an annual data series, retail food baskets' contents are fixed over relatively long periods of time. Estimates of farm share denote the proportion of the consumer's dollar earned by agriculture for a basket of foods representing what American households bought at a particular period in time.

In reporting annual estimates, ERS assumes that retail food baskets are similar to the foods which BLS prices for calculating the CPI (U.S. city-average series). The value of a retail basket in year t can then be approximated as the product of its base year value, Q'_{r}P_{r0}, and an adjustment factor to account for retail price inflation. This adjustment factor is the ratio of the CPI for the commodity in question at time t, CPI_{t}, to the same CPI in the base year of the data series, CPI_{0}. ERS uses the following formula to approximate the farm share of a retail basket:

FS_{t} = Q'_{f}P_{ft} / (Q'_{r}P_{r0})(CPI_{t}/CPI_{0})

BLS publishes separate CPIs for major food groups. To estimate the farm share of fresh vegetables, for example, ERS uses the CPI for fresh vegetables. This approximation makes annual calculations more convenient to implement as researchers do not need to collect prices for individual foods at retail every year. Only the CPI for the food group is required to update the value of the retail basket.

Once published, market basket statistics may be revised because:

- Updated estimates of average prices received by farmers for their commodities are available (these updates may be available as much as a year or two after preliminary figures).
- Conversion factors may be adjusted. For example, improved efficiency in food packaging and shipping may reduce waste and spoilage. If so, members of the food marketing system could buy a smaller quantity from farmers to provide the same quantity at retail.

The following discussion focuses on the market baskets for dairy products, fresh fruit, and fresh vegetables.

### Dairy Products

ERS calculates the farm share as well as the farm-to-retail price spread for a basket of dairy products representing what U.S. households purchased in 2003 for at-home consumption. ERS provides estimates for each year since 2000.

#### Identifying a retail and an agricultural basket

*Retail food basket*

To identify the retail dairy basket, ERS uses the diary portion of the CE to determine how much money American households spent on several types of foods. On average, households participating in the 2003 diary spent about $328 for dairy products over the course of the year. These expenditures included $112.58 for fluid milk, $13.95 for fluid cream, $18.08 for butter, $97.04 for cheese, $58.09 for ice cream and related foods, and $28.26 for "other dairy products" like yogurt.

ERS uses data from Nielsen's Homescan panel to derive the quantities purchased and prices of products. In 2003, sample households divided their fluid milk expenditures among whole milk (26 percent), reduced-fat (54 percent), skim (20 percent), and buttermilk (under 1 percent). For each type of milk, the total amount households spent was divided by the number of gallons they bought. For example, reduced-fat milk sold for about $2.61 per gallon, on average. It was assumed that 54 percent of the $112.58 spent on fluid milk by CE households was for reduced-fat milk, or $60.29. Based on the per-gallon price, households bought 23.06 gallons of reduced-fat milk in 2003.

Spending on cheese was divided among four categories. The largest category included Cheddar, Colby, Swiss, Mozzarella, Muenster, and other relatively hard, natural cheeses. In total, these cheeses accounted for about 64 percent of the $97.04 that Nielsen households spent on cheese. Thus, CE households were assumed to have spent $61.66 for 17.02 pounds at $3.62 per pound.

The second category of cheese included in the food basket is Cottage cheese, accounting for about 8 percent of Nielsen household expenditures on cheese. CE households were estimated to have spent $7.68 for 4.99 pounds of low-fat Cottage cheese at $1.54 per pound.

Cream and Neufchatel comprise the third category of cheese in the retail food basket. These cheeses account for about 7 percent of Nielsen household expenditures on cheese. Thus, CE households were estimated to have spent $6.87 for 2.63 pounds of creamed cheese at $2.61 per pound.

Processed foods account for the balance of cheeses in the retail food basket. Natural cheeses can be blended together along with other ingredients to make processed cheese, processed cheese food, and processed cheese spread. In total, these three foods accounted for about 21 percent of Nielsen household expenditures on cheese products. CE households were estimated to have spent $20.83 for 7.77 pounds of these processed cheese products at $2.68 per pound.

Based on these estimation procedures, CE households would have bought 11.19 pounds of fluid cream, 33.91 pounds of ice cream, 4.9 pounds of ice milk, 1.6 pounds of sherbet, 8.5 pounds of butter, and 21.53 pounds of yogurt.

*The agricultural basket and its value relative to the retail food basket*

Once the contents of the retail food basket are specified, ERS estimates the contents of a corresponding agricultural basket. How much raw farm milk would be needed to supply the milkfat in the foods? To answer this question, several assumptions were made about the foods in the retail basket and conversion factors applied. The following assumptions are in addition to those outlined in Conversion Factors Used in Calculating Farm Share Statistics:

- The whole milk in the retail food basket contains, on average, 3.3 percent milkfat. By contrast, stores sell a variety of reduced-fat milk, such as 1-percent and 2-percent. The reduced-fat milk in the basket is assumed to have an average fat content of 1.5 percent, buttermilk is assumed to contain 1 percent, and skim milk 0.1 percent fat.
- Fluid cream is categorized according to the product's fat content. Heavy cream is at least 36 percent fat and half-and-half is between 10.5 and 18 percent fat. Fluid cream in the retail basket is assumed to have an average fat content of about 20 percent.
- The relatively hard natural cheeses that American households buy are assumed to contain as much milkfat as Monterey cheese. Fat solids from milk account for about 28 percent of the total weight of Monterey, nonfat solids account for another 28 percent, and moisture accounts for 44 percent of the product's weight. Natural cheeses are often classified by moisture content, which relates to their hardness. Monterey contains less fat per pound than harder cheeses like Cheddar and Colby, but it has more fat per pound than softer cheeses like Muenster and Mozzarella.
- Processed cheese food with a fat content of 23 percent was used to represent the amount of milkfat in all types of processed cheese products that American households buy. Processed cheese food contains less fat than processed cheese on a per pound basis, but more than processed cheese spread.

*Milkfat Estimates*

Combining the amount of milkfat in all products in the food basket yields a basket containing fat from about 740 pounds of milk. To produce the retail basket, manufacturers and processors are assumed to source slightly more milk than the 740 pounds needed to supply the fat contained in this basket. This combined amount was inflated by 2.5 percent, and it was assumed that they purchased 759 pounds (740 / 0.975). The additional 2.5 percent of milk accounts for the likelihood that some milk is wasted as dairy products are processed, packaged, and transported.

The 2.5 percent additional milk purchased also provides a cushion to account for variation in the composition of farm milk. The conversion factors used in this analysis assume all farm milk to be 3.7 percent fat and 8.62 percent nonfat solids on a per pound basis. In reality, the composition of milk varies. For example, the fat content of milk averaged 3.68 percent in 2006, 3.66 percent in 2005, and 3.67 percent in 2004. Variation in the fat and nonfat components affects the quantities of dairy products that can be produced.

The additional milk bought by processors and manufacturers is also needed to satisfy the small amount of milkfat contained in coproducts.

*Farm Value*

Farm receipts are estimated as the product of farm prices and the quantity of milk bought by processors. The average price received by farmers for all milk is reported monthly in Agricultural Prices, a publication of USDA's National Agricultural Statistics Service (NASS). ERS calculates a simple average of NASS's 12 monthly prices. In 2003, this average was $12.53 per 100 pounds of milk. Estimated farm receipts are thus $95 (759 x $0.1253).

ERS's retail basket, however, includes nonfat solids from only about 573 pounds of milk. It follows that nonfat solids from 167 pounds of milk (740 - 573) would remain for making coproducts. Not all of these solids are likely to be wasted; most are used for producing other foods even though these foods are not included in the retail food basket. It was assumed that nonfat solids from about 108 pounds of milk are recovered from cheese production and used to make 9.91 pounds of dry whey. It is also assumed that the skim solids from about 59 pounds of milk are used to make 5.25 pounds of nonfat dry milk.

The total value added to the 759 pounds of milk bought from farmers is the sum of the amount added to the parts consumed in producing the retail basket and the amount added to the parts consumed in making coproducts. To compare the retail price of the food basket with the farm value of only the milk parts consumed in making it, the farm value of any coproducts needs to be subtracted from farm receipts.

ERS uses wholesale prices for dry whey and nonfat dry milk to place an initial value on the coproducts. Monthly data published by USDA's Agricultural Marketing Service (AMS) are used to calculate a simple average of the 12 months for each coproduct. In 2003, wholesale prices averaged $0.80 per pound for nonfat dry milk and $0.17 per pound for dry whey.

