Estimation Methods
Import Shares
Characteristics of Estimates
Data and Methodology

As the U.S. population has grown in both number and ethnic diversity, the volume and variety of food consumed and imported in the United States has increased correspondingly.  In 2009, U.S. food consumption totaled 654 billion pounds, or more than 2,100 pounds per capita. Of this amount, imports accounted for 17 percent (110 billion pounds), or 358 pounds per capita.  Imports account for an increasing share of food consumed in the United States, much of which cannot be produced domestically due to climate conditions and seasonality of crop production.  Also, some imported foods cost less to produce that their domestically grown counterparts.  Consumers prefer an increasingly wider selection of food products, such as tropical fruit, vegetables, nuts, premium coffee, wines and beers, cheeses, confections, grain products, and processed meats.  As more food producers and processors, both domestic and foreign, vie to supply these products, policymakers need to know which commodities are more dependent on imports.

One approach to estimating the percentage of imports as a share of food consumption uses the volume of food imports and consumption, which is measured as the ratio between the physical weights of food imports to total food consumed.  Another approach uses values (measured in dollars) for food imports and consumed food. This study examines the differences in these two procedures.  Policy analysts and other researchers can use this information to gauge the growing importance of imported products as supplements to the food supplied domestically by U.S. farmers and food manufacturers.  The sections that follow describe the methods and data used for each approach and their strengths and weaknesses, provide import share estimates using both approaches, and then analyze the differences between the two sets of results.

Estimation Methods

Analysts use two approaches to estimate the import share of food consumption.  The first is based on the volume (i.e., the physical weight) of imports divided by the volume of domestic consumption for each food group or product.  The second is based on the values of imports and food consumed domestically.  In both approaches, the units of measure for imports and consumption should be the same.  For example, since most U.S. red meat imports are measured in carcass weight, red meat consumption should also be measured in carcass weight.  Likewise, since food imports are generally priced at wholesale value, food consumption should be valued at the producer price level, not at retail value.

The Volume Method:

There are a number of data requirements for estimating import shares based on volume.  In the ERS food availability and disappearance database, the physical weights of imported processed products are converted into their equivalent fresh or farm weights.  Thus, import units are compatible with consumption units which are also measured in fresh or farm weights.  Liquid or quantity measures for food are converted into weight equivalents using conversion factors or average weight per unit.  The ERS food availability database already provides most of these conversions.  Food consumption (or, disappearance) volumes are estimated by adding imports to production, then subtracting exports.  Nonfood uses, such as industrial materials, pet food, seed, feed, and losses in storage, handling, and transport, are removed.

To estimate the import share of U.S. food consumption based on volume, analysts divide the physical weight of imports for each food group or their aggregate by the physical weight of the corresponding food group or aggregate consumed in the United States. Import volumes and consumption estimates of all food groups are found in the ERS Food Availability Data System. The foods derived from farm animals are: Red meats, poultry meat, eggs, fish and shellfish, dairy products, and animal fats.  Foods obtained from plants are: Vegetables oils, nuts, fruits (fresh, frozen, canned, and dried), fruit juices, wine, vegetables (fresh, canned, frozen), legumes (dry), food grains, sugar and sweeteners, tropical products (coffee, cocoa, tea, spices), and beverages.

The Value Method:

Foods are classified as unprocessed or processed for the purpose of valuing their cost of production.  The value of unprocessed foods is based on prices received by farmers and is represented by farm cash receipts as estimated by ERS from NASS data.  These foods cover all farm commodities that are sold for fresh consumption or for processing into food products.  Thus, total U.S. farm cash receipts for food commodities include the cost of materials used by food manufacturers.  As a result, the estimated value of processed foods is represented by value added in food manufacturing since the cost of materials is already included in farm cash receipts for food commodities.  The value of food consumption (or use) is estimated as production value plus imports, then minus exports.  The production value for unprocessed foods is their corresponding farm cash receipts, whereas the production value of processed foods is their farm receipts plus value added by manufacturers.  Commodity and food stocks are not accounted for given that the change between beginning and ending stocks do not significantly affect the annual value of total food consumption.

The import share of food consumption based on value is the ratio of imported food value to domestic consumption value.  Again, no estimates of food losses due to waste or spoilage are accounted for in calculating food consumption value.  Assuming that food losses apply proportionally to both imports and consumption, they are expected to offset each other in the import/consumption ratio.  While stocks of food commodities owned by farmers are excluded from farm cash receipts, processed food inventories are not accounted for because data on the value of stocks of imported food products are not available.

