Loss-Adjusted Food Availability Documentation
The ERS Loss-Adjusted Food Availability Data are derived from
ERS per capita food availability data adjusted for food spoilage,
plate waste, and other losses to more closely approximate actual
per capita intake. This data series is recommended for daily
estimates of the per capita number of:
- calories, and
- food pattern equivalents of the five major food groups
plus the amounts of added sugars and sweeteners and added fats and
oils. These estimates were previously called servings and
MyPyramid equivalents, and the data series was previously
known as the Food Guide Pyramid Servings data.
This data series is considered to be a preliminary series
because ERS has a set of initiatives underway to update and
document the underlying loss assumptions. Data can be accessed
through Excel
spreadsheets which provide all of the current loss assumptions
and the structure for the data series.
Coverage of the Data
Per capita calorie consumption and food pattern
equivalents are estimated for more than 250 agricultural
commodities from 1970 to the most recent year of data available.
Per capita data are reported for both individual commodities and
aggregated food groups. The data for individual commodities are
aggregated into food groups to facilitate comparison with
recommendations for average daily intakes for the U.S.
population.
History Behind the Data
In the mid-1990s, ERS conducted a major effort to expand the
usefulness of the Food Availability Data System for diet and
nutrition monitoring by converting annual food availability data
into daily individual intake data and by adjusting the data for
estimated spoilage and losses in the home and marketing system. The
release of the Food Guide Pyramid in 1992 provided
researchers with a new framework for assessing U.S. dietary
status--one that went beyond a traditional approach (adequacy of
individual nutrients) to a food-based approach linking diet and
chronic disease risk.
Adjusting the food availability data for spoilage and waste is
key to the integrity of the Loss-Adjusted Food Availability Data
series, but some data were prone to error and poorly documented.
Unknown errors could be introduced into the series by adjusting for
losses. Previous studies have limited documentation of food loss
estimates at different marketing levels such as retail and consumer
levels or for individual food commodity groups (for example,
peaches, corn, or beef).
ERS first attempted to estimate losses from data on food
available for consumption in three selected sectors of the
marketing system-retail stores, foodservice institutions, and the
home. ERS gathered existing food loss coefficients from published
reports and discussions with commodity experts and then applied
these coefficients to the Food Availability Data System for 1995.
See Estimating and Addressing America's Food
Losses in the January-April 1997 edition of Food
Review magazine. The article focused on understanding the
magnitude of food losses at the retail, foodservice, and consumer
levels and looked for solutions to reduce these losses through food
recovery, recycling, and education. Losses were estimated for over
250 individual foods and commodity groups aggregated into 10 food
groups.
In 1998, ERS released a second report, A Dietary
Assessment of the U.S. Food Supply: Comparing Per Capita Food
Consumption with Food Guide Pyramid Serving
Recommendations. This publication applied the loss
coefficients from the previous study to a broader time period
(1970-96), with the assumption that the loss rates remained
constant over time. Servings based on the 1996 Food Guide
Pyramid, Home and Garden Bulletin Number 252 were calculated
for the same 250 individual foods and commodity groups, which were
aggregated into five Pyramid food groups, plus added sweeteners and
added fats and oils.
Another major effort to revise, refine, and restructure the data
system was completed in 2005, and the new system was launched on
the ERS website on February 1, 2005. For the first time, this new
data system included spreadsheets for each of the hundreds of
commodities in the Loss-Adjusted Food Availability Data, and the
spreadsheets presented all of ERS's food loss assumptions.
Additionally, each fruit and vegetable has a separate spreadsheet
for different processing types. For example, apples have
spreadsheets for fresh, frozen, dehydrated/dried, and canned apples
as well as for apples made into juice. ERS provided data users with
access to the core spreadsheets in this data series to increase
transparency of the loss assumptions, with a footnote on each
spreadsheet stating that these assumptions were first estimates
intended to serve as a starting point for additional research and
discussion.