Because wholesale prices include processing costs, these costs are subtracted from average wholesale prices to estimate the farm value of a pound of each coproduct using data published by AMS. In 2003, per pound processing costs were $0.14 for nonfat dry milk and $0.159 for dry whey.

The farm value of ERS's agricultural basket equals farm receipts less the farm value of coproducts. For 2003, farm value is estimated to be: $91.48 or 95 - (5.25 x (0.80 - 0.14)) - (9.91 x (0.17 - 0.159). Farm value can then be compared with the $328 spent for the retail basket. The spread between farm and retail prices is $328 - $91.48 = $236.57, and farm share is (91.48 / 328) = 27.89 percent.

Retail foods - product | Retail foods - pounds | Farm commodities - product | Farm commodities - pounds | Coproducts - product | Coproducts - pounds | |||
---|---|---|---|---|---|---|---|---|

Butter | 8.5 | Milk | 758.5 | Nonfat dry milk | 5.25 | |||

Buttermilk | 2.25 | Dry whey | 9.91 | |||||

Cottage cheese | 4.99 | |||||||

Cream | 11.19 | |||||||

Cream cheese | 2.63 | |||||||

Ice cream | 33.91 | |||||||

Ice milk | 4.9 | |||||||

Low-fat milk | 198.79 | |||||||

Monterey cheese | 17.02 | |||||||

Processed cheese food | 7.77 | |||||||

Sherbet | 1.6 | |||||||

Skim milk | 71.52 | |||||||

Whole milk | 89.19 | |||||||

Yogurt | 21.53 | |||||||

Note: Some numbers have been rounded. Source: USDA, Economic Research Service, Price Spreads from Farm to Consumer data product. |

#### Calculating annual estimates

Having specified the contents of the retail and agricultural baskets, ERS uses a formula to estimate the farm share of dairy food prices in years other than the base year of the data series, 2003. The following example shows implementation for 2005.

Step 1:

Estimation begins by plugging the BLS-calculated CPI for dairy and related products into the ERS formula previously specified. The CPI was 167.9 in 2003 and 182.4 in 2005. Since 2003 is the base year of the data series, CPI_{0} = 167.9. For t= 2005, it further follows that 1.086 = (182.4 / 167.9) which is an adjustment factor for retail price inflation.

Step 2:

The adjustment factor is next multiplied by the value of the retail basket in the base year (Q'_{r}P_{r0} ). The result is what CE households spent for dairy products in 2003: $328.

Step 3:

The value of the denominator in the farm share equation is estimated at ($328 x 1.086) = $356.33, which yields the approximate value of the retail basket in 2005.

Step 4:

Farm prices and coproduct values are next collected from Agricultural Prices; and AMS, respectively. For t = 2005, the prices in the numerator (P_{ft}) need to be updated from prior yearly values. However, neither the quantities (Q_{f}) nor the methods for calculation change from those used to estimate farm value for 2003. For t = 2005, using the updated values of P_{ft}, the updated farm value of the agricultural basket net coproducts, Q'_{f}P_{ft}, is $104.41.

Step 5:

Farm share is computed by dividing farm value by the value of the retail basket. As noted in Step 4, in 2005, farm value was $104.41, which amounts to about 29.3 percent of the approximate value of $356.33 for the retail basket.

Step 6:

ERS also reports indices for retail cost, farm value, and the farm-to-retail price spread. Each is expressed as an index with a value of 100 in the base year (2003) of the data series. These indices show how the variable in question has changed over time.

#### Retail Cost Index

The retail cost index shows retail price trends over time. This is the ratio of the CPI in year t to the same CPI in the base year (2003), CPI_{t} / CPI_{0}, multiplied by 100. For t= 2005, the retail cost index is (1.086 x 100) = 108.6.

#### Farm Value Index

The farm value index shows farm value trends over time. This is the ratio of the farm value in year t to the farm value in 2003, Q_{f}P_{ft} / Q_{f}P_{f0}, multiplied by 100. As already shown, farm value was $91.48 in 2003 and $104.41 in 2005. It follows that the 2005 farm value index was (($104.41 / $91.48) x 100) = 114.13 in 2005.

#### Farm-to-Retail Price Spread Index

The farm-to-retail price spread index measures changes in the difference between the basket's retail and farm values. This is a ratio of the spread in a year to the spread in the base year of the data series, (Q_{r}P_{rt} - Q_{f}P_{ft}) / (Q_{r}P_{r0} - Q_{f}P_{f0}), multiplied by 100. In 2005, the spread between retail and farm values was ($356.33 - $104.41) = $251.92. It was ($328 - $91.48) = $236.52 in 2003. The value of the farm-retail spread index was ($251.92 / $236.52) x 100 = 106.51 in 2005.

#### Interpretation of Farm Value and Supporting Indices

The three reported indices show trends in each variable over time. For example, farm share grew from 27.89 percent in 2003 to 29.3 percent in 2005. The three indices suggest a reason for this change. They show growth in both farm value and in the amount of value added to farm commodities as measured by the farm-to-retail price spread. The values of both indices exceeded 100 in 2005. However, the farm value index grew relatively faster (114.13 versus 108.6).

### Fresh fruit and vegetables

ERS calculates the farm share as well as the farm-to-retail cost spread for baskets of fresh fruit and fresh vegetables that are representative of U.S. household purchases for at-home consumption between 1999 and 2003. ERS provides estimates for each year since 1997.

#### Identifying a representative market basket

### Retail food baskets

ERS created food baskets to represent what households bought for at-home consumption in 1999 and 2003. The final retail baskets are an average of the 1999 and 2003 baskets.

ERS uses the Consumer Expenditure Survey to help determine the contents of retail food baskets. The diary portion of the CE reports how much money American households spent on several types of foods. For example, households spent, on average, $148.51 for fresh vegetables in 1999, including $18.92 for lettuce, $26.91 for tomatoes, $28.35 for potatoes, and $74.33 for "other fresh vegetables."

Because the CE does not contain data on prices or quantities purchased, ERS obtains this information for individual fresh fruits and fresh vegetables from Nielsen's Homescan panel. Households participating in this panel kept a record of their purchases at retail foodstores using a scanner installed in their home. After a shopping trip, panelists would re-scan purchased items or manually enter information on products lacking a bar code. The sample available for this study contains data for 7,200 households in 1999 and 8,833 households in 2003.

Based on the Nielsen data, a representative household, for example, split its 1999 CE expenditures of $18.92 for lettuce into $11.73 for iceberg and $7.19 for romaine ("iceberg" lettuce represents all purchases of head lettuce, which accounts for about 62 percent of the value of all lettuce purchased; "romaine" represents all purchases of leafy lettuce). Using Nielsen national average prices, ERS then estimates quantities purchased. For example, since iceberg lettuce averaged $0.78/lb., a representative household in 1999 could have bought 15.04 lb. with its $11.73.

A similar process was used to estimate quantities of fresh potatoes, fresh tomatoes, and "other fresh vegetables" purchased. To select items for inclusion in "other fresh vegetables", Nielsen data were used to rank foods by expenditure share. The top 12 foods are selected, and then the $74.33 spent by households in the CE on other fresh vegetables in 1999 is allocated among the selected 12 foods. For example, broccoli accounted for 7.46 percent of Nielsen panelist expenditures on the 12 other fresh vegetables, so 7.46 percent of the $74.33 spent on other fresh vegetables in the CE ($5.55) was allocated to expenditures on broccoli. Because the price of broccoli averaged $0.88 per pound at retail, it was further inferred that a representative household in 1999 bought 6.3 pounds.

ERS repeats the above procedures using data from 2003. The final market baskets for fresh fruit and fresh vegetables were created by averaging the contents of the 1999 and 2003 baskets.

### Agricultural Baskets

Once the contents of the fresh fruit and fresh vegetables retail food baskets are specified, ERS uses conversion factors to estimate the contents of their corresponding agricultural baskets. For each food in a retail basket, conversion factors inflate the retail quantity by the amount necessary to compensate for waste and shrinkage that occurs as goods are prepared for retail sales. For example, the market basket for fresh vegetables contains 21.11 pounds of carrots, and ERS estimates that farmers must supply 1.031 pounds of carrots for every 1 pound supplied by marketers at retail. Some carrots may spoil and, perhaps, others may need trimming.

See Conversion Factors Used in Calculating Farm Share Statistics

Vegetable | Retail quantity | Farm quantity |
---|---|---|

Asparagus | 2.00 | 2.19 |

Bell peppers | 6.47 | 7.04 |

Broccoli | 6.71 | 7.29 |

Cabbage | 7.51 | 8.08 |

Carrots | 21.11 | 21.76 |

Cauliflower | 2.18 | 2.37 |

Celery | 5.34 | 5.74 |

Corn on the cob | 4.38 | 4.76 |

Cucumber | 6.79 | 7.38 |

Iceberg lettuce | 15.37 | 16.53 |

Agaricus mushrooms | 3.12 | 3.32 |

Onions | 24.22 | 25.77 |

Potatoes | 82.92 | 86.37 |

Romaine lettuce | 7.97 | 8.57 |

Sweet potatoes | 4.67 | 5.19 |

Tomatoes | 20.91 | 24.60 |

Fruit | Retail quantity | Farm quantity |
---|---|---|

Apples | 34.07 | 35.49 |

Cantaloupe | 11.25 | 12.23 |

Cherries | 2.11 | 2.29 |

Grapefruit | 15.07 | 15.54 |

Grapes | 15.89 | 17.47 |

Honeydew melon | 1.80 | 1.96 |

Kiwifruit | 0.91 | 1.00 |

Lemons | 6.01 | 6.27 |

Oranges | 25.02 | 25.79 |

Peaches | 8.87 | 9.43 |

Pears | 3.87 | 4.07 |

Plums | 2.46 | 2.59 |

Strawberries | 8.27 | 8.99 |

Watermelon | 19.75 | 21.95 |

#### Calculating an annual price series

Having identified the contents of retail food and agricultural baskets, ERS estimates annual statistics by using information on consumer and farm prices as well as the formula previously specified.

When calculating statistics for fresh vegetables, for example, the value of the denominator in the formula is estimated by using the CPI for that commodity. In 2001, the CPI for fresh vegetables was 230.6. To estimate the retail price of the market basket in, say 2005, the retail value of the market basket in the base year ($162) is multiplied by the appropriate CPI for that year (271.1), divided by its 2001 value (230.6). That is, $162 x (271.7 / 230.6) = $190.87.

ERS uses prices received by farmers for their commodities to update the value of the farm basket for each year of the data series. In 2005, the total value of all the contents of the fresh vegetables farm basket was $48.32, which amounts to about 25 percent of the estimated price of $190.87 for the retail basket.

See also:

How Low has the Farm Share of Retail Food Prices Really Fallen?## How ERS Calculates Farm Share for Individual Foods

ERS estimates the farm share of the retail price for selected foods. These calculations compare the retail price of a food with the farm value of the commodities used to manufacture it. Data on prices at retail and at the farm gate are needed to make these comparisons. ERS results are sensitive to the prices adopted for analysis. Because two different stores may sell the same food for different prices, the farm share of a consumer's dollar would not likely be the same at both stores.

### Data

For retail prices, ERS obtains data on national average prices from the Bureau of Labor Statistics (BLS), the Florida Department of Citrus, and the National Consumer Panel (NCP). Two market research companies—Information Resources, Inc. and Nielsen—jointly maintain the NCP, a panel of households that is representative of the continental United States both demographically and geographically. Participating households are given a scanner to keep at home. After a shopping occasion, panelists use these scanners to record their purchases, including the quantities bought as well as the names and locations of the stores where purchases were made.

For prices at the farm gate, ERS relies primarily on data published by ERS—and other USDA agencies such as the National Agricultural Statistical Service (NASS)—and the Agricultural Marketing Service (AMS). These data are available in publications such as Agricultural Prices or from the agencies' websites.

Conversion factors specify how much of a farm commodity is used in a food's manufacture as well as the amount of coproducts that are produced along with this food. For example, conversion factors specify how much milk farmers supply for each pound of Cheddar cheese sold by marketers. Conversion factors also state how much dry whey can be made with the nonfat milk solids from this milk that are not consumed in making the cheese.

### Formula

The formal equation used to calculate the farm share of different foods is: FS_{t} = Q'_{f}P_{ft} / P_{rt}where FS_{t} is the farm share at time t. The argument in the denominator of the formula is the retail price of the food at time t, P_{rt}. The first argument in the numerator, Q_{f}, is a vector of quantities that includes amounts of agricultural goods used to make the food as well as amounts used to make coproducts. Agricultural inputs enter Q_{f} as a positive number. Coproducts enter as a negative number. P_{ft} is a vector of unit prices for the agricultural goods and coproducts in Q_{f}.

Once published, farm share statistics may be revised for a variety of reasons:

- Updated estimates of prices received by farmers for their commodities become available as much as a year or two after preliminary figures.
- Conversion factors may be adjusted. For example, improved efficiency in food packaging and shipping may reduce waste and spoilage. If so, food marketers could buy a smaller quantity from farmers to provide the same quantity at retail.

ERS's processes for estimating the farm share of a food are illustrated for selected dairy products, fresh fruits, fresh vegetables, processed fruit and vegetables such as orange juice, and products like flour and sugar that are processed from field crops.

### Dairy Products

#### General

ERS's calculations for whole milk, ice cream, and Cheddar cheese rely on U.S.-city average retail price data from BLS; on the other hand, calculations for butter rely on average retail price data from the NCP. In 2018, the monthly price of a gallon of whole milk ranged from a low of $2.84 in July to a high of $2.96 in January. For this data series, a simple average of monthly average prices is calculated—for 2018, the average is $2.90.

To estimate the farm value of whole milk, butter, and ice cream, ERS uses data generated by AMS from the Federal milk order program. This program sets minimum prices for fluid-grade milk. Not all milk is priced at the same level. Instead, there is a classified pricing system in which the minimum amount paid for milk is determined by how the milk is used. There are four classes: Class I is defined as milk for beverage products; Class II includes milk used to make fluid cream, yogurt, ice cream, and other perishable foods; Class III includes creamed and hard cheeses; and Class IV products are butter and dried milk.

Because these estimates rely on data generated through the administration of the Federal milk order program, ERS's farm share estimates for whole milk, butter, and ice cream do not reflect conditions in all parts of the United States. The current Federal milk order program consists of 11 Federal orders. For example, parts of New England and the Mid-Atlantic States make up the Northeast Order. California's Federal Milk Marketing Order officially began operations in 2018. Some other places, however, are not covered by a Federal order.

For all four dairy products considered here—whole milk, ice cream, Cheddar cheese, and butter—ERS's objective is to estimate the contribution from dairy farmers' milk earnings to retail prices. Estimates of farm value and farm share do not include farm receipts for other types of ingredients included in dairy products, such as sugar or fruit contained in ice cream.

ERS statistics for whole milk, ice cream, and Cheddar cheese also do not account for over-order payments. Dairy farmers and their cooperatives can bargain to sell milk at prices higher than the regulated minimums. For example, fluid milk processors and dairy product manufacturers may be willing to pay a premium to receive a certain quality and quantity of milk, at a specified time and location. Dairy farmers may or may not receive higher prices for their milk in these cases. However, to the extent that cooperatives and independent producers with market power can negotiate over-order payments that increase payments to dairy farmers, ERS's calculations based on minimum class prices would underestimate the farm value of dairy products.

ERS calculates the farm share of the retail price of:

#### Milk, whole fluid

Though marketed as "whole milk," farm milk generally has a small amount of cream removed during processing. A gallon of whole milk is assumed to contain 3.3 percent fat and weigh 8.6 pounds. ERS's process for calculating the farm share of this product is illustrated using data for 2018.

Step 1:

The retail price of a gallon of whole milk averaged $2.90 in 2018. Farm share is the ratio of average retail price to the amount received by farmers for milk components (skim and fat solids) consumed in production.

Under the Federal milk order program, processors must pay at least the Class I price for farm milk used in making beverage products.

Step 2:

To estimate the farm value of whole milk, ERS first estimates the farm value of one component—Class I skim milk. To begin, ERS calculates a simple average of the monthly base skim milk price reported by AMS—$6.23 per hundredweight in 2018. Then, the required Class I differential is added to the average skim milk price. AMS adds this differential to the base skim milk price to generate the minimum amount of money processors must pay for Class I skim milk. Though the value of the differential varies geographically, a principal pricing point is reported by AMS for each of the 11 Federal orders. ERS calculates a weighted average of the reported figures. These weights are the amount of Class I milk sold in the order divided by the amount marketed in all 11 orders. In 2018, this weighted average was about $2.84 per hundred pounds of skim milk.