Unprocessed foods consist of the following groups: Food grains (wheat and rice), peanuts, vegetables and melons, fruits and tree nuts, meat animals, dairy milk, poultry and eggs, and fish and marine.  The groups under processed foods are: Grains and oilseed milling products, sugar and confections, preserved fruit and vegetables, dairy products, meat products, fish and seafood, bakery products, other processed foods, and beverages.  Each is identified by a 4-digit NAICS code (North American Industrial Classification System) with corresponding value added in manufacturing and import values.  The data sources are the Annual Survey of Manufactures (U.S. Census Bureau), the International Trade Administration's trade database (U.S. Department of Commerce), and ERS farm cash receipts.

Import Shares

Estimates of the import share of total U.S. food consumption based on volume and value were found to be not significantly different from each other (tables 1 and 2).  The import share estimate in 2009 is 16.8 percent based on volume and 17.3 percent based on value.  In the early 1990s, the share estimates from volume and value were both around 11 percent.  Over the next two decades both import shares gradually increased until the shares based on value surpassed the shares based on volume in 2008.  Based on volume, the import share of animal products climbed

•  Table 1 - Import shares of US food consumption using the volume method
•  Table 2 - Import shares of U.S. food consumption using the value method

from 5 percent to 6 percent over the same period, led by fish and shellfish.  For plant products, the import share went from 16.8 percent in 1990 to 25.6 percent in 2009.

The share of imports relative to the volume of food consumption in 1989 was 11 percent, 13 percent in 1997, 15 percent in 2004, and 16.8 percent in 2009.  These increasing import shares reflect the 2.6-percent average annual growth of per capita food imports, from 227 pounds in 1989 to 358 pounds in 2009.  During the same period, per capita food consumption (including imports) grew by 0.3 percent on average annually, from about 1,998 pounds in 1990 to 2,127 pounds in 2009.  As the U.S. population expanded from 247 million in 1989 to 307 million in 2009, imported food grew by 54 billion pounds in two decades even as domestic food production increased by 107 billion pounds.  In fact, U.S. food production has grown by an average 1 percent annually over the past 20 years-about the same pace as population growth during that period.

Based on volume, import shares by food group mostly show upward trends.  The share of imported grains and grain products in consumption climbed from 9 percent in 1989 to 13 percent in 2009.  Similar increasing import shares are observed for fruits and nuts, vegetables, sweeteners, wine, and beer.  Among animal product groups, only fish and shellfish have large import shares which have significantly grown over the past two decades.

Import shares of food consumption based on value differ between unprocessed and processed products.  While both product groups started with similar import shares of 10 to 11 percent in 1990, the import share of unprocessed foods has grown faster to 19 percent in 2009 versus 16 percent for processed foods.  This indicates increasing amounts of imported fresh fruits, nuts, and vegetables for direct consumption.  Domestic food manufacturers are also importing more raw and intermediate food commodities for further processing.  Indeed, U.S. imports of bulk and intermediate food products grew from $6 billion in 1990 to $17 billion in 2009 (or in volume terms, from 6.6 to 12.3 million metric tons).

The growth patterns of the two import share measures reflect, in part, the rising cost of imports as the dollar depreciated in purchasing power beginning in 2003 (fig. 1).  The cost of imported foods has also increased significantly since 2006 as world food prices followed farm commodity and petroleum prices to higher levels.  The United States continued to import high-value products such as coffee, cocoa, sugar, coconut and palm oils, and wine, whose prices were rising as the dollar kept depreciating.  The United States also continues to import increasingly larger quantities of fruits and vegetables and other horticulture products, a group that has not experienced any decline in annual import value since 1992, before falling in the recession year of 2009.  In fact, U.S. agricultural import unit values have risen by an average 6.8 percent rate annually from 2003 to 2008.  This contrasts sharply with less than 1 percent import cost inflation

on average in the preceding 6 years (1997 to 2002).  Based on these differing inflation rate patterns in recent years, it can be inferred that import prices are distorting import shares based on value by imposing an upward bias relative to import shares based on volume.