By 2005, the level of documentation of ERS's food loss
assumptions ranged from little to none for estimates at the retail
and customer levels to substantial for estimates at the farm level
and for the nonedible share for each food. These loss assumptions
were based on data and studies from the mid-1970s or earlier, but
the food marketing system has changed dramatically since then,
including innovations in processing technology and unprecedented
growth in the foodservice sector. Despite these limitations, the
food loss estimates were the best available and continued to be
used in assumptions in ERS's Loss-Adjusted Food Availability Data
System. For these and other reasons, ERS recognized the need to
update and improve all loss assumptions for each commodity for
three general types of losses:
- losses at the primary level (for example, farm to retail
weight).
- losses at the retail level, such as in supermarkets, megastores
like Walmart, and other retail outlets, including convenience
stores and mom-and-pop grocery stores. Losses in restaurants and
other foodservice outlets are not included.
- losses at the consumer level. This includes losses for food
consumed at home and away from home (for example, restaurants and
fast food outlets) by consumers and food services. There are two
components:
- "Nonedible share" of a food, such as an asparagus stalk or
apple core. Data on the nonedible share are from the National
Nutrient Database for Standard Reference, compiled by USDA's
Agricultural Research Service (U.S. Department of Agriculture,
2007).
- Cooking loss and uneaten food, such as plate waste from the
edible share.
Initiatives To Update the ERS Loss-Adjusted Food Availability
Data
ERS's long run goal for the Loss-Adjusted Food Availability Data
series is to review, update, and document each loss estimate for
each covered commodity for the most recent year of data possible,
and to ascertain if any of these loss estimates have changed since
1970 (the first year in the data series). It was necessary to
update and improve these estimates in a series of initiatives
because of the diverse nature of the three types of loss
assumptions--farm to retail, retail, and consumer levels--and
because of resource limitations. ERS has three multipart
initiatives underway to update the three types of loss assumptions
for the several hundred commodities. Although some of the
initiatives have been completed and the resulting updated loss
estimates are now used in the Loss-Adjusted Food Availability Data
series, other initiatives and subcomponents are still underway and
still other initiatives may be necessary to fill in remaining data
gaps. Hence, the data series is considered to be preliminary.
1. Losses at the primary level-farm to retail
weight
Under a cooperative agreement, ERS and the University of
Minnesota's Food Industry Center (TFIC) compiled revised
agricultural conversion factors from farm to retail. Loss estimates
are sometimes called conversion factors, particularly when
describing how a farm commodity is transformed into a
consumer-ready product (for example, fresh chicken to boneless
fresh chicken). Through information from a series of industry
interviews, TFIC updated the conversion factors for the main
categories of meats and poultry as well as for several fruits and
vegetables.
In 2007, the cooperative agreement with TFIC was completed and a
new cooperative agreement was started with Pennsylvania State
University and the International Life Sciences Institute (ILSI) to
review TFIC estimates, collect data on the remaining commodities
not covered by TFIC (for example, grains, fats, and dairy
products), and explore areas of concern, such as conversion factors
identified as most likely to change in the near future. Information
from this study is being used by some ERS commodity analysts to
update the supply and disappearance spreadsheets--the foundation of
the Food Availability Data System.
2. Losses at the retail level
In September 2007, ERS obtained updated food loss estimates at
the retail/institutional level to the consumer level (for example,
from supermarkets) for fresh fruits, vegetables, meat, poultry, and
seafood through a competitive grant with the Perishables Group,
Inc. (PG). PG compared supplier shipment data with point-of-sale
data from stores in large, national supermarket retail chains to
identify loss percentages. PG supplemented this with qualitative
information from retail contacts. The updated loss estimates from
this study had little impact on per capita food loss estimates
because the new estimates were generally close to the previous loss
assumptions. The updated loss estimates were incorporated into
ERS's Loss-Adjusted Food Availability Data series in February 2009
and are documented in the report, Supermarket Loss Estimates for Fresh Fruit,
Vegetables, Meat, Poultry, and Seafood and Their Use in the ERS
Loss-Adjusted Food Availability Data.
