National Academies Press: OpenBook

Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary (2015)

Chapter: 2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses

« Previous: 1 Introduction and Background
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

2

The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses

The purpose of the first technical session of the workshop, as summarized in this chapter, was to introduce the current methods, data, and uses of the food availability (FA) and loss-adjusted food availability (LAFA) data provided to the public through the Economic Research Service (ERS) Food Availability Data System (FADS). Cheryl Christensen (ERS) moderated the session and introduced the two speakers, also from ERS: Mark Jekanowski and Jean Buzby. The first section of this chapter reports on Jekanowski’s description of the FA data structure and uses, followed by Buzby’s description of the food loss estimates and the LAFA data structure and uses. The final section of this chapter summarizes the open discussion between the speakers and audience.

STATEMENT OF MARK JEKANOWSKI FOOD AVAILABILITY DATA STRUCTURE AND USES

Jekanowski described the core data series that ERS provides on FA as a general proxy for consumption. He explained the mechanics of how the estimates are prepared, how they are used, and some of their strengths and weaknesses.

He described FA as a massive dataset, with more than 200 common food categories (commodities), including grains (oats, rye, wheat, etc.); dairy products (cheeses, dry, fluid, frozen, etc.); meats (fish, poultry, red meat); eggs; sweeteners (caloric, by type: sugar, honey, high-fructose corn syrup, dextrose); peanuts and tree nuts; coffee, tea, and cocoa; spices;

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

vegetables (more than 50 types: fresh, frozen, canned); and fruits (about 40 types: fresh, canned, dried, frozen, juice). FADS is the only source of time-series FA data, and, for many items, the data extend back to 1909. He noted that it is a very rich source of data for tracking a proxy of food consumption or dietary changes over time.

Jekanowski pointed out that the FA data do not provide a direct measure of consumption. Instead, they provide a measure of disappearance of a food commodity from the supply chain, with the result often referred to informally as per capita consumption. Per capita consumption by commodity, computed from the FA data, overstates actual consumption because it does not account for waste or loss along the retail marketing chain.

Jekanowski used a flowchart to illustrate FA supply and disappearance (see Figure 2-1). For each commodity, the data system relies on annual measures of U.S. agricultural production and stocks (inventories) at the farm level from the National Agricultural Statistics Service (NASS) and on estimates of U.S. imports and exports from the Census Bureau’s trade data. Total supply is the sum of beginning stocks, production, and imports. Disappearance from supply is the aggregate of ending stocks,

images

FIGURE 2-1 Commodity supply and disappearance flowchart. Accounting relationships that illustrate food availablity supply and disappearance.
NOTE: BEA = Bureau of Economic Analysis, NASS = National Agricultural Statistics Service, and WASDE = World Agricultural Supply and Demand Estimates.
SOURCE: Prepared by M. Jekanowski for presentation at the workshop.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

exports, food use, and an estimate for farm and industrial use. The data to estimate farm and industrial (nonfood) use, if available, come from a variety of sources depending on the commodity, and include products used on the farm for seed, feed, or industrial uses such as ethanol or biofuels.1 If an estimate of food use is available, as it is for a few commodities, the estimate for farm and industrial use is the residual computed as supply minus the aggregate of exports, ending stocks, use by the food industry, and use by any other industry for which data are available, such as biofuels. For commodities like wheat and for various fats and oils, usage by the food industry was historically measured directly and published in the Census Bureau’s Current Industrial Reports (CIR), a data series terminated in 2012 due to budget constraints.

Supply and use balance sheets are used to estimate domestic disappearance from supply during a year, and each provides the estimate of food availability for that commodity. In Figure 2-1, the box labeled Domestic Disappearance on the far right provides an estimate for the amount of food that was available for consumption.

Jekanowski said the supply and use balance sheets on which FADS is based are used for all of USDA’s World Agricultural Supply and Demand Estimates (WASDE).2 They provide routine ongoing estimates of supply and demand conditions that assist analysts in better understanding price conditions and in forecasting production and market outcomes. In addition to FADS, commodity balance sheet data also feed directly into farm income forecasts and provide a fundamental way to look at the agricultural economy.

He illustrated FADS methodology by showing three examples of how food availability is estimated, for a grain (oats, see Table 2-1), a meat product (beef, see Table 2-2), and a vegetable (fresh carrots, see Table 2-3).

In Table 2-1, the spreadsheet for oats includes data for the most recent year then available, 2011.3 Although data for oats go back to 1921, Jekanowski said he included fewer years on his table for space considerations.

The first column under Supply is production, based on NASS acreage and yield estimates for oats. Data for imports, the second column under Supply, come from Census Bureau trade information, while the third column is NASS information on stocks of oats in storage at the beginning of the given year. Total supply is the aggregate of production, imports, and beginning stocks for the marketing year.

________________

1The Energy Information Administration provides information on biofuels production.

2For information on WASDE, see: http://www.usda.gov/oce/commodity/wasde/ [July 2014].

3The data were updated on the website to 2012 at about the time of the workshop.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

Under Disappearance, the first column is exports from the Census Bureau’s trade statistics. The third column is stocks of oats in storage at the end of the marketing year, in millions of bushels.4 Historically through 2011, total food disappearance of oats came from the CIR series. As mentioned above, the Census Bureau used to conduct quarterly surveys of the milling industry that reported production volumes of different types of grain-based flours, which provided a more direct measure of food availability for oats and other grains.

Returning to the second column under Disappearance, Jekanowski said nonfood use for oats is primarily feed and seed. Since total disappearance is the sum of exports, ending stocks, nonfood use, and food use—and total disappearance balances with total supply—the difference between supply and the aggregate of exports, ending stocks, and food use provides an estimate for nonfood use as the residual use category. Without a direct estimate of food use, he said, it cannot be separated from residual use and would be lumped together with feed and seed use. The spreadsheet for oats is similar to those for wheat and several other commodities that relied on the CIR for direct food availability. Most of these have not been updated to 2012 because of the loss of the CIR, but in a few cases, he noted, food use has been updated by extending long-term trends and using data provided by industry. NASS is planning to start collecting some of the data formerly provided by the CIR by the end of 2014, he said.

Finally, the total disappearance quantity in bushels is converted to pounds and divided by the U.S. population to give a grain equivalent estimate of per capita availability. This is multiplied by an adjustment factor5 of 0.60 (in this case) to adjust for milling rates and to get per capita availability on a product equivalent basis for oats.

Table 2-2 shows a similar spreadsheet for beef, he explained. The concept is the same, but with all columns measured in pounds, starting out with measures of production,6 imports, and beginning stocks,7 to give total supply for beef. For Disappearance, the columns are exports, shipments to U.S. territories (also from the Census Bureau’s trade data), and ending stocks. ERS does not provide an estimate of nonfood use for beef, although some nonfood uses exist.8 Total beef supply less exports,

________________

4The difference between stocks at the beginning of a year and the stocks at the end of a year represents amounts of product that were available to be consumed during that year.

5The adjustment factor varies for different types of grains or different types of products.

6Production of meat comes from NASS data on slaughter from three sources: slaughter under federal inspection, other commercial slaughter, and slaughter on farms. See http://www.ers.usda.gov/data-products/food-availability-(per-capita)-data-system/foodavailability-documentation.aspx#meat [June 2014].

7Stocks of meat products are amounts in cold storage at a particular point in time.

8Pet food and rendering, for example.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

TABLE 2-1 Example of Oats Supply and Use

Year U.S. Population, Jan. 1 (millions) Supply (millions of bushels)
Production Imports Beginning Stocks Total Supply
1998 274.626 166.0 107.7 74.0 347.7
1999 277.790 146.2 98.6 81.4 326.2
2000 280.976 149.5 106.0 76.0 331.6
2001 283.920 117.6 96.0 72.7 286.3
2002 286.788 116.0 95.1 63.2 274.3
2003 289.518 144.4 89.7 49.8 283.9
2004 292.192 115.7 90.3 64.9 270.9
2005 294.914 114.9 91.2 57.9 264.0
2006 297.647 93.5 106.2 52.6 252.3
2007 300.574 90.4 123.3 50.6 264.3
2008 303.506 89.1 114.6 66.8 270.5
2009 306.208 93.1 94.9 84.1 272.1
2010 308.833 81.2 85.1 80.3 246.7
2011 310.939 53.6 94.1 67.6 215.3

aTotal food availability = total supply minus the sum of exports, nonfood use, and ending stocks.

bConversion factor from grain to product equivalent = 0.60.

shipments to U.S. territories, and ending stocks gives a value for carcass beef availability in any given year.

For animal products, Jekanowski explained, the spreadsheet is somewhat more complicated because three units of measure can be of interest: a carcass equivalent, a retail equivalent (retail cuts of beef that include bone9), and a boneless equivalent. The boneless equivalent allows the user to put all meats on an equal basis, for purposes of evaluating consumption and calorie intake.

The last two columns of the beef spreadsheet provide the adjustment factors to use to convert carcass weight to retail, and carcass weight to boneless. These three categories of beef availability are used to compute

________________

9The bone is not consumed, but it is purchased by the consumer.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Disappearance (millions of bushels) Food Availability
Export Nonfood Use Ending Stocks Totala (millions of bushels) Per Capita (lbs.)
Grain Equivalent Oat Products Equivalentb
1.7 207.6 81.4 57.0 7.5 4.5
1.8 191.6 76.0 56.8 7.4 4.4
1.7 200.4 72.7 56.7 7.3 4.4
2.8 161.1 63.2 59.2 7.5 4.5
2.6 161.7 49.8 60.2 7.6 4.5
2.5 154.2 64.9 62.4 7.8 4.7
2.7 147.2 57.9 63.0 7.8 4.7
2.1 146.4 52.6 62.9 7.7 4.6
2.6 134.6 50.6 64.5 7.8 4.7
2.9 128.6 66.8 66.0 7.9 4.7
3.3 115.4 84.1 67.6 8.0 4.8
2.2 122.8 80.3 66.8 7.9 4.7
2.9 108.5 67.6 67.7 7.9 4.7
2.4 98.6 55.0 59.4 6.9 4.1

SOURCE: Prepared by M. Jekanowski for presentation at the workshop. Data from ERS Food Availability Data System: http://www.ers.usda.gov/data-products/food-availability-(percapita)-data-system.aspx [July 2014].

the three categories of per capita beef availability by dividing by the U.S. population. Similar methods and spreadsheets are used for all common meat products in addition to beef, such as pork and poultry.

Jekanowski also presented a spreadsheet example for fresh carrots (see Table 2-3). In general, there is less available information concerning the vegetable industry for a given year. In particular, because vegetables are perishable, no stocks carry over from one year to another. Estimates of production (the product of acreage and yield) are available from NASS, and the Census Bureau has trade data. There is no nonfood use for most vegetables. The supply and use balance sheet consists only of production10 plus imports to yield supply, minus exports to yield total disappear-

________________

10With declining budgets in 2012, some NASS surveys for production of fresh vegetables were terminated. Fortunately for FADS, these surveys were reinstated.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

TABLE 2-2 Example of Beef Supply and Use

Year U.S. Population, July 1 (millions) Supply (millions of lbs.)
Production Imports Beginning Stocks Total Supply
2005 296.186 24,787.0 3,599.0 637.0 29,023.0
2006 298.996 26,256.4 3,084.7 571.0 29,912.1
2007 302.004 26,523.2 3,052.2 630.0 30,205.4
2008 304.798 26,663.6 2,538.1 630.0 29,831.7
2009 307.439 26,067.7 2,626.2 642.0 29,335.9
2010 309.750 26,411.9 2,297.0 565.0 29,273.9
2011 312.009 26,291.7 2,056.5 585.0 28,933.2
Food Availability

 

Year U.S. Population, July 1 (millions) Total (millions of lbs.)
Carcassa Retail Boneless
2005 296.186 27,658.8 19,361.1 18,503.7
2006 298.996 28,054.8 19,638.3 18,768.6
2007 302.004 28,042.4 19,629.7 18,760.4
2008 304.798 27,060.9 18,942.6 18,103.7
2009 307.439 26,703.1 18,692.1 17,864.3
2010 309.750 26,262.9 18,384.0 17,569.9
2011 312.009 25,399.0 17,779.3 16,992.0

aCarcass food availability = total supply minus the sum of exports, shipments to U.S. territories, and ending stocks.

ance from supply for carrots. Dividing by the total population provides a measure of per capita carrot availability. For carrots, the adjustment factor of 0.97 accounts for the losses from farm production to retail. Jekanowski explained that the adjustment accounts for things like creating baby-cut carrots and other moderately processed products.

Jekanowski noted that one of the most powerful uses of these data is to examine availability trends over time, which he illustrated with Figures 2-2, 2-3, and 2-4. He noted that Figure 2-2 shows a dramatic increase in

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Food Disappearance (millions of lbs.)
Exports Shipments to U.S.Territories Ending Stocks
698.0 95.2 571.0
1,144.9 82.4 630.0
1,434.0 99.0 630.0
1,996.00 132.9 642.0
1,934.8 133.0 565.0
2,299.0 127.0 585.0
2,784.8 149.4 600.0

 

Per Capita (lbs.) Factors (%) for Converting Weight to
Carcass Retail Boneless Retail Boneless
93.4 65.4 62.5 0.70 0.669
93.8 65.7 62.8 0.70 0.669
92.9 65.0 62.1 0.70 0.669
88.8 62.1 59.4 0.70 0.669
86.9 60.8 58.1 0.70 0.669
84.8 59.4 56.7 0.70 0.669
81.4 57.0 54.5 0.70 0.669

SOURCE: Prepared by M. Jekanowski for presentation at the workshop. Data from ERS Food Availability Data System: http://www.ers.usda.gov/data-products/food-availability-(percapita)-data-system.aspx [July 2014].

chicken availability over time, a decline in the availability of beef, and relatively stable availability of pork. He clarified that Figure 2-2 presents data on a retail equivalent basis. It is often presented on a boneless equivalent basis that reduces availability for all products, but not by the same amount. For example, chicken availability would shift down more than beef or pork, since bones account for a larger proportion of chicken available at the retail level.

He noted that Figure 2-3 shows steady increases in wheat flour avail-

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

TABLE 2-3 Example of Fresh Carrots Supply and Use

Year U.S. Population, July 1 (millions) Supply (millions of lbs.)
Production Imports Total Supply
1998 276.115 2,706.8 179.2 2,886.0
1999 279.295 2,661.7 184.8 2,846.5
2000 282.385 2,708.0 167.5 2,875.5
2001 285.309 2,783.9 201.4 2,985.3
2002 288.105 2,586.5 190.2 2,776.7
2003 290.820 2,696.4 187.2 2,883.6
2004 293.463 2,628.0 215.2 2,843.2
2005 296.186 2,654.5 196.6 2,851.1
2006 298.996 2,429.0 248.4 2,677.4
2007 302.004 2,443.0 245.5 2,688.5
2008 304.798 2,456.5 276.6 2,733.1
2009 307.439 2,216.3 298.5 2,514.8
2010 309.750 2,323.7 322.9 2,646.6
2011 312.009 2,201.2 393.6 2,594.8

aConversion factor from farm to retail = 0.97.
SOURCE: Prepared by M. Jekanowski for presentation at the workshop. Data from ERS Food Availability Data System: http://www.ers.usda.gov/data-products/food-availability-(percapita)-data-system.aspx [July 2014].

ability through the mid- to late-1990s. At about that time, the Atkins diet, a low-carbohydrate diet, became popular, and the diet’s effect can be seen in the data. A fairly sharp turnaround in wheat availability occurred from the late 1990s through the early 2000s that ultimately leveled off.

Jekanowski noted that a common perception is that healthier diets might mean greater vegetable consumption. However, he said Figure 2-4 shows this trend does not show up in the availability of types of vegetables. The figure shows a sharp increase in availability of broccoli, and relatively stable availability of cauliflower and asparagus with some ups and downs. He reminded the audience that in 1990, President George H.W. Bush talked about not liking broccoli, around the same time as the dip in broccoli availability. Although he said he did not know whether

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Disappearance (millions of lbs.) Food Availability
Exports Total (millions of lbs.) Per Capita (lbs.)
Farm Retaila
255.5 2,630.5 9.5 9.2
262.3 2,584.2 9.3 9.0
276.5 2,599.0 9.2 8.9
309.1 2,676.2 9.4 9.1
351.6 2,425.1 8.4 8.2
330.6 2,553.0 8.8 8.5
283.9 2,559.3 8.7 8.5
284.7 2,566.4 8.7 8.4
253.1 2,424.3 8.1 7.9
257.5 2,431.0 8.0 7.8
274.3 2,458.8 8.1 7.8
244.1 2,270.7 7.4 7.2
244.0 2,402.6 7.8 7.5
238.9 2,355.9 7.6 7.3

the dip was related to the President’s remarks or just coincidence, broccoli availability has increased since then.

Jekanowski next discussed advantages of the supply and use approach. First, the approach provides a full accounting of commodity use and supply for all commodities. Volumes and shares of U.S. supply that come from imports are known, as are volumes carried over from one year to the next. The data give analysts the ability to put food availability into the context of the overall food economy. He said that because the time series of the data is so long and has been estimated consistently, it is very powerful for identifying long-term dietary trends, even accounting for the fact that it is only a proxy for, and likely overestimates, consumption.

He went on to say FA data are very commonly used in estimating

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

images

FIGURE 2-2 Trends in per capita meat availability.
SOURCE: Prepared by M. Jekanowski for presentation at the workshop. Data from USDA’s Economic Research Service Food Availability Data System. Available: http://www.ers.usda.gov/data-products/food-availability-(per-capita)-data-system.aspx [July 2014].

complete demand systems, especially for meats, and for estimating elasticities for beef, pork, and chicken. By relying on existing data, FADS is less data-intensive than trying to track actual consumer purchases.

Jekanowski reminded the audience that use of scanner data in FADS has been mentioned. In his view, it would be useful to consider such new approaches, at least as a backup or as a companion to the current methodology. He noted one complication with scanner data is that, in most cases, they are product-based (e.g., baked goods or pizza) rather than commodity-based (ingredients). Accounting for all the different ingredients in the vast array of consumer products is not an easy task.

According to Jekanowski, an additional advantage is that FADS data measure the total amount of supply going into the food industry without regard to how it is used within that industry. As a result, FADS is immune to changes in the style or form of consumer food purchases and uses. He noted that direct surveys of consumers would likely entail significant levels of self-reporting biases. People tend to overestimate the amount of healthy foods they consume and underestimate the amounts of snacks and desserts.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

images

FIGURE 2-3 Trends in per capita wheat flour availability.
SOURCE: Prepared by M. Jekanowski for presentation at the workshop. Data from ERS Food Availability Data System. Available: http://www.ers.usda.gov/dataproducts/food-availability-(per-capita)-data-system.aspx [July 2014].

As a final advantage, he said the FADS data are based almost entirely on widely available public data sources. As a result, FADS is very transparent.

He then described what he sees as limitations. First, FADS takes a commodity focus, starting at the farm level with production and inventories, and staying at that level. It provides no information about specific food items, such as individual retail cuts of meat or different kinds of processed foods. Clearly, he said, from a health policy perspective, better knowledge about consumption patterns would be useful.

Second, FADS does not capture all common food categories. For example, total grain consumption is known, but not how much of that grain is consumed as whole versus highly refined grains. Some categories are omitted because there are no good data. For example, game meat, home garden production, and niche markets such as soy foods are excluded because of a lack of data.

Third, the data are available only annually and at a national level. While there are likely seasonal consumption patterns for different foods, FADS data cannot be used to analyze seasonal effects. The data also do

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

images

FIGURE 2-4 Trends in per capita availability of selected vegetables.
SOURCE: Prepared by M. Jekanowski for presentation at the workshop. Data from USDA’s Economic Research Service Food Availability Data System. Available: http://www.ers.usda.gov/data-products/food-availability-(per-capita)-data-system.aspx [July 2014].

not support analysis of regional consumption patterns or an analysis of demographic or socioeconomic patterns of consumption.

Fourth, there is a lag between the date of the data and when the data are made available to the public. At the time of the workshop (April 2014), for instance, ERS had just released the data for 2012. Because FADS relies on data from other sources (particularly NASS and the Census Bureau), it cannot be finalized until all components from all sources are available and final.

An additional limitation he highlighted is that food availability overstates consumption because it does not account for waste or spoilage. Finally, he said, since FADS requires data from many sources, it is vulnerable to decisions by other agencies that impact the availability or content of their data, such as the Census Bureau decision to terminate its CIR series. FADS relied heavily on these data for consumption of many grains, added fats and oils, and items like margarine and salad dressing. ERS is hopeful that, within a year or so, the data will be available from NASS to report on those categories once again, he said.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

STATEMENT OF JEAN BUZBY LOSS-ADJUSTED FOOD AVAILABILITY DATA

Buzby described the LAFA data series that was developed by Kantor et al. (1997) to help account for the fact that the core FA data series overstates the amount of food consumed. She stressed ERS refers to the LAFA series as preliminary because the underlying loss assumptions and estimates continue to be refined. She noted that the LAFA series takes into account the substantial quantities of food that go uneaten because of spoilage, moisture loss, plate waste, and other reasons from farm to plate. The primary goal is to more closely approximate actual consumption.

She provided the ERS definitions of food loss and food waste:

  • Food loss represents the edible amount of food, postharvest, that is available for human consumption but is not consumed for any reason. It includes cooking loss and natural shrinkage (e.g., moisture loss); loss from mold, pests, or inadequate climate control; and food waste.
  • Food waste is a component of food loss and occurs when an edible item goes unconsumed, as in food discarded by retailers due to color or appearance and plate waste by consumers.

Buzby emphasized that food waste in the LAFA series is just one component of food loss, and that ERS does not have estimates for all different components of food loss. She then provided examples of where food loss can occur at different levels. At the farm level, preharvest losses can be due to severe weather, disease, or predation from insects, birds, and animals. Losses during harvest can be caused by machinery or production problems, as well as business decisions to leave portions of a field unharvested. Postharvest loss refers to loss after harvest, or in the case of milk, after a cow has been milked. Estimates of production from NASS are supposed to be net of farm loss.

She said losses at the processing and wholesaling levels can arise from discarding substandard products, such as bruised fruit or oddly shaped vegetables that do not meet supermarkets’ quality standards. Losses can also arise from shrinkage, poor handling, cold storage failure, or transportation problems. There are also cooking and preparation losses, food removed from the system because of food safety concerns, and plate waste that occurs due to differing tastes and preferences or preparation of more food than needed. The bottom line, she said, is many different reasons account for food loss from the farm to the fork. At this time, ERS only produces summary statistics for losses at the retail level (supermarket) and the consumer level.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

Buzby illustrated the loss adjustment for fresh carrots (see Table 2-4). Loss adjustment starts from farm-level food availability from FADS, illustrated in Table 2-3. There are adjustments for three types of losses: (1) loss from the farm (primary) to retail, (2) loss at the retail (supermarket) level, and (3) loss at the consumer level. Consumer loss is further split into a nonedible share and other losses (e.g., cooking and uneaten food). All loss adjustments are expressed as a percentage.

The second column in Table 2-4,11 referred to as primary weight, is the weight in pounds of per capita farm availability of fresh carrots from the FA spreadsheet (Table 2-3). The third column shows the 3 percent loss from primary to retail weight used to calculate retail food availability in Table 2-3. The fourth column shows the retail food availability of fresh carrots.

The next column shows retail loss, or loss at the supermarket, to be 5.1 percent for all years. After accounting for supermarket loss, the result is consumer weight. There are two adjustments made at the consumer level. The first is the nonedible share, estimated to be 11 percent for carrots. Buzby noted that ERS has very good data on the percentage of a food that is nonedible by commodity from the USDA’s National Nutrient Database for Standard Reference. The second is other losses (cooking and uneaten food), estimated to be 34 percent.

Buzby clarified the total loss at all levels (49 percent) is not the sum of the different losses because losses are taken sequentially. After all losses are accounted for, the result is per capita availability adjusted for loss and is presented in three different units: pounds per year, ounces per day, and grams per day.

The final columns in the spreadsheet in Table 2-4 use conversion factors to get calories per cup equivalent or grams per cup equivalent, she explained. The next-to-last column shows calories available per day and food pattern equivalents (a measure of the number of servings) available per day. These figures are frequently compared against federal dietary recommendations.

Buzby stated that an aggregate view of food loss is provided in a new ERS report that she coauthored (Buzby, Wells, and Hyman, 2014). It provides loss-adjusted estimates in terms of weight, value, and, for the first time, calories. In 2010, according to the report, the aggregate of loss at the retail and consumer level was 133 billion pounds, valued at $162 billion (using retail prices from the Nielsen Homescan data) and accounted for 141 trillion calories. The report also presents annual per capita LAFA, both per year and per day.

________________

11Table 2-4 shows preliminary data for 2010. The data for 2010 were revised in Table 2-3. The two tables also use different numbers of significant digits.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

Buzby provided examples of how LAFA estimates are used. For example, Figure 2-5 illustrates the most popular fruits in terms of loss-adjusted availability in 1970 versus 2011. In 1970, apples were the most popular fruit in terms of the amount available for consumption. However, by 2011, bananas had caught up and surpassed apples. Watermelon and grapes moved up in popularity from 1970 to 2011.

She went on to describe a chart showing LAFA in calories by category in 2010 (see Figure 2-6). This figure shows flour and cereal products provided 610 calories per day for the average American in 2010, more than any other food group, followed by added fats and oils (not including naturally occurring fats and oils, such as the fat in meat). Availability of caloric sweeteners (added sweeteners) was 400 calories per day, excluding naturally occurring sugars, such as fructose and fruit.

Buzby referred to Figure 2-7 to illustrate per capita loss-adjusted data for the five food groups in 2012 as a percentage of the federal dietary recommendations for a 2000-calorie diet. The average U.S. diet falls short of USDA’s MyPlate recommendations for vegetables, dairy, and fruit. On average, Americans consumed more than the recommended amounts of meat and grains in 2012. Looking back to 1970, she noted an increase in loss-adjusted availability of fruits and vegetables, even though availability still does not come close to the MyPlate recommendations.

Buzby shared data on food loss at the retail and consumer level for a variety of commodities in 2010 (see Table 2-5). As noted earlier, the estimated total postharvest food loss was 133 billion pounds, or 31 percent of the food supply. Total loss at the retail level was 43 billion pounds, or about 10 percent of the food supply. At the consumer level, losses were almost 90 billion pounds, or 21 percent of the food supply.

Food losses vary by commodity, as Buzby illustrated in Figure 2-8. She explained that the figure shows the amount of food loss for each food group by the length of the bar, measured in billion pounds. For each bar, the two colors show the amount of retail loss (yellow) and consumer loss (blue). She noted the split in losses between retail and consumer, as well as the variance in total losses by food group. For example, for grain products, 39 percent of the loss occurred at the retail level and 61 percent occurred at the consumer level. Added fats and oils was the only food group where a larger portion of loss occurred at the retail level than at the consumer level. Dairy products had the largest loss at the retail level, while vegetables had the largest loss at the consumer level.

Figure 2-9 depicts the three food groups with the highest share of food loss in the United States in 2010 (plus a residual “other” category), as measured by amount (weight in pounds), value (measured in dollars), and calories. If measured in pounds, the top three food groups in terms of loss are dairy, vegetables, and grains. If measured by value, the top three

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

TABLE 2-4 Loss-Adjusted Food Availability for Carrots (per capita)

Year Weighta (lbs/year) Loss from Primary to Retail % Retail Weight (lbs/year) Loss from Retail to Consumer %
1998 9.53 3 9.24 5.1
1999 9.25 3 8.98 5.1
2000 9.20 3 8.93 5.1
2001 9.38 3 9.10 5.1
2002 8.42 3 8.16 5.1
2003 8.78 3 8.52 5.1
2004 8.72 3 8.46 5.1
2005 8.66 3 8.40 5.1
2006 8.11 3 7.86 5.1
2007 8.05 3 7.81 5.1
2008 8.07 3 7.82 5.1
2009 7.39 3 7.16 5.1
2010 7.61 3 7.38 5.1

 

Per Capita Availability Adjusted for Loss
Lbs/Year Ozs./Day Grams/Day Calories per Cup Equivalent
4.82 0.21 5.99 52.0
4.68 0.21 5.82 52.0
4.66 0.20 5.79 52.0
4.75 0.21 5.90 52.0
4.26 0.19 5.29 52.0
4.44 0.19 5.52 52.0
4.41 0.19 5.48 52.0
4.39 0.19 5.45 52.0
4.10 0.18 5.10 52.0
4.07 0.18 5.06 52.0
4.08 0.18 5.07 52.0
3.74 0.16 4.65 52.0
3.85 0.17 4.79 52.0

aPrimary weight for carrots pertains to per capita farm availability.

SOURCE: Prepared by J. Buzby for presentation at the workshop. Data from USDA’s Economic Research Service Food Availability Data System. Available: http://www.ers.usda.gov/data-products/food-availability-(per-capita)-data-system.aspx [July 2014].

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Consumer Weight (lbs/year) Loss at Consumer Level  
Nonedible Share % Other (cooking loss and uneaten food) % Total Loss, All Levels %
8.77 11.00 34.0 49
8.51 11.00 34.0 49
8.47 11.00 34.0 49
8.63 11.00 34.0 49
7.75 11.00 34.0 49
8.08 11.00 34.0 49
8.02 11.00 34.0 49
7.97 11.00 34.0 49
7.46 11.00 34.0 49
7.41 11.00 34.0 49
7.42 11.00 34.0 49
6.80 11.00 34.0 49
7.00 11.00 34.0 49

 

Grams per Cup Equivalent Calories Available Daily Food Pattern Equivalents Available Daily (cups)
128.0 2.4 .047
128.0 2.4 .045
128.0 2.4 .045
128.0 2.4 .046
128.0 2.2 .041
128.0 2.2 .043
128.0 2.2 .043
128.0 2.2 .043
128.0 2.1 .040
128.0 2.1 .040
128.0 2.1 .040
128.0 1.9 .036
128.0 1.9 .037
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

images

FIGURE 2-5 Most common fruits available for U.S. consumers, 1970 versus 2011.
SOURCE: Updated by J. Buzby for the workshop, based on http://www.ers.usda.gov/data-products/chart-gallery/detail.aspx?chartId=30486 [July 2014].

images

FIGURE 2-6 Loss-adjusted food availability (LAFA) in calories per day by category, 2010.
SOURCE: Available: http://www.ers.usda.gov/data-products/food-availability%28per-capita%29-data-system/summary-findings.aspx#.U8QhKkA_wwc [July 2014].

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

images

FIGURE 2-7 U.S. diet in five food groups as compared with USDA MyPlate recommendations.
NOTE: Rice data were discontinued and thus were not included in the grains group. Loss-Adjusted Food Availability data serve as proxies for food consumption.
1Based on a 2,000-calorie diet.
SOURCE: Updated by J. Buzby for the workshop, based on figure in http://www.ers.usda.gov/data-products/food-availability-%28per-capita%29-data-system/summary-findings.aspx#.U8QhKkA_wwc [July 2014].

food groups in terms of loss are meats, dairy, and vegetables. If measured by calories, the top three groups in terms of loss are grains, added sugar and sweeteners, and added fats and oils—the ingredients in calorie-dense foods, she pointed out. Comparing the first two charts in Figure 2-9, on a value basis the meat, poultry, and fish category constitutes 30 percent of the total value, but only 12 percent of the total weight, because foods in this group tend to cost more per pound than many other foods.

Buzby turned her presentation to ERS initiatives for improving the LAFA data series, noting that the ERS long-run goal is to update the data series by reviewing, updating, and documenting each loss estimate for each individual commodity to the most recent year of data available.

In the past two decades, ERS had two cooperative agreements to update the farm-to-retail weight loss factors, one with the University of Minnesota’s Food Industry Center (TFIC), and the other with Pennsylvania State University and the International Life Sciences Institute (ILSI). ERS commodity analysts are using some of these estimates in both the supply and use spreadsheets and the loss-adjusted spreadsheets.12

________________

12See the following link for more detail: http://www.ers.usda.gov/data-products/foodavailability-(per-capita)-data-system/loss-adjusted-food-availability-documentation.aspx [June 2014].

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

TABLE 2-5 Estimated Total Food Loss in the United States, 2010

  Losses from Food Supplya (billion pounds)
Commodity Retail Consumer Total
Dairy products   9.3 16.2   25.4
Vegetables   7.0 18.2   25.2
Grain products   7.2 11.3   18.5
Fruit   6.0 12.5   18.4
Added sugar and sweeteners   4.5 12.3   16.7
Meat, poultry, and fish   2.7 12.7   15.3
Added fats and oils   5.4   4.5     9.9
Eggs   0.7   2.1     2.8
Tree nuts and peanuts   0.2   0.3     0.5
Total 43.0 89.9 132.9

aTotals may not add due to rounding.

SOURCE: Prepared by J. Buzby for presentation at workshop, based on Buzby, Wells, and Hyman (2014, p. 12).

images

FIGURE 2-8 Quantity losses at the consumer and retail levels for nine food categories (measured in billion pounds).
   aIncludes loss in the home and away-from-home locations. Includes cooking shrinkage and uneaten foods.
SOURCE: Buzby, Wells, and Aulakh (2014).

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

images

FIGURE 2-9 Annual food loss of top three food groups measured by amount, value, and calories.
SOURCE: Buzby, Wells, and Aulakh (2014).

Buzby went on to say that for the losses at the retail level in 2005-2006, ERS sponsored retail-level loss assessments for fresh fruits, vegetables, meat, poultry, and seafood, and it adopted the new methods in February 2009 (see Buzby et al., 2009). ERS recently received new data from the Perishable Groups, now part of Nielsen, for 2011 and 2012, and is in the process of reviewing this new information. Buzby noted that the new data include qualitative information from produce, meat, and seafood managers about where and why food loss occurs at the retail level. She said many retail-level loss estimates need updating and documenting, particularly added fats and oils, added sugars and sweeteners, fluid milk and dairy products, grain products, processed fruits and vegetables, eggs, and peanuts and tree nuts. Retail losses for some of these commodities have not been updated since 1997 (see Kantor et al., 1997).

Through a grant with RTI International, ERS obtained loss estimates at the consumer level for most of the commodities in the LAFA data series and adopted them in the LAFA data in August 2012 (see Muth et al., 2011). However, she noted, not all consumer-level loss estimates were reviewed and revised at that time, and many could be revisited, such as dry edible beans, peas and lentils, and certain commodities in the following food groups: fruits and vegetables, beverage milks, grains, sugar and sweeteners, and added fats and oils.

Many of the challenges associated with the core FA data described by Jekanowski in the previous section also apply to the LAFA series, Buzby said. Additionally, she stated, the preliminary loss-adjusted series has some of its own challenges. Buzby noted that data limitations prevent ERS from estimating total food loss across all commodities at the farm level and at the farm-to-retail levels. Although ERS could consider doing

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

an exploratory analysis to see if they could do a summary total for retail, this would require a new study, she explained.

Buzby summarized additional challenges and potential opportunities with the LAFA series. First, as noted previously, the loss-adjustment percentage estimates are the same each year from 1970 to the most recent year, with a few exceptions. (Beef is one such exception, where some conversion factors, from the carcass weight to the boneless weight, changed between 1986 and 1996, reflecting a closer trimming away of fat.13) She said validating and updating adjustment factors is a potential opportunity for improvement.

Another limitation she noted is that the data series does not adequately reflect retail-level food donations to food banks and other charitable organizations, or the transfers of unsold food to thrift shops for sale at lower prices. This means that some food currently classified as a loss may be consumed, and thus is not a food loss.

Buzby questioned the structure of the data series, in particular the point at which the inedible share of food is removed. She asked whether the ERS approach is as consistent and accurate as possible. She also said users have said they would like to see the consumer-level food loss estimates split into loss at home and loss away from home. She noted that the same could be said for food available at home versus food available away from home.

Other users have asked for finer levels of consumer-level loss detail, such as plate waste, or cooking loss, but these requests may require data that do not yet exist. She suggested a possible solution might be the addition of a separate column for cooking loss, although this addition would require work to verify its feasibility. She said the series could also be improved by better accounting for processed foods, especially in food imports and exports.

Buzby reminded the audience that unlike the core FA series, the LAFA series only goes back to 1970. She stated that the LAFA data expressed in terms of calories and food pattern equivalents are particularly important. ERS also uses the embedded loss assumptions to estimate the amount, value, and calories of food loss at the retail and consumer levels.

She said that, like the FA estimates, the LAFA series serves as a proxy for actual consumption for over 200 commodities in the United States. LAFA provides estimates for individual commodities and food groups. There are some commodities for which FA data are available, but LAFA data are not. For example, coffee, tea, and cocoa are not part of the major food groups tracked in LAFA.

She ended by saying that like the FA data series, the LAFA series is

________________

13This change is illustrated in Table 3-1 in Chapter 3, but for pork rather than beef.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

particularly useful for studying food consumption trends, and pointed to a later session in the workshop (summarized in Chapter 3) for a discussion of uses of the data. She pointed out the additional limitation of keeping up with changes that impact assignment of losses to a sector. For example, in recent years, supermarkets are providing more fruits and vegetables in cut-up form. This packaging probably shifts loss formerly taken at the consumer level to the retail (supermarket) level and has not yet been accounted for in ERS estimates, she said.

OPEN DISCUSSION

Sarah Nusser (Iowa State University) asked whether the new data from NASS on consumption of grains and oils that will replace the Census Bureau’s discontinued Current Industrial Reports (CIR) will use the same methodology. Jekanowski replied that for the most part, NASS is planning to replicate what the Census Bureau produced. However, the agency is rethinking what data can be collected and what would be useful to collect. NASS is surveying the major USDA users, such as ERS and the World Board, to make sure the new survey will satisfy the needs of the greatest number of users. For example, NASS may collect information on the amount of grain sold from mills as whole versus refined grain, data the Census Bureau did not collect.

Laurian Unnevehr (University of Illinois at Urbana-Champaign) noted that the level of commodity disaggregation in the FA data has increased over time. As an example, she said she found 10 years of data on kale in the vegetable series. She asked how the FA data series has evolved and whether it is because NASS is collecting more detailed data on things like horticultural crops. Jekanowski replied that if kale and other foods are reported in FADS, it is because the data are available from NASS. He reminded the audience that because of budget cuts, NASS planned to suspend reporting on most of their vegetable production data for the 2012 production season. ERS was concerned about the potential impact on the FADS data, but NASS reinstated most of those reports.

Helen Jensen (Iowa State University) asked if FADS incorporates grains produced and put in multi-ingredient or composite products that are exported or imported. For example, she asked, if wheat is put into cookies that are exported, is that counted as a loss? Buzby replied that FA and LAFA data are commodity-based; as such, they include foods like wheat flour and oats but exclude multi-ingredient foods like pasta, breads, and cookies. She noted a study (Batres-Marquez and Jensen, 2002) that analyzed imports and exports of processed grain products as the only work done in this area, but agreed, with increasing imports and exports of processed foods, it is becoming a more important issue. She said it might

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

be possible to make such an adjustment at the retail level, but the details would need to be worked out. She said this is an area where the LAFA data could be improved.

Josef Schmidhuber (Food and Agriculture Organization [FAO]) referred participants to his later presentation on the FAO approach to identifying imports and exports of commodities in processed foods (summarized in Chapter 4). A commodity like wheat is not eaten as wheat but in products like noodles and bread, he pointed out. The challenge is not just accounting for traded processed products, but many foods, such as breakfast cereals, are composite foods of many ingredients, and it is not clear how spreadsheets can account for them. He noted that FAO is confronted with the same problem, but for more than 180 countries, including the United States.

The FAO approach to dealing with multi-ingredient or processed foods is quite complex, he said. Methods to account for such foods are needed not only for trade, but also for all elements of the balance sheet. Processed products are also in storage, and all are consumed. Wheat can be stored as grain, but it can also be stored as flour and cookies. Buzby agreed that the lack of data on composite or multi-ingredient foods is an issue for the FADS estimates of food availability and loss.

Schmidhuber asked whether FADS considers availability to occur at the farm level or at the retail level and whether the system makes any adjustment for waste at the farm level. Jekanowski replied that for the most part, FADS starts with availability at the farm level. There are some minor adjustments, for example, going from carcass weight to retail weight to boneless weight, that will account for loss at the retail level. However, it all derives from the farm level.

Schmidhuber reminded the audience of Jekanowski’s statement that food availability is sometimes computed as a residual, noting that FAO faces the same problem. In an ideal situation, food consumption would be estimated directly, not as a residual. He noted that food consumption is the least elastic of all forms of consumption, at least in the United States. If it is computed as a residual, the estimate includes all the uncertainty that is actually in all the other variables of the balance. He noted this is one of the elements that the FAO would like to change in its system. Jekanowski replied that the way food availability is computed in FADS reflects the data that are available. He reminded the audience that for most of the major grains, oil, and seeds, nonfood use was the residual because the CIR data provided food consumption directly.

Schmidhuber noted the adjustment factors and loss coefficients at various stages in FADS are constant over time, but in reality, they are not constant because of such factors as weather or the magnitude of a crop (bumper or failure). He asked whether FADS is considering accounting

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

for changes in adjustment factors over time. Buzby agreed with the need, noting the impact of new technology such as packaging for fresh fruit to enhance shelf life. ERS calls their loss-adjusted data preliminary because of these data limitations, she explained. She underscored the size of the task of computing time-varying adjustment factors because of the many commodities, many adjustments, and little data. ERS is in the process of analyzing new data on food loss at the supermarket level from the Nielsen’s Perishables Group for 2011 and 2012, which may potentially be used to update the current 2005 and 2006 estimates. With these data, she said, ERS may be able to compare the 2005-2006 estimates with the 2011-2012 estimates to obtain a snapshot of change.

Schmidhuber praised the FADS presentation of its aggregates in pounds, calories, and value14 and asked whether FADS might ever consider aggregating food losses in terms of CO2 equivalents. He noted the biggest externality of waste in developed companies is in having a too-deep resource footprint, such as too many carbon greenhouse gas emissions. Buzby said that Venkat (2012) documented some of the work to translate food use and losses to climate change impact, but she is not yet sure how ERS will be able to make use of that work.

In answer to a question from Harry de Gorter (Cornell University), Buzby said the food loss estimates consider only the retail (grocery stores, restaurants, small corner shops, hotels, restaurants, hospitals, schools, and so on) and consumer levels. She clarified that at the consumer level, food loss includes both food at home and food away from home. She said that separating these two types of consumer-level losses could be done in the future.

De Gorter referred to Buzby’s statement that total postharvest loss is 31 percent of supply and asked how much larger the figure would be if losses on the farm and between farm and processing to retail were included. Buzby explained, in general, that there are very little data on food loss at the farm level. She referred to one study (Kader, 2005), which published estimates of farm loss for fruits and vegetables, but said the data are spotty and not all are recent. She noted internal discussions about an exploratory analysis of the potential for better estimating farm-to-retail level loss. Using cheese as an example, however, she said coming up with estimates would be complicated.

A participant asked whether ERS has had the opportunity to break down the data based upon regional differences that might lead to commodity loss differences, such as rural versus urban, or even geospatial differences. Buzby said data in both the FA and LAFA series are national

________________

14See Buzby, Wells, and Hyman (2014).

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

only, with no demographic breakdowns. She suggested a food consumption survey would be a better fit for this application.

Morvarid Bagherzadeh (Organisation for Economic Co-operation and Development) remarked the supply and use balance sheets are a great input to her work. She asked Jekanowski about reconciling classifications in trade and production. For example, she noted that the