Given that the value of U.S. food consumption in 2009 amounted to an estimated $443 billion at the producer price level, and that the corresponding consumption volume is 654 billion pounds, a pound of food consumed in 2009 cost an average 68 cents at wholesale.  In 1989, the cost was 46 cents per pound.  This 2-percent average annual cost increase reflects both domestic price inflation and import price inflation of food products consumed in the United States, with import prices contributing an increasing share over the past 20 years.  Thus, since the share of imports in U.S. food consumption was around 17 percent in 2009, then about 17 percent of wholesale food price inflation in 2009 is explained by or attributed to import prices.  If import prices rose faster than domestic food prices, imported foods will account for more than 17 percent of the wholesale price inflation in 2009.  This can occur if the dollar depreciated in price-adjusted terms against the currencies of countries that supply U.S. food imports.  Hence, one role that an import share of food consumption estimate can play is to gauge the relative contributions to overall inflation of imported food prices as opposed to domestic-produced food prices.

Among the food groups, the highest import shares are for fresh fish and shellfish, fruits and nuts, sweeteners, and wine and beer.  The import share for tropical products (coffee, cocoa, tea, and spices) is nearly 100 percent since domestic production is close to zero.  Other examples of near 100-percent import shares are bananas, tropical oils (palm, coconut, and palm kernel), olive oil, and cashew nuts.  Among fruits and vegetables, the highest import shares are for apple juice, table grapes, and fresh tomatoes.  Other foods with increasing import shares include lamb meat, melons, orange juice, frozen potatoes (French fries), and candy.  While some products are imported due to lack of domestic production, others have increased over time as demand for offseason produce, preferences for premium quality, and the variety of imported food products have climbed along with consumer incomes.

Foods with the lowest import shares, or foods with the highest share of domestic production, include poultry meat and eggs, milk and other dairy products, and bulk grains.  Nevertheless, 21 percent of rice consumed by Americans is imported as well as significant shares of grain products such as pasta, noodles, flour/dough mixes, and other bakery input products.  Thirty percent of wine and 14 percent of beer are now imported.  Only 4 percent of U.S. pork and cheese consumption is imported, and only 3 percent of peanuts, head lettuce, and wheat flour.  Other products that are produced abundantly on domestic farms relative to foreign suppliers include almonds, apples, oranges, canned tomatoes, and fresh mushrooms.  Some food imports are produced in foreign countries by U.S.-owned enterprises, such as Dole pineapples, Del Monte canned fruit, or Chiquita bananas.

Characteristics of Estimates

Import shares can be calculated from either volumes or values of imports and food consumption.  Each method has advantages and disadvantages.  An advantage in using volumes or quantities is that the effect of prices on import share is excluded.  If import and consumption volumes remain unchanged from one year to the next, import shares should be the same.  However, if import prices rise relative to domestic prices (perhaps due to a depreciated home currency), import shares based on value will be higher even if import volumes did not change in proportion to total food consumption.  Disadvantages of using volumes (physical weights) to estimate import share include the need to transform liquid and other non-weight quantities into weight-equivalent units, as well as converting product or processed weights into farm or fresh weight units.  Another shortcoming is that water-laden products such as fresh produce are heavier than grains, nuts, or spices that have little moisture content such that in $1 million worth of each commodity, for example, fresh produce will typically weigh more than the low-moisture crops.  As a result, the aggregate volume of different commodities is biased toward heavier water-laden crops.

If the value of food consumption is based on consumer food expenditures, its import share will be underestimated because of value added in the form of retail and food service margins, overhead expenses, and labor costs.  Since import values are generally based on wholesale prices, they do not reflect food service or retail costs.  Thus, in order for import share to be calculated from consistent or similar units, food consumption value has to reflect wholesale or producer prices.  If the shipment value of domestically processed foods is used in estimating the value of food consumption, the cost of materials (raw and intermediate inputs for food processing) has to be discounted to avoid double-counting farm receipts for food commodities sold to processors/manufacturers.  That is, total U.S. farm cash receipts are added to value-added in domestic food manufacturing to estimate food consumption value at producer prices, after adding imports, and subtracting exports.

In terms of value, U.S. imports of consumer-ready foods (including fish and shellfish) are more than twice those of imported bulk and intermediate products that are not ready for consumption without further processing.  In volume terms, the ratio is smaller since their relative weights are closer to each other.  In this case, the import share of consumer-ready foods based on value will be greater than their import share based on volume.  This bias is corrected by ERS commodity analysts by converting the product weight of imported processed foods into farm or fresh weight in order to ensure consistency in estimators such as import share of consumption.  Nevertheless, as the United States imports an increasing amount of high-value products such as processed meats, wine, cheese, coffee, chocolate, olive oil, or tree nuts, the overall import share based on value will rise relative to import share based on volume.  That is, if the rate of increase in imports due to expensive products exceeds the growth rate of total consumption in value terms, whereas their corresponding volumes have a smaller growth differential, then the overall import share of consumption based on value will climb faster than the import share based on volume.

A seemingly logical way to estimate import share of consumption from values is to multiply wholesale or producer prices by their respective food import and consumption volumes.  This procedure, however, is flawed because it is mathematically equivalent to using volumes.  More critically, processed food imports have value added, which is not accounted for when volumes are used.  While food import and consumption values (computed as volume multiplied by price) are affected by both demand/supply and price/exchange rate changes, their volumes are only affected directly by demand/supply forces, and indirectly by price/exchange rate fluctuations.  That is, price changes alter import and consumption demand (as well as production and export supply), which in turn move the respective quantities demanded or supplied (i.e., volumes).

Given that food import and consumption volumes are more readily available (in the ERS food availability database), calculating import share from them is straightforward.  For import share based on values, due to no available data at wholesale prices, food consumption value has to be roughly estimated from farm cash receipts, value added in food manufacturing, and net imports.  Since farm sales receipts are recorded by commodity whereas value added is by industry, import shares estimates by food group combine commodity data with industry data in measuring food consumption value.  Thus, for the sake of expediency and relative accuracy, using food import and consumption volumes are the better choice.  While there are sharply contrasting costs in terms of data-gathering and calculation effort between the two import share estimation methods, their results should be mutually consistent if measurement errors are minimal.  Indeed, judging from the estimation results, the import share estimators-one based on volume and the other on value-are found not to be significantly different from each other.

Estimating the dollar value of U.S. food consumption at the producer or wholesale level provides only an approximate measure of the actual value.  This is one disadvantage of using values to calculate the import share of consumption (in addition to the difficulty in obtaining the various input data).  If the volume of food consumed is unchanged from one year to the next but prices have increased during the same period, then the value of food consumption will have expanded.  Only if the value of imports grew by the same percentage as the value of consumption will the import share of consumption based on values remain unchanged in this case, as it would if based on volume.  Thus, the effect of changes in domestic prices of food commodities, as well as the effect of changes in exchange rates on import costs, can distort the import share of consumption based on value if corresponding volumes do not change.

The estimated value of food consumed in the United States in 2009 was $450 billion, of which processed foods accounted for $297 billion and unprocessed foods accounted for $152.9 billion.  It should be noted that the value for unprocessed food consumption is estimated from the sum of U.S. farm cash receipts for all food commodities and imported foods, minus food exports.  The double-counting of farm sales receipts for food commodities intended for processing (i.e., not consumed as unprocessed but counted) is corrected by using only value added by food manufacturers to estimate the value of U.S. processed food production.  This adjustment in valuing processed food production helps avoid overestimation of food consumption value that effectively leads to underestimating import share of consumption.

Some imports compete with domestic products, such as grain products, processed meat, cheese, sugar, wine, and beer.  If the responsiveness of import demand to price changes is high, the analysis of contributions of import prices relative to domestic prices to overall food price inflation will be dynamic.  That is, if import prices of competing products rise faster than their domestic counterparts, consumer preference is expected to shift more in favor of the domestic products over at least the short term.  This change in preference will be reflected in a lower import share of consumption for that food group in the following year (if using annual data), which implies a relatively smaller contribution of import prices to overall food price inflation.  Thus, the estimated import share of consumption each year serves effectively as a weight that can be used to calculate a composite wholesale price which is a weighted average of import prices and domestic prices.

Measuring import share using volumes is more straightforward because the import and food consumption (or disappearance) data are readily available from the ERS online database.  In addition, this approach avoids the effect of changing prices in the absence of volume change.  Alternatively, using dollar values to calculate import shares requires one data source for imports and two other sources for estimating the value of U.S.-consumed food.  Estimating food consumption from production values of processed and unprocessed foods and their net imports is a cumbersome and indirect procedure, which also raises the likelihood of measurement and calculation errors.  Despite these differences in data and methods, the volume-based and value-based measures resulted in comparable import shares that show similar upward trends.  Although import shares by food groups may differ between the two measurement methods, their estimated aggregate import shares for total food disappearance were found to be not significantly different.

Data and Methodology

ERS maintains an online database of the food available for consumption in the United States.  The ERS Food Availability (Per Capita) Data System includes historical supply-and-use tables for most of the food commodities that make up the American diet (e.g., beef, chicken, eggs, fruit and vegetables). The data cover imports and domestic food disappearance (consumption and waste).  Estimates of the import share of the volume of each food item or food group are derived from this database.  For import shares based on value, estimates are calculated using the U.S. Census Bureau's Annual Survey of Manufactures, which provides data on value added in the manufacture and processing of farm commodities.  ERS's farm cash receipts estimates provide data on production value for unprocessed food commodities.  USDA's Global Agricultural Trade System (GATS) database provides data on import values.

In ERS's Food Availability (Per Capita) Data System, foods consumed in the United States are categorized into ten groups: four are produced from farm animal products and six are produced from plant products.  U.S. food imports are divided into these groupings as well:

Animal products:

• Red meat
• Poultry and eggs
• Dairy products
• Fish and shellfish

Plant products:

• Grains and products
• Fruit and nuts
• Vegetables
• Sweeteners
• Tropical products (coffee, cocoa, tea, spices)
• Wine and beer

Nonfood agricultural products, such as animal feed or industrial materials, are excluded from these food groups.  A comprehensive list of all U.S. food imports is available at ("Value of U.S. food imports, by food group" in 14 Excel tables).  The ERS Food Availability database provides data for volume of production, imports, exports, stocks, and disappearance (consumption and waste) for each food commodity.  The ten food groups can also be classified as either unprocessed or processed.  Within these groups, unprocessed foods that are consumable without further processing include fresh fruit, fresh vegetables, tree nuts, and eggs.  The remaining foods are classified as processed or manufactured products before final sale to consumers, including all food commodities that are cut, frozen, dried, canned, bottled, preserved, prepared, juiced, or fermented.  All meat, dairy, grain, oilseed, and sweetener products are considered processing commodities as they require further processing before consumption.

Industry data defined by NAICS codes (North American Industry Classification System) are used for the production, import, and export values of processed foods in estimating the value of U.S. food consumption at the wholesale level.  Farm cash receipts as estimated by ERS are used as values of unprocessed food commodities (much of which are inputs in food processing).  U.S. Census trade values for unprocessed food commodities are then combined with farm cash receipts to calculate consumption values of unprocessed foods.  Finally, consumption values of processed and unprocessed foods as well as values of imported processed and unprocessed foods are added together to calculate the combined import share for processed and unprocessed foods.  Thus, the aggregate import share for all consumed foods reflects commodity as well as industry data.

The following food groups represent non-manufactured products of the crop production industry (NAICS 111) whose import and export values and corresponding farm cash receipts are used in estimating the import share of food consumption:

• Grains and oilseeds (NAICS 1111)
• Vegetables and melons (NAICS 1112)
• Fruits and tree nuts (NAICS 1113)
• Beef cattle; swine (NAICS 1121, 1122)
• Poultry; sheep and goats (NAICS 1123, 1124)
• Farm fish and products (NAICS 1125)
• Fish and marine products (NAICS 114)

The source for import and export data for these food groups is the U.S. Department of Commerce's International Trade Administration, which identifies product groups by 4-digit NAICS codes.  This is the same source for trade data of the manufactured food industry (NAICS 311) whose product groups are listed below.  The Customs value for imported merchandise is the transaction value at port of entry into the United States.  These include packing costs, selling commissions, and license fees incurred by the importer.  Excluded costs are transportation, insurance, and other related expenses before reaching the port of entry.   Customs duties and other federal taxes paid by the buyer upon entry are also excluded from the imported good's customs value.

• Grain and oilseed milling (NAICS 3112)
• Sugar and confections (NAICS 3113)
• Preserved fruit and vegetables (NAICS 3114)
• Dairy products (NAICS 3115)
• Meat products (NAICS 3116)
• Fish and seafood (NAICS 3117)
• Bakery products (NAICS 3118)
• Other processed food (NAICS 3119)
• Beverages (NAICS 3121)

The production value for beverages is corrected to remove value added in distilleries (for spirits and liquor).  Value added by food manufacturers is obtained for these NAICS codes from the U.S. Census Bureau's Annual Survey of Manufactures.  These food industry data are aggregated in estimating the import share of the value of total food and beverage consumption in the United States, as shown in the summary table below.  For each food aggregate (or group), imports are divided by that food aggregate's total consumption, which includes those imports.

To credibly compare these value-derived estimates with volume-based estimates of food consumption, supply and use accounting was critical to both approaches.  As applied to commodities, food consumption equals total supply (production plus imports) minus exports.  In this equation, food consumption is effectively a residual use variable which implicitly includes imports.  That is, food consumption is a domestic use variable.

Last updated: Wednesday, May 30, 2012

For more information contact: Alberto Jerardo