PG did not have appropriate data to update the retail-level loss
assumptions for specific grains, dairy, and added fats and oils in
the ERS Loss-Adjusted Food Availability Data series, and for fruits
and vegetables other than in their fresh form (such as canned,
frozen, and juice).
3. Losses at the consumer level
Under a grant with ERS, RTI International calculated updated
consumer-level loss estimates for cooking loss and food loss from
the edible share of food. In the first stage of this grant, RTI
International reviewed studies on food loss at the consumer level
and completed a small sample of restaurant interviews. The report
from this effort, Exploratory Research on Estimation of
Consumer-Level Food Loss Conversion Factors 
, concluded that there have been few published
research studies on consumer-level food loss in the United States,
and much of the published material was released by ERS.
In the second stage of this grant, RTI International used a
numerical estimation method to calculate consumer-level food loss
estimates using Nielsen Homescan data and National Health and
Nutrition Examination Survey (NHANES) data. ERS then analyzed how
the Loss-Adjusted Food Availability per capita data would change if
the proposed RTI estimates of consumer-level food loss were
incorporated into the data series. The full report, Consumer-Level Food Loss
Estimates and Their Use in the ERS Loss-Adjusted Food Availability
Data was published in January 2011.
In August 2012, ERS incorporated RTI International's "best
estimate" of consumer-level food loss conversion factors into the
loss-adjusted food availability data series. In cases where RTI
International's best estimates were unavailable, the ERS estimate
was used. ERS did not adopt the RTI estimate for fresh grapefruit.
Additionally, RTI International's best estimates for cane and beet
sugar were used for high fructose corn syrup (HFCS), glucose, and
dextrose.
In addition to these initiatives, ERS welcomes suggestions to
expand and improve its loss estimates.
Constructing the Data
The current ERS per capita food availability data were converted
into daily per capita food pattern equivalents comparable
to those identified in USDA's MyPyramid
Equivalents Database (MPED) using a multi-stage process. Each
commodity was assigned to one of five major food groups (fruits,
vegetables, meat, dairy, and grains), or to one of two additional
groups for added fats, oils, and added sweeteners. The data were
adjusted for spoilage and other losses by subtracting estimated
losses from the consumption weight reported in the food
availability data. Loss was estimated at several different stages
in the marketing system (farm to retail, retail, and consumer).
Nonedible portions of all foods-seeds, pits, and inedible
peels-were also subtracted from the data. The data were converted
from pounds per capita per year to grams per capita per day to be
comparable to food pattern equivalents.
For each food supply commodity, a food pattern
equivalent was defined, with size based on USDA's
MPED and weight based on USDA's Nutrient Database for
Standard Reference (NDB). For example, the MPED defines 1
cup of sliced, raw apple as a 1-cup equivalent of fruit and the NDB
indicates that 1-cup of sliced apple with skin weighs 109
grams.
After defining food pattern equivalent weights for each
commodity, daily per capita consumption--adjusted for loss and
nonedible parts--was converted into grams and divided by the
assigned food pattern equivalent. Food pattern
equivalent weights for individual commodities were aggregated
to total daily amounts for the five major USDA food groups, plus
the amounts for added sugars and sweeteners and for added fats and
oils. Aggregated amounts for each food group were then compared
with the amount recommended in USDA's Food
Patterns of the Dietary
Guidelines for Americans, 2010. These recommendations are
broken into 12 calorie levels ranging from 1,000 to 3,200 calories
per day. Because data are not available on the distribution of
Americans among each of the 12 calorie levels in the Dietary
Guidelines, ERS used the 2,000-calorie-per-day reference level
in the analysis to be consistent with the level used throughout the
examples in the Dietary Guidelines and on the Nutrition
Facts labels found on most packaged foods.
Limitations of the Data
As with the basic food availability data, the Loss-Adjusted Food
Availability Data series does not measure actual consumption or the
quantities ingested. This is because neither series is based on
direct observations of individual intake (see Food Availability
Documentation). Therefore, data are not available by
socioeconomic, demographic, and geographic (State, regional, or
city) breakdowns, and it is not known if such data exist.
The limited ability of researchers to measure food loss
accurately suggests that actual loss rates may differ from the
assumptions used in this data series. Estimates of farm to retail,
retail, and consumer level food losses may be understated or
overstated due to limitations in the underlying published studies.
Food loss, particularly at the consumer level, is by nature
difficult to measure accurately. Participants in household surveys
on food waste tend to be highly "reactive"--changing their behavior
during the survey period instead of acknowledging how much food
they typically discard-or misstating their true levels of product
discard. Studies that observe food loss by inspecting landfill
garbage are also prone to errors. Such studies are not nationally
representative and may not account for food fed to pets and other
animals, put in garbage disposals, or composted at home.
Plate waste studies, such as for schoolchildren at lunchtime, often
target only a slice of the total U.S. population, and the findings
cannot be extrapolated to other demographic categories.
Food loss for individual commodities, in particular, may vary
over time, yet the ERS data currently do not capture these changes.
Some of the apparent increase in food loss probably stems from
increased waste and more trimming of food. Processed foods, such as
frozen dinners, are generally more trimmed than if the raw
ingredients were prepared at home. Smaller households, with
increased away-from-home eating, may also have more waste. On the
other hand, new food technologies and food production and
processing practices, such as improvement in the preservation of
bread, may reduce food losses over time. Additionally, although ERS
has relatively well-documented data for the loss assumptions for
the nonedible share, the Loss-Adjusted Food Availability Data
series is not designed to identify where in the food production,
marketing, and consumption chain the nonedible share was removed
from food commodities.
Usefulness of the Data
Even with some limitations, both the per capita food
availability data and the per capita loss-adjusted food
availability data are useful for economic analyses because they
serve as indirect measures of trends in food use. In other words,
both data series provide an indication of whether Americans, on
average, are consuming more or less of various foods over time.
By converting food availability data into daily individual
food pattern equivalents comparable to the Dietary
Guidelines' recommendations, ERS has contributed to existing
dietary assessment research. For example, the 2008 report Dietary Assessment of Major
Trends in U.S. Food Consumption, 1970-2005 examined major
trends in the amount of food available for consumption in the
United States between 1970 and 2005 and estimated whether Americans
are meeting Federal dietary recommendations for each of the major
food groups. Researchers and policymakers can use the Loss-Adjusted
Food Availability Data series to measure changes in food
consumption behavior over time relative to major nutrition
education or policy initiatives.
Because the loss-adjusted data were derived from data for raw
and semi-processed agricultural commodities rather than for final
food products, food pattern equivalents can be readily
converted back to farm-level data. This eases the translation of
dietary recommendations into production and supply goals for
farmers and the food industry. This time series can be used as a
baseline to project future trends in food demand and to compare
these trends with recommendations for USDA food group
intakes. For example, a 2006 ERS report Possible Implications for U.S. Agriculture from
Adoption of Select Dietary Guidelines provides one view of
the potential implications for U.S. agriculture if Americans fully
met the recommendations in the 2005 Dietary Guidelines for
Americans and MyPyramid Plan for fruit, vegetables,
milk, and whole grains. A straightforward extrapolation using ERS
loss-adjusted food availability data for these food groups suggests
that the potential long-term impact on food demand and production
in the United States could be substantial.
The data are also useful for helping researchers better
understand the differences and similarities between the food supply
data and the National Health and Nutrition Examination Survey
(NHANES), which measures food products actually eaten by
individuals. In essence, the food pattern equivalent
estimates allow researchers to compare the amount and types of food
available in the food supply with information on actual food
intakes by Americans.