Step 3:

To estimate minimum farm value for the butterfat in whole milk, ERS uses the monthly advanced butterfat pricing factor reported by AMS. The Class I differential is added to the pricing factor. In 2018, this pricing factor averaged $2.52 per pound, and the Class I differential was $0.0284 per pound.

Step 4:

Combining the above values for the skim and fat solids in milk, ERS next calculates the farm value of 100 pounds of Class I milk that is 96.7 percent skim milk and 3.3 percent fat. If purchased at minimum regulated prices, the milk has a value of $17.19 or (6.23 + 2.84) x (0.967) + ( $2.52 + 0.0284) x (3.3).

A gallon of milk is assumed to weigh 8.6 pounds, but processors purchase slightly more farm milk for every gallon of milk marketed at retail. This extra amount accounts for waste and spoilage that can occur in assembling and processing milk. It is assumed that losses through waste and spoilage equal 2 percent of the value of what farmers sell to marketers. Thus, in 2018, marketers were assumed to pay farmers (($17.19 x 0.086) / 0.98) = $1.51 for each gallon of milk marketed at retail, or 52 percent of the retail price of $2.90.

#### Butter

Cream is the primary ingredient in butter. Butter manufacturers may buy cream with less than the full fat content of farm milk or purchase farm milk and skim off the cream. One pound of butter is assumed to include 0.803 pounds of fat and 0.01 pounds of skim solids. ERS's process for calculating the farm share of the retail price of butter is illustrated using data for 2013.

The retail price of butter averaged $2.70 per pound in 2013. This estimate is based on the price paid by NCP households for boxed, salted, regular (not whipped or churned) butter. The farm share of a pound of butter is the ratio of this average retail price to the price farmers received for the milk components—skim and fat solids—used in making this type of butter.

Under the Federal milk order program, processors must pay at least the Class IV price for milk used to make butter.

Step 1:

To estimate the farm value of a pound of butter, ERS begins with the Class IV price of skim milk. In 2013, the regulated minimum Class IV skim milk price averaged $13.71 per hundredweight, or just under $0.14 per pound. How much of this milk must manufacturers purchase to acquire the 0.01 pounds of skim solids in a pound of butter? Since 1 pound of this type of milk contains about 0.09 pounds of skim solids, manufacturers must buy 0.11 pounds (0.01/0.09).

Step 2:

Manufacturers also must buy 0.803 pounds of fat for every pound of butter they make. ERS uses the AMS data series to determine the farm value of 0.803 pounds of fat from milk. In 2013, the Class IV butterfat price averaged about $1.66 per pound.

Step 3:

Combining the above values for skim and fat solids in a pound of butter, ERS then calculates the farm value of cream, assuming that skim and fat solids were purchased at regulated minimum prices. This estimate is (0.803 x $1.66) + (0.11 x $0.1371) = $1.35.

Step 4:

ERS makes adjustments to account for waste and spoilage that may occur as butter is produced. These losses are assumed to equal 2 percent of what farmers sell. Thus, in 2013, to cover for waste and spoilage, manufacturers must buy cream with a farm value of $1.38 ($1.35 / 0.98) for each pound of butter marketed at retail.

Step 5:

ERS estimates the farm share of the retail price of butter in 2013 to be 51 percent ($1.38 / $2.70).

#### Cheddar cheese

Fluid milk is curdled to make natural Cheddar cheese. ERS uses the Van Slyke formula to determine the amount of milk in a pound of this cheese. Assuming that the milk has an average fat content of 3.5 percent, the Van Slyke formula indicates that manufacturers must purchase 10.3 pounds of milk for every pound of Cheddar cheese made. Along with cheese, manufacturers also produce 0.5 pounds of the coproduct dry whey.

ERS’s procedure for calculating the farm share of a pound of Cheddar cheese is illustrated using data for 2011, when the retail prices averaged $5.42 per pound.

Under the Federal milk order (FMO) program, processors must pay at least the Class III price for milk used to make cheese.

Step 1:

ERS collects minimum monthly prices for milk manufactured into Class III products under the Federal milk order program. A simple mean of these monthly prices is calculated to estimate annual prices. In 2011, cheese manufacturers were required to pay, on average, at least $18.37 per hundredweight of Class III milk (or $0.1837 per pound).

Step 2:

Given that manufacturers must purchase 10.3 pounds of milk for every pound of Cheddar cheese they make, ERS estimates that gross farm receipts are about $1.89 per pound of Cheddar cheese produced ($0.1837 x 10.3).

Step 3:

Monthly dry whey prices published by NASS/USDA are then used to value coproducts from cheese production. In 2011, NASS reported that monthly average prices for dry whey ranged from $0.3789 to $0.6538 per pound. These wholesale prices, however, reflect both the value of the milk components in whey, and the value added to these components through processing. ERS uses AMS's reported estimates of processing costs to isolate the value for only the milk components. In 2011, it cost $0.1991 to process 1 pound of whey, and the monthly average cost of dry whey (net processing costs) was $0.3333.

Step 4:

Altogether, the net farm value of a pound of Cheddar cheese in 2011 was about $1.72, calculated as $1.89 - (0.5 x $0.3333), which equals 32 percent of the retail price of $5.42.

#### Ice cream, regular

Ice cream is made from a mix that includes fluid milk and cream. Ice cream plants may buy these inputs from fluid milk processors. ERS estimates farm share of the retail price for 1 half-gallon of regular ice cream weighing 2.25 pounds. Fat solids account for 12 percent and skim solids account for 10 percent of the product's weight. Thus, regular ice cream contains 0.27 (2.25 x 0.12) pounds of fat from milk and 0.225 (2.25 x 0.1) pounds of skim solids.

ERS's process for calculating the farm share of the retail price of regular ice cream is illustrated below using data from 2013, when retail prices averaged $4.99 per pound.

Under the Federal milk order program, marketers must pay at least the Class II price for milk used in ice cream.

Step 1:

ERS first estimates the farm value of skim milk used in producing ice cream. Since 1 pound of skim milk contains about 0.09 pounds of skim solids and ERS's ice cream product contains 0.225 pounds, marketers must buy 2.5 (0.225/0.09) pounds of skim milk for each container of ice cream they make. To put a value on this milk, ERS calculates a simple average of the monthly Class II skim milk price reported by AMS—for 2013, the average is $14.07 per hundredweight, or $0.1407 per pound.

Step 2:

Marketers also must buy 0.27 pounds of fat from milk for every container of ice cream produced. To determine the farm value of this fat, ERS again uses the AMS data series. In 2013, the Class II butterfat price averaged about $1.67 per pound.

Step 3:

Combining the above values for the skim and fat solids in a container of ice cream, ERS then calculates the farm value of the cream under the assumption that skim and fat solids were purchased at regulated minimum prices. This estimate is $0.80, or ((0.27 x $1.67) + (2.5 x $0.1407)).

Step 4:

Adjustments are next made to account for waste and spoilage that tend to occur as ice cream is made. ERS assumes that these losses equal 2 percent of what marketers buy. Thus, in 2013, to cover for waste and spoilage, marketers must buy cream with a farm value of $0.82 ($0.80 / 0.98) for each container of ice cream marketed at retail.

Step 5:

ERS estimates that the farm share of the retail price of ice cream in 2013 was 16 percent ($0.82 / $4.99).

### Field Crops

#### General

ERS's calculations for flour, vegetable (soybean) oil, and sugar rely on national average retail prices (U.S. city-average price data) published by BLS. Annual figures are estimated by taking a simple average of the reported monthly prices.

For prices at the farm gate, publicly available data published by ERS, NASS, and AMS are used.

Coproducts account for a substantial share of overall farm receipts for flour, vegetable (soybean) oil, and sugar. For example, wheat is milled to produce flour, and its coproducts—bran and wheat middlings—may be sold for use in food products or animal feed. Similarly, sugar extracted from sugar beets and sugarcane yields coproducts such as molasses and livestock feed.

Data on coproduct prices are obtained mostly from USDA agencies, including NASS. For sugar, data from the Census Bureau are used. ERS calculates the farm share of the retail price of one pound of:

#### All-purpose white flour

Wheat is milled to produce flour. Wheat kernels, also known as wheat berries, are seeds from which new wheat plants may grow. Within a kernel are endosperm and wheat germ. Roller milling involves crushing and gradually reducing wheat kernels to produce flour which consists primarily of endosperm. Coproducts of flour milling include bran (the hard, outermost shell of the wheat kernel) and wheat middlings (bran, germ, and endosperm remnants). These coproducts may be sold for use in food products or animal feed.

Milling different classes of wheat yields different types of flour. Many products, including all-purpose white flour, may include several classes of wheat (depending on the region of the country where the flour is produced). However, all-purpose flour is primarily produced from hard red winter (HRW) wheat. For the purposes of computing the farm value and farm share of the retail price of all-purpose flour, we assume that it is made from 100 percent HRW.

Step 1:

Monthly retail prices for all-purpose white flour, per pound, are obtained from BLS, published in conjunction with the BLS Consumer Price Index reporting program. In 2008, the monthly price of all-purpose white flour ranged from a low of $0.421 per pound in January to a high of $0.544 per pound in July. For this data series, a simple average of monthly average prices is calculated—for 2008, the average is $0.5067.

Step 2:

What are the farm receipts for the amount of wheat used to make one pound of flour? Milling wheat yields approximately 73 percent flour and 27 percent coproducts, so producing 1 pound of flour requires 1.37 pounds of wheat (1 / 0.73). Monthly data on the farm price per bushel of hard red wheat are available from Agricultural Prices (published by NASS). In 2008, the monthly farm price of hard red winter wheat averaged $7.6642 per bushel. Since one bushel weighs 60 pounds and 1.37 pounds of wheat are required, the reported price is divided by 43.8 (60 / 1.37) to estimate farm receipts. In 2008, the gross farm value of a pound of all-purpose white flour was $0.175.

Step 3:

A mix of bran and middlings is produced in conjunction with the flour, and the value of these coproducts must be subtracted from the gross farm value. Prices of these two coproducts are very similar, and, indeed, often identical. Because Kansas City primarily mills hard red winter wheat, the price of middlings per ton for Kansas City is used to calculate the aggregate coproduct value. This information is obtained from the Livestock and Grain Market News Portal, published by AMS, using the midpoint of the reported low and high bid levels. In 2008, the monthly price of middlings in Kansas City averaged $134.44 per ton. The price is divided by 2000 to obtain the coproduct price per pound—$0.0672 per pound.

Step 4:

What amount of coproducts is produced along with a pound of flour? Since the wheat-to-coproduct yield is 27 percent, about 0.37 pounds of coproducts can also be produced from 1.37 pounds of wheat that millers buy from farmers to produce a pound of flour (1.37 x 0.27 = 0.37). To estimate net farm value, the middlings price is multiplied by 0.37 and this product subtracted from the gross farm value. For 2008, net farm value is $0.175-(0.37 x $0.0672) = $0.1501.

Step 5:

The farm value share is determined by dividing the net farm value by the retail price. The farm value share for 2008 is $0.1501 /$0.5067=29.63 percent.

#### Vegetable (Soybean) Oil

Soybeans are a major oilseed and widely used to manufacture vegetable oil. Crude vegetable oil extracted from soybeans must be further refined for human consumption.

ERS assumes an 8-percent loss from refining crude oil (*Agricultural Handbook* No. 697, table 34, pg. 44). Thus, in order to supply 1 pound of refined oil, manufacturers need about 1.09 pounds (1 ÷ 0.92) of crude oil. One gallon of refined oil is further assumed to weigh 7.7 pounds. ERS estimates the farm value of one gallon of vegetable oil based on farm receipts for the soybeans purchased from farmers to produce this amount of vegetable oil. This value is then adjusted for the value of coproducts and, finally, compared with an estimate of national-average retail prices.

Data from 2016 are used in the example for vegetable oil.

Step 1:

Using IRI InfoScan data, ERS estimates that consumers paid $6.51 per gallon of soybean oil at retail stores in 2016.

Step 2:

The farm-level price of soybeans is available in *Agricultural Prices* which is published monthly by the National Agricultural Statistics Service (NASS). These figures are reported on a dollar-per-bushel basis. In August 2016, soybeans cost $9.93 per bushel or, equivalently, $0.17 per pound ($9.93 ÷ 60).

Step 3:

Crushing soybeans produces three products—oil, meal, and hulls. The amount of oil extracted from a bushel of soybeans (oil yield) is obtained from table 9 of the ERS *Oil Crops Yearbook*—11.68 pounds in August 2016.

Step 4:

The next step is to determine the share of oil in a bushel of soybeans. To do so, ERS divides the figure in Step 3 by the total weight of a bushel of soybeans in pounds. Using August 2016, for example, divide 11.68 by 60 to determine that 19.47 percent of a bushel of soybeans was allocated to soybean oil production.

Step 5:

The number of pounds of soybeans needed to produce 1.09 pounds of crude (1 pound of refined) soybean oil in August 2016 can now be estimated as 5.6 (calculated as 1.09 ÷ 0.1947).

Step 6:

The number of pounds of soybeans needed to produce 1 gallon (7.7 pounds) of refined soybean oil in August 2016 can now be estimated as 43 (calculated as 5.6 x 7.7).

Step 7:

Gross farm value is calculated by multiplying the U.S. farm price of soybeans in pounds (obtained in Step 2) by the number of pounds of soybeans required to make one gallon of refined soybean oil (Step 6). For August 2016, multiply $0.17 by 43 to obtain a gross farm value of $7.12.

Step 8:

The percentage of the value of a bushel of soybeans that is attributed to coproducts (soybean meal and hulls) is next obtained from table 9 of the *Oil Crops Yearbook*. In August 2016, it is reported that meal and hulls accounted for 68 percent of the total value of a bushel of soybeans. When this share is multiplied by the gross farm value (Step 7), the result is $4.83 for the value of coproducts in 43 pounds of soybeans.

Step 9:

Net farm value is calculated by subtracting the value of coproducts from the gross farm value. In August 2016, ERS estimates the net farm value of one gallon of soybean oil to be $7.12 - $4.83 = $2.29 per pound.

Step 10:

After repeating steps 2 through 9 for each month of a year, ERS calculates a simple average of the 12 monthly farm value estimates. For 2016, that average is $2.25.

Step 11:

The farm value share is the net farm value of refined soybean oil divided by the average retail price. For 2016, ERS's estimate is 34.5 percent (100 x ($2.25 ÷ $6.51)).

#### White (refined) sugar

Because sugar sold in retail stores in the United States may have been extracted from either sugar beets, sugarcane, or both, a weighted average is used of the farm values of refined sugar from both sources. Moreover, firms processing sugar beets or sugarcane may produce coproducts such as molasses and livestock feed.

Quantities of coproducts obtained along with sugar are estimated using data published by the Census Bureau for 1977-2007. In the Census of Manufactures, the Census Bureau publishes the value of both the sugar and the coproducts—molasses and livestock feed manufactured (Table 6a, "Products Statistics"). These data are used to estimate the share of industry shipments attributable to coproducts. Before 1997, the SIC Code was used for data from two industries—Raw Cane Sugar (SIC 2061) and Beet Sugar (SIC 2063). Since 1997, the NAICS system is used, relying on data from two industries—Sugarcane Products (NAICS311311) and Beet Sugar (NAICS 311313).

Regardless of the industry or classification system, the same procedures are used to develop coproduct estimates. First, the total value of industry shipments is estimated as the value of sugar and coproduct shipments, excluding only the value of shipments not specified by kind (n.s.k.). Next, the value of only coproduct shipments is summed. Finally, the sum of the coproduct shipments is divided by the value of total industry shipments. Mathematically,

CR=C/(T-NSK), where:

CR=coproduct ratio for a given industry

C=total coproduct value of shipments for a given industry

T=Total value of shipments for a given industry

NSK=Total value of shipments not specified by kind

The share of the value of sugar (cane or beet) processing attributed to coproducts appears to have changed only a little over the last 30 years. The share ranged from 4.84 percent to 7.48 percent for cane sugar processing and from 7.47 percent to 13.04 percent for beet sugar processing. In years where data were incompletely reported (specifically, 2002 for sugar beets 2007 for sugarcane), the coproduct ratios used in sugar computations were determined by averaging the estimated coproduct ratios for each Census from 1977 to 2007, for years and industries where data were completely reported. The average coproduct ratios were 6 percent for sugarcane and 10 percent for sugar beet processing.

Step 1:

The retail price per pound of bagged sugar (all sizes) is obtained from the average retail price series published by BLS. These figures are produced in conjunction with the monthly Consumer Price Index (CPI) reporting program. Annual estimates are calculated by taking simple averages of the monthly data. In 2007, this figure came to $0.51 per pound.

Step 2:

The annual average price of U.S. sugarcane (per ton) is obtained from Table 13 ("Sugarcane for sugar: price per ton, by State") of the Sugar and Sweeteners Yearbook (SASY). In 2007, the annual average price received by farmers was $29.40 per ton (2000 pounds) or, $0.0147 per pound.

Step 3:

In 2007, total production of sugarcane for making sugar was 28.273 million tons. The data are obtained from SASY Table 15 (U.S. sugarcane area, yield, production, sugar output, recovery rate, and sugar yield per acre).

Step 4:

Total production of raw cane sugar cane also comes from Table 15 of SASY. In 2007, this figure was 3.454 million tons.

Step 5:

The amount of raw cane sugar extracted per pound of sugarcane is estimated by dividing the figure in Step 4 by the figure in Step 3. In 2007, manufacturers extracted 3.454 / 28.273 = 0.1222 pounds of cane sugar (on average) from each pound of sugarcane processed.

Step 6:

USDA's Conversion Factors and Weights and Measures (Statistical Bulletin No. 616, March 1979) reports that refining 1.07 pounds of raw sugar produces one pound of refined sugar. This estimate is divided by the figure in Step 5 to further obtain an estimate of the quantity of sugarcane required to produce one pound of refined cane sugar. In 2007, food processors bought an estimated 8.756 (1.07 / 0.1222) pounds of sugarcane for each pound of cane sugar they produced.

Step 7:

The gross farm value per pound of cane sugar is estimated by multiplying the figure in Step 6 by the farm price of a pound of sugarcane (Step 2). For 2007, this value is 8.756 x $0.0147 = $0.1287.

Step 8:

Using data published by the Census Bureau, coproducts are estimated to account for 6 percent of the farm value of sugarcane. To estimate the net farm value of one pound of cane sugar, the gross farm value obtained in Step 7 is multiplied by 0.94 (1 - 0.06). In 2007, the net farm value of one pound cane sugar was about $0.12.

Step 9:

The sequence of steps employed for sugarcane is now repeated for sugar beets. First, the U.S. sugar beet price received by farmers is obtained from SASY Table 12 ("Sugar beet price per ton, by State and United States"). In 2007, the annual average price received by farmers was $42 per ton (2000 pounds) or, $0.021 per pound.

Step 10:

Sugar beet production is reported in SASY Table 17 ("U.S. sugar beet area, yield, and production"). In 2007, sugar beet production totaled 31.834 million tons.

Step 11:

The quantity of sugar obtained from sugar beets on a raw value basis is also recorded in SASY Table 17, and was 4.721 million tons in 2007.

Step 12:

The procedure for calculating the yield for sugar beets is analogous to that used for sugar cane. Here, the figure obtained in Step 11 is divided by the figure reported for Step 10. In 2007, manufacturers extracted 4.721 / 31.834 = 0.1483 pounds of beet sugar from each pound of sugar beets processed.

Step 13:

Sugar yield on a refined basis is then calculated using the same conversion factor (1.07) used for sugarcane. Thus, divide 1.07 by the result obtained in Step 12. In 2007, food processors bought an estimated 1.07 / 0.1483 = 7.2152 pounds of sugar beets for each pound of beet sugar produced.

Step 14:

The gross farm value per pound of beet sugar is estimated by multiplying the figure in Step 13 by the farm price of a pound of sugar beets (Step 9). For 2007, this value is 7.2152 x $0.021 = $0.1515.

Step 15:

Using data from the Census Bureau, coproducts are estimated to account for 10 percent of the farm value of sugar beets. To estimate the net farm value of one pound of beet sugar, we multiply the gross farm value obtained in Step 14 by 0.90 (1 - 0.10). In 2007, the net farm value of one pound beet sugar was about $0.14.

Step 16:

The final step in calculating the farm value of sugar is to take a weighted average of net farm values of refined cane and beet sugar. These weights are the percentages of U.S. sugar production derived from sugarcane and sugar beets, both of which are obtained from SASY Table 16 ("U.S. beet and cane sugar production"). In 2007, the weighted average, net farm value of a pound of refined sugar at a retail store is (0.421 x $0.12) + (0.579 x $0.14) =$0.13.

Step 17:

Finally, the farm value share is calculated by dividing the net farm value determined in Step 16 by the retail price of sugar reported in Step 1. For 2007, farm share was 0.13 / 0.51 = 0.25 or, 25 percent.

#### White pan bread

White pan bread is baked in long, narrow pans, and many manufacturers typically slice and package this type of bread in a plastic bag. While each manufacturer may use a different recipe (formula), ERS bases its calculations on a representative formula made available through the Department of Grain Science and Industry at Kansas State University. The farm-derived ingredients in 1 pound of white pan bread include 0.5964 pounds of flour, 0.0119 pounds of soybean oil, and 0.0508 pounds of corn syrup (dry weight). ERS assigns a zero farm value to the bread’s other ingredients that include water, yeast, salt, emulsifiers, and calcium propionate.

Step 1:

Retail stores commonly sell white pan bread in loaves ranging from 16 to 32 ounces. In conjunction with its reporting of the Consumer Price Index, BLS publishes monthly average prices at retail stores on per pound basis. Each month’s price is a broad average that is derived from prices for all brands and loaf sizes. For this data series, a simple average of the monthly average prices is calculated—for 2012, the average is $1.4177 per pound.

Step 2:

The farm value of each ingredient is estimated. White, all-purpose flour is the primary farm-derived ingredient; the data and procedures used to calculate its farm value are detailed above (all-purpose white flour). In 2012, that value was $0.1364 per pound. Since each pound of bread is assumed to contain 0.5964 pounds of flour, ERS estimates that the farm value of the flour in the retail product was $0.0814 (calculated as 0.5964 x $0.1364).

Step 3:

To estimate the farm value of the soybean oil in bread, ERS adapted the procedures described above for estimating the farm value of stick margarine. Again, it is assumed that a 4-percent loss occurs from refining crude oil (vegetable oil, as extracted from oilseeds). In order to supply one pound of refined oil, manufacturers therefore need 1.04167 pounds (1.00 / 0.96) of crude oil. In 2012, the farm value of one pound of refined soybean oil was $0.4654. The farm value of the oil in one pound of white pan bread was likewise $0.0055 (calculated as 0.0119 x $0.4654).

Step 4:

Data and procedures used to calculate the farm value of corn syrup are detailed in Table 31b of ERS's Sugar and Sweeteners Yearbook. In 2012, the farm value of corn syrup was $0.1238 per dry pound. Thus, the farm value of the corn syrup in one pound of white pan bread was $0.0063 (calculated as 0.0508 x $0.1238).

Step 5:

The farm value of all three farm-derived ingredients is next totaled. In 2012, the farm value of white pan bread was $0.0814 + $0.0055 + $0.0063 = $0.0932.

Step 6:

The farm share of the retail price is determined by dividing the farm value by the retail price. The farm value share for 2012 is $0.0932 / $1.4177 = 0.0657, or about 7 percent.

### Fresh Fruit

ERS estimates the farm share of seven types of fresh fruit—apples, grapefruit, grapes, lemons, oranges, pears, and strawberries. To illustrate the process behind the calculations, consider apples in 2006 when the retail price of red delicious apples averaged $1.07 per pound. Farm prices averaged $0.30 per pound. Assuming that 4 percent of farm commodity volume is lost as farm apples make their way through the marketing chain, farmers must supply 1.04167 pounds (1/0.96) of apples per pound sold by marketers. The farm share equation is: (1.04167 x 0.30) / 1.07 = $0.29.

Although the values for conversion factors, retail prices, and farm prices vary for other types of fresh fruit, ERS follows the same process for calculating farm share.

The farm share statistics presented here are reproduced from the Fruit and Tree Nuts Situation and Outlook Yearbook, which is published annually by ERS.

### Fresh Vegetables

ERS estimates the farm share of 4 types of fresh vegetables—broccoli, a head of iceberg lettuce, potatoes, and field-grown tomatoes. For iceberg lettuce, the retail price averaged $0.87 per pound in 2005. Farm prices averaged just under $0.16 per pound. Under the assumption that 7 percent of farm commodity volume is lost as the lettuce proceeds through the marketing chain, farmers must supply 1.075 pounds (1 / 0.93) of lettuce per pound sold by marketers. Farm share is thus (1.075 x 0.16) / 0.87 = $0.19.

Although the values for conversion factors, retail prices, and farm prices vary for the other three types of fresh vegetables, the same process for calculating the farm share is followed.

The farm share statistics presented here are reproduced from the Vegetables and Melons Tree Nuts Situation and Outlook Yearbook, which is published annually by ERS.

### Processed Fruit and Vegetables

#### Orange Juice

Orange juice is extracted from a variety of fresh oranges such as Valencia and Navel. Most of the weight of juice ready for consumption is from water. The soluble solids in orange juice consist primarily of sugars, with some citric acid and minerals; these soluble solids are measured in degrees brix. Fresh squeezed juice averages about 11.8^{o} brix, meaning that the juice is 11.8 percent soluble solids. The brix level of frozen concentrate (FC) sold at retail stores is at least 41.8^{o} brix; however, after adding three containers of water per container of concentrate, the resulting consumable juice is also 11.8^{o} brix. The same quantity of fresh oranges is needed to produce fresh-squeezed (not frozen concentrate or NFC) juice as frozen concentrate.

Fresh oranges are cleaned and graded at juice processing plants; then the juice is extracted. A processor making NFC products may pasteurize and bottle the extracted juice in a glass container, paperboard carton, or other consumer-ready package. For FC products, most of the water is removed and the concentrated juice is chilled. This type of concentrate is commonly stored at 65^{o} brix and may be separated into different tanks according to the variety of orange used in production. The processor may later blend concentrate from various tanks according to customer specifications. Before shipping to retailers, the processor may also add filtered water to reduce the brix level to 41.8^{o}.

Coproducts such as citrus pulp pellets (mostly cattle feed), D-limonene, and molasses can be made from the peel, pulp, and seeds that remain after producing orange juice. During the 2007-08 season when prices for citrus pulp pellets were low, industry executives stated that making coproducts was merely the least expensive way to dispose of leftover orange parts. A study by the University of Florida estimated that the value of juice (including orange and grapefruit) produced in the State in 2007-08 was $3.45 billion and the value of coproducts was $136 million. However, due to the paucity of data available from year-to-year on the quantity and value of coproducts, ERS was unable to incorporate coproduct values into the calculations. Thus, in years when coproducts values are substantial, a comparison of farm receipts for oranges with retail orange juice prices will overstate the farm share. This is because retail prices alone understate the final value of products made from the oranges.

Step 1:

The Florida Department of Citrus (FDOC) publishes average retail prices for both NFC and FC juice online (www.fdocgrower.com). For example, during the 2009-10 season—from October 4, 2009 through October 2, 2010—orange juice sold as concentrate at retail stores cost $4.57 per gallon after dilution by the consumer with three parts water. NFC juice sold for $6.53 per gallon at retail.

Step 2:

The FDOC reports the quantity of concentrate that can be processed each season from fresh Florida oranges. The yield varies from season to season based on characteristics of the orange crop. In the 2009-10 season, one 90-pound box of fresh oranges yielded 1.58 gallons of concentrate at 42^{o} brix.

Step 3:

According to conversion factors published in the ERS Fruit and Tree Nuts Yearbook, ERS estimated that 0.2476 gallons of concentrate at 42^{o} brix makes one gallon of drinkable juice at 11.8^{o} brix. The information in step 2 further implies that 0.1567 boxes of fresh oranges (0.2476 / 1.58) were needed to produce a gallon of consumable orange juice during the 2009-10 season.

Step 4:

What do farmers receive for the oranges in one gallon of consumable orange juice? ERS publishes season-average prices received by Florida farmers per 90-pound box of oranges (equivalent-on-tree returns) in the Fruit and Tree Nuts Yearbook (table C22). Growers receive different prices for oranges marketed for fresh use versus oranges sold for processing. Farmers earned $6.72 per 90-pound box of processing oranges during the 2009-10 season. Farm receipts for the oranges in one gallon of consumable juice likewise totaled $1.053 that season (0.1567 x $6.72).

Step 5:

The farm share of retail prices is determined by dividing each product's farm value by its retail price, with the ratios of the prices then stated in percentage terms. For the 2009-10 season, ERS estimated the farm share of FC at 23 percent (based on $1.053 / $4.57) and the farm share of NFC at 16 percent (based on $1.053 / $6.53).

## Conversion Factors Used in Calculating Farm Share Statistics

ERS's comparisons of farm and retail prices for individual foods and baskets of foods require estimating the quantity of farm commodities in retail products and accounting for coproducts in some cases.

Conversion factors behind ERS's estimates of farm share and farm-to-retail price spreads are described below for individual dairy products, fresh fruits, and fresh vegetables. Conversion factors for flour, sugar, and margarine are described in the detailed examples under field crops.

### Dairy Products

Dairy products are made primarily from milk. For this data series, it is assumed that dairy products come from cow's milk, but some products could have been manufactured from goat, sheep, or another animal's milk.

Milk is composed of milkfat (fat) and nonfat (skim) solids. Fat is assumed to account for 3.7 percent of the milk's weight, skim solids (proteins, lactose, and minerals) account for 8.62 percent, and the remainder is water. When manufacturing some dairy products, a significant amount of either fat or nonfat solids can remain for making coproducts.

The conversion factors used for dairy products are based on two publications:

Weights, Measures, and Conversion Factors for Agricultural Commodities and Their Products. Agricultural Handbook No. 697. U.S. Department of Agriculture, Economics Research Service in cooperation with the Agricultural Marketing Service, the Agricultural Research Service, and the National Agricultural Statistics Service, 1992.

Conversion Factors and Weights and Measures For Agricultural Commodities and Their Products. Statistical Bulletin No. 616. U.S. Department of Agriculture, Economics, Statistics, and Cooperatives Service, 1979.

The conversion factors are described in greater detail for specific dairy foods:

- Fluid milk and cream
- Butter
- Cheese
- Ice cream, ice milk, and fruit sherbet
- Nonfat dry milk
- Whey
- Yogurt

#### Fluid milk and cream

Fluid milk resembles farm milk. In processing fluid milk, some fat-rich cream is removed from farm milk. It is assumed that a gallon of whole milk contains 3.3 percent fat. When making reduced-fat milk, such as 1-percent and 2-percent, much more cream is removed. In 1 pound of 2-percent milk, the fat that remains is assumed to come from 0.541 pounds of raw milk (0.02 / 0.037).

Cream, removed from raw milk by centrifugal separation, may be used to produce other dairy products including fluid cream. There are many varieties of fluid cream defined according to fat content. Heavy cream, for example, contains at least 36 percent fat. One pound of heavy cream therefore contains as much fat as 9.237 pounds of raw milk (0.36 / 0.037).

Other types of fluid cream include half-and-half which has a fat content of 10.5 to 18 percent.

#### Butter

Butter can be made by churning fluid cream. Federal regulations require that butter have a minimum fat content of 80 percent, although creameries allow for a margin of error. Assuming 1 pound of butter contains 0.803 pounds of fat, fat from 21.702 pounds of raw milk (0.803 / 0.037) would be required to manufacture 1 pound of butter.

Nonfat solids account for only 1 percent of the weight of butter. One pound of butter contains nonfat solids from 0.116 pounds of milk (0.01 / 0.0862).

#### Cheese

Fluid milk is curdled to make cheese. Curdling separates the milk into solid curds and liquid whey. Most of the fat remains in the solid curds which are then used to make cheese.

Cheddar cheese is a hard cheese. It is assumed that moisture accounts for only 39 percent of the product's weight (the standard maximum); fat and skim solids account for the remaining 61 percent. Moreover, it is assumed that there are 0.305 pounds of fat and skim solids each in 1 pound of cheese. Given the fat and skim solids in a pound of milk, fat from 8.243 pounds of milk would be in 1 pound of cheddar cheese (0.305 / 0.037). Skim solids from 3.538 pounds of milk would also be included (0.305 / 0.0862).

Monterey cheese is assumed to have slightly more moisture than Cheddar cheese, with moisture accounting for 44 percent of the weight. In 1 pound of Monterey cheese, there are 0.28 pounds of fat and skim solids each. The fat from 7.568 pounds of milk is in 1 pound of Monterey cheese (0.28 / 0.037) as are skim solids from 3.248 pounds of milk (0.28 / 0.0862).

Far less moisture is removed from cream cheese, a soft cheese that is not ripened. Moisture may account for up to 55 percent of the weight of cream cheese and fat for 37 percent. It follows that the fat from 10 pounds of milk (0.37 / 0.037) is used to produce 1 pound of cream cheese.

Moisture may account for up to 82.5 percent of the weight of low fat cottage cheese and fat for 2 percent. It follows that the fat from 0.541 pounds (0.02 / 0.037) of milk is used to produce 1 pound of low fat cottage cheese.

Cheddar cheese and cream cheese are both examples of natural cheeses. Some natural cheeses can be blended together to make processed cheese. For example, Cheddar and Colby can be combined to make American cheese.

Cheesemakers blend other ingredients with natural cheeses including emulsifiers that melt the final product evenly when heated. Anhydrous milkfat, cream, salt, flavorings, and other ingredients may also be added. Federal standards distinguish between processed cheese, processed cheese food, and processed cheese spread according to the amounts of natural cheese, fat, and moisture they contain. The fat content of processed cheese is closest to that of natural cheese. ERS assumes that processed cheese food has 23 percent fat and processed cheese spread has 20 percent fat. A pound of processed cheese food would therefore contain the fat from 6.216 pounds (0.23 / 0.037) of fluid milk.

#### Ice cream, ice milk, and fruit sherbet

Ice cream is made by combining milk and cream in a mix that may also include ingredients such as sugar, egg yolks, and stabilizers. After homogenization and pasteurization, the mix is cooled and flavorings can be added. Air is then incorporated into the mix during freezing, which results in the volume of ice cream exceeding the volume of the mix (known as overrun). Some amount of overrun is necessary for ice cream to have a soft texture as opposed to the solid texture of ice. Fruits or other foods may also be added before the final product is packaged and hardened further in a freezer.

Federal regulations require that ice cream weigh at least 4.5 pounds per gallon and have a minimum fat content of 10 percent. However, many popular brands have less overrun and a higher fat content, such as 12 or 16 percent. Regular ice cream has more overrun and less fat than premium brands of ice cream do.

Consider 1 half-gallon of regular ice cream weighing 2.25 pounds. Fat solids account for 12 percent and skim solids 10 percent of the product's weight. It follows that the fat from 7.29 pounds (2.25 x (0.12 / 0.037)) of milk is included in the product. This same container of ice cream also includes the skim solids from 2.61 pounds (2.25 x (0.1 / 0.0862)) of milk.

Ice milk has a lower fat content than ice cream does. Fat is assumed to account for 4 percent of the weight of ice milk. Thus, 1 pound of ice milk contains the fat from 1.081 pounds of fluid milk (0.04 / 0.037).

Fruit sherbet is assumed to be 2 percent fat. One pound of this product contains the fat from 0.541 pounds of fluid milk (0.02 / 0.037).

#### Nonfat dry milk

Removing water from skim milk produces nonfat dry milk. A pound of nonfat dry milk contains only 0.008 pounds of fat. Fat from 0.216 pounds of milk (0.008 / 0.037) is likewise contained in 1 pound of nonfat dry milk. This product is rich in protein and other skim solids which constitute 96.2 percent of the product's weight. As 1 pound of raw milk contains about 0.0862 pounds of these solids, the skim solids from 11.16 pounds of raw milk (0.962 / 0.0862) are contained in 1 pound of nonfat dry milk.

#### Whey

When fluid milk is curdled to make cheese, it separates into solid curds and liquid whey. Whey is low in fat, but high in lactose, protein, vitamins, and some minerals. Water can be removed to make dry whey which is added to animal feeds, nutritional supplements, and infant formulas, among other products. Fat solids are assumed to account for 1.2 percent and skim solids 94.3 percent of the weight of dry whey. Thus, 1 pound of dry whey includes fat from 0.324 pounds of milk (0.012 / 0.037) and the nonfat solids from 10.94 pounds (0.943 / 0.0862) of milk.

#### Yogurt

Bacteria are added to low-fat milk to make yogurt which may be sweetened and flavored to offset its natural bitterness. Fruit may also be added. It is assumed that fat from 0.452 pounds of milk is used in making 1 pound of yogurt.

### Fresh Fruit

Conversion factors for fresh fruit reflect spoilage and trimming. Much of the weight of a farm commodity is lost during transportation and at other points in the marketing chain. For grapes, some clusters may be trimmed to remove spoiled or damaged grapes. It is assumed that marketers discard about 9 percent of what they buy from farmers. Thus, the remaining retail weight is 91 percent of the original farm weight. Farmers therefore supply 1/0.91 = 1.099 pounds of grapes for 1 pound sold by retailers.

Although the values for ERS's conversion factors vary by type of food, the same process is followed for all fresh fruit calculations. Below is a table of conversion factors for fresh fruit:

Fruit | Percentage loss | Conversion factor | ||||||
---|---|---|---|---|---|---|---|---|

Apples | 4 | 1.042 | ||||||

Cantaloupe | 8 | 1.087 | ||||||

Cherries | 1 | 1.087 | ||||||

Grapefruit | 3 | 1.031 | ||||||

Grapes | 9 | 1.099 | ||||||

Honeydew melon | 8 | 1.087 | ||||||

Kiwifruit | 9 | 1.099 | ||||||

Lemons | 4 | 1.042 | ||||||

Oranges | 3 | 1.031 | ||||||

Peaches | 6 | 1.064 | ||||||

Pears | 5 | 1.053 | ||||||

Plums | 5 | 1.053 | ||||||

Strawberries | 8 | 1.087 | ||||||

Watermelon | 10 | 1.111 | ||||||

Source: USDA, Economic Research Service, ERS Food Availability (Per Capita) Data System. |

### Fresh Vegetables

Conversion factors for fresh vegetables also reflect spoilage and trimming. For romaine lettuce, some damaged leaves may need to be removed. It is assumed that marketers discard about 7 percent of what they buy from farmers. The remaining retail weight is 93 percent of the weight at the farm gate. Farmers must likewise supply 1.075 (1 / 0.93) pounds of romaine lettuce for 1 pound sold by retailers.

Though the values of the conversion factors vary by the type of food, the same process is followed for all fresh vegetable calculations. Below is a table of conversion factors for fresh vegetables:

Vegetable | Percentage loss | Conversion factor | ||||||
---|---|---|---|---|---|---|---|---|

Asparagus | 9 | 1.099 | ||||||

Bell peppers | 8 | 1.087 | ||||||

Broccoli | 8 | 1.087 | ||||||

Cabbage | 7 | 1.075 | ||||||

Carrots | 3 | 1.031 | ||||||

Cauliflower | 8 | 1.087 | ||||||

Celery | 7 | 1.075 | ||||||

Corn on the cob | 8 | 1.087 | ||||||

Cucumber | 8 | 1.087 | ||||||

Iceberg lettuce | 7 | 1.075 | ||||||

Agaricus mushrooms | 6 | 1.064 | ||||||

Onions | 6 | 1.064 | ||||||

Potatoes | 4 | 1.042 | ||||||

Romaine lettuce | 7 | 1.075 | ||||||

Sweet potatoes | 10 | 1.111 | ||||||

Tomatoes | 15 | 1.176 | ||||||

Source: USDA, Economic Research Service, ERS Food Availability (Per Capita) Data System. |

## Data Visualization

In the data visualization, ERS shows how farm-to-retail price spreads can be used to analyze trends in U.S. food markets. In the last story point, ERS uses data from the Agricultural Baseline Database to depict the projected behavior of farm commodity and retail prices over the next decade. The Agricultural Baseline Database provides long-run, 10-year projections from USDA's annual long-term projections report, which is published in February every year. The database covers projections for major field crops (corn, sorghum, barley, oats, wheat, rice, soybeans, and upland cotton) and livestock (beef, pork, poultry and eggs, and dairy) from the February 2015 report. Sufficient data to forecast price spreads are not available. However, the data visualization provides projections for selected commodities used to calculate ERS’s price spreads and for the Consumer Price Index to give an indication of what might happen to farm commodity and aggregate retail prices over the next decade. These trends are computed as year-over-year percentage changes derived from data in the projections report.