- Methods Used to Produce Data
- Sources of Data (Public)
- Sources of Data (Proprietary)
- Sources of Error
- Restricted Access PP-Suite Products
- How the PP-NAP differs from other ERS Data Products
- Additional Literature
The Purchase to Plate National Average Prices (PP-NAP) provides estimated prices for individual food and beverage items reported consumed by participants in What We Eat in America, the dietary component of the National Health and Examination Survey (WWEIA/NHANES). The prices are derived using the Purchase to Plate Crosswalk (PPC) that links retail grocery scanner data from Information Resources, Inc. (IRI) with the USDA Food and Nutrient Database for Dietary Studies (FNDDS). The PP-NAP focuses on foods reported at least 10 times in WWEIA/NHANES.
The PP-NAP can be used to estimate the cost of food for the WWEIA/NHANES study participants’ reported food intake and was used by the USDA in the 2021 reevaluation of the Thrifty Food Plan (TFP). The TFP is used as the basis for Supplemental Nutrition Assistance Program (SNAP) benefits.
The PP-NAP has several limitations:
- The unit prices are national averages and do not represent the prices that WWEIA/NHANES participants (or any other consumers) face in a given locality, at a certain time, or for a specific food item.
- Unit prices are based on the retail cost of food purchased at grocery stores and other stores that sell groceries and do not include adjustment for the labor, rent, and utilities associated with restaurant food. The prices also do not include sales tax or account for any edible food loss.
- Due to data limitations, the PP-NAP calculations include some but not all “private label” (store brand) items. Thus, the unit prices in the PP-NAP that are affected by this may overestimate true average prices, because these private label items may cost less than the average of the brands and private labels that are included.
- Some seemingly similar foods in the WWEIA/NHANES survey may have different unit prices, due to the preparation method. If a convenience food was available, the PP-NAP generally used the convenience food. For example, in the PP-NAP 2015/16 data, some coffee products are priced using the assumption that the item is made at home using ground coffee, while others assume the price is based on a pre-made drink (such as a bottled coffee latte).
- Individual prices change between cycles of PP-NAP as new products come on the market, older products are removed, and consumers change their preferences for package sizes.
The PP-NAP can be linked to WWEIA/NHANES data by joining the food_code variable with the corresponding WWEIA/NHANES variable. The PP-NAP also contains the price per 100 edible grams (price_100gm), other columns that cover the method used to calculate the price (method, method_description), and whether the food can be found in NHANES or only FNDDS. WWEIA/NHANES food quantity data are reported in edible grams.
The following variables are available for each 2-year data period:
- Year – Indicates the data period of NHANES where the food_code is found
- food_code – Code assigned for the food in the FNDDS, which is the same code used in WWEIA/NHANES
- mod_code – Code assigned to a modification of a food_code. Use food_code and mod_code when linking to WWEIA/NHANES (PP-NAP 2011/2012 only)
- food_description – Description of the food associated with food_code
- method – Number of the method used to calculate the price
- method_description – Name of the method used to calculate the price (the publications under “Price Construction” in the Additional Literature section provide additional details)
- NHANES – Indicates why the food_code is included in the PP-NAP (not in PP-NAP 2011/2012):
- Base – Food identified at the beginning of the price creation process
- Top 90 – Participants in WWEIA/NHANES, reported eating this food or beverage at least 10 times (PP-NAP 2013/2014 only)
- Extra – Has a direct link to IRI grocery scanner data and exists in the FNDDS only or the food was added later in the process to facilitate the next revision of the USDA Food Plan market baskets
- Alcohol – Alcoholic beverage (PP-NAP 2013/2014 only)
- price_100gm – The price per 100 edible grams of the food in nominal dollars for the years of the IRI data used to estimate the price
The datasets (described below) used to build PP-NAP are USDA’s What We Eat in America (WWEIA), the dietary component of the National Health and Nutrition Examination Survey (NHANES), the USDA Food and Nutrient Database for Dietary Studies (FNDDS), the Information Resources, Inc. (IRI) retail grocery scanner data, and the Purchase to Plate Crosswalk (PPC). The generalized process for construction of the national average prices is as follows:
- Select USDA food codes representing foods or beverages that were reported 10 or more times in the WWEIA cycle, or are needed for the Thrifty Food Plan (TFP) market basket update.
- Review recipes in the FNDDS of the selected foods to determine whether recipe modifications are needed to accommodate convenience foods and include standard substitutions in the recipe.
- Calculate the total purchase weight sold and expenditure for each item in the IRI data.
- Link scanner data Universal Product Codes (UPCs) (barcodes) to food codes in the USDA nutrition databases using the PPC.
- Convert purchase weight to edible weight for each UPC using the PPC conversion factors.
- Sum the total expenditure and edible weight across UPCs for each USDA food code.
- Calculate the average price per 100 grams of the USDA food codes by dividing expenditure by edible weight.
- Apply average unit prices for the USDA food codes to the recipes identified in step two and calculate the unit price of the FNDDS foods selected in step one.
The resulting prices are nationally representative of foods reported eaten by WWEIA/NHANES participants. The resulting price for any particular food may be derived from multiple brands and private labels, package sizes, flavors, and types. The IRI grocery scanner data represent individual products as purchased in a grocery store. The WWEIA/NHANES codes and supporting recipes in the FNDDS are more general (e.g., BBQ sauce versus a brand-specific BBQ sauce). More information on each of these datasets, including links to the data, is included in the Additional Literature section.
What We Eat in America/National Health and Nutrition Examination Survey/ (WWEIA/NHANES) – NHANES is a biennial program of surveys aimed at assessing the health and nutrition of non-institutionalized adults and children living in the United States. Of particular interest is the WWEIA survey that includes one or two 24-hour dietary recall interviews for each respondent. The USDA then calculates the nutritional content for the foods reported—including the vitamin and mineral content, and the food pattern equivalents (food group quantities). Respondents describe the quantity of the food or beverage, and this information is uniformly converted to grams of food or beverage. Note that the grams of food are only the edible portion—after the skin, peels, seeds, bones, and shells are removed and the food is cooked.
Food and Nutrient Database for Dietary Studies (FNDDS) – The FNDDS provides the nutrient content for the foods in the WWEIA survey. USDA determines the average nutrient content of these foods by developing recipes. Some recipes are a single ingredient (e.g., an apple, milk, orange juice, cooked chicken with no salt added), while others include multiple ingredients, and some incorporate standardized ingredient recipes. For example, cooked chicken might be used in several recipes. The recipes are used in the PP-NAP to estimate the cost of the food, but revisions are needed to incorporate convenience foods, include commonly used substitutions, and include all forms (e.g., fresh, frozen, dried, shelf-stable, cooked, raw, and purchased with or without refuse (skins, peels, seeds, bones, shells)).
Information Resources, Inc. (IRI) InfoScan – The IRI InfoScan dataset is a record of retail food store sales from across the country at the UPC level. Participating stores include standard grocery stores, mass merchandisers or super centers, convenience stores, drug stores, and dollar stores. The IRI Product Dictionary (PD) offers additional information at the UPC level—including weight, flavor, and nutrients from the nutrition facts label. This nutrient information, however, is lacking in comparison to USDA nutrition databases. The weight is the purchase weight, so items like fresh fruits and vegetables may contain refuse (e.g., peels, seeds, skins), while other items (such as raw meat, pasta, and raw rice) would gain or lose weight when the consumer cooks the food. More information about the data, including the statistical properties of these data are available.
There are three known sources of error in the PP-NAP: the quality of the matches, the choice of recipe used to price the food in NHANES, and issues with the scanner data.
- Matching – Although matches between the scanner data and the USDA data are done using computer-assisted methods, each match is reviewed by trained nutritionists. For consistency, a sample of each nutritionist’s review is reviewed by a senior nutritionist. The stated error rate for the matches is less than 5 percent, but the true error rate is likely closer to zero. Remaining errors in the matches are most likely due to errors in the product dictionary.
- Recipe choice – The FNDDS recipes developed for WWEIA/NHANES were used as a starting point. However, most of these recipes had to be modified and, in some cases, a prepared product was used. When comparing prices of two closely related products (e.g., hot coffee made from coffee grounds versus a mocha-latte purchased ready-to-drink), the price differences are not considered errors.
- Retail scanner data:
- Errors in the product dictionary – In rare cases, the product information is incomplete, or the product is categorized incorrectly.
- Missing stores - Compared to Economic Census data, InfoScan represents half of all grocery sales but only represents 15 percent of stores. This coverage varies by parts of the country.
- Stores may not be representative of the national composition of stores – ERS is developing store weights to account for some of these shortcomings.
- Private label products may be underrepresented – Some chains do not share their private label (store brand) data with ERS at a sufficient level of detail to estimate prices.
The following products are restricted access because they include IRI’s proprietary data. Information on accessing the IRI data and these products is available on the Using Proprietary Data page.
The Purchase to Plate Crosswalk (PPC) allows researchers to measure how well consumer purchases or store sales adhere to Federal nutrition guidance by calculating Healthy Eating Index (HEI) scores. It does this by linking retail foods in the IRI household and store scanner data with foods found in the FNDDS. The PPC provides a crosswalk between about 350,000 retail foods reported in IRI grocery scanner data in a given year to about 3,200 foods included in the FNDDS in the closest matching year(s). The retail foods in IRI data are typical grocery items and are ready-to-eat, ready-to-heat, or ingredients used to prepare meals and snacks. The foods in the FNDDS are “prepared” foods (e.g., cooked, peeled, etc.), ready-to-eat, reported “as consumed,” or in some cases raw but without the inedible bones, seeds, shells, or skins included. To allow users to align purchase quantities with the nutrients in the USDA data, the PPC provides conversion factors associated with each grocery item that converts the item’s purchased weight to the edible (“as consumed”) weight.
The Purchase to Plate Price Tool (PPPT) estimates food prices for the foods WWEIA/NHANES participants report consuming. It includes the program and data that produces the PP-NAP, following the steps described in Methods Used to Produce Data. The tool allows researchers to estimate food prices for a subset of the IRI grocery scanner data.
The Purchase to Plate Ingredient Tool (PPIT) produces data that indicates how much of each item in the recipe should be purchased to prepare 100 grams of the food. Estimated price information for the ingredients is also included.
The Fruit and Vegetable Prices data product looks at the average prices for more than 150 commonly consumed fresh and processed fruits and vegetables, including the retail price per pound and the price per edible cup equivalent. The prices in this data product are also constructed from the IRI grocery scanner data, however the scanner data foods used are fewer and more focused than the PP-NAP. For example, organic fruits and vegetables are not included and package sizes are limited. The PP-NAP includes all package sizes and organic foods. In addition, the PP-NAP food costs are tied to codes in the FNDDS while the Fruit and Vegetable Prices are not. Finally, the PP-NAP covers a more diverse set of foods than the 150 most frequently consumed fruits and vegetables.
The Food Expenditure Series tracks total expenditures at food-at-home (FAH) and food-away-from-home (FAFH) establishments by different retail and food service type. The Food Expenditure Series uses the retail sales approach that provides the most direct measure of total purchases by final users. The primary data sources are four annual surveys reported by the U.S. Department of Commerce, Bureau of the Census that provide information on sales for industries that sell food: Annual Retail Trade Survey (ARTS), Service Annual Survey (SAS), Annual Wholesale Trade Survey (AWTS), and Annual Survey of Manufactures (ASM). These data are not broken down by individual foods. The annual average consumer FAH expenditure data are used to validate the PP-NAP. The PP-NAP allows users to develop a distribution of total food costs based on the 2 days of dietary recall data in NHANES.
The Food Price Outlook uses the U.S. Bureau of Labor Statistics Consumer Price Index (CPI) and Producer Price Index (PPI) and other information to forecast the price changes of major categories of food. The PP-NAP provides food costs for specific foods for past years.
The National Household Food Acquisition and Purchase Survey (FoodAPS) is a comprehensive survey of all food acquisitions made by participating households over a 1-week period. The data also contain links between UPCs and the FNDDS. The PP-NAP provides average prices for items in the FNDDS, while FoodAPS records the price paid by the household for a specific grocery or FAFH item.
The Food Consumption and Nutrient Intakes data product lists the average daily ingested amount of the major food groups, added sugars, and discretionary fats and oils. These data are reported for different age and income levels, as well as FAH and FAFH. Quantities are based on the WWEIA/NHANES survey. The PP-NAP provides national average prices for foods in the WWEIA/NHANES survey, so that researchers can use additional data in WWEIA/NHANES for their research questions.
These publications provide background on how National Average Prices are constructed:
- Estimating Prices for Foods in the National Health and Nutrition Examination Survey: The Purchase to Plate Price Tool
- Development of the Purchase to Plate Crosswalk and Price Tool: Estimating Prices for the National Health and Nutrition Examination Survey (NHANES) Foods and Measuring the Healthfulness of Retail Food Purchases
IRI InfoScan and Consumer Network
Please see ERS’s Using Proprietary Data page for more information on the IRI databases—including how to gain access to the IRI data and considerations when using the data and comparisons of IRI InfoScan and Consumer Network database to other databases.
USDA Nutrition Database
The PP-NAP uses the recipes in the Food and Nutrient Database for Dietary Studies (FNDDS) to create prices.
USDA Purchase to Plate Crosswalk (PPC)
The following report includes information on linking the IRI grocery scanner data to the USDA FNDDS, including matching methodology and conversion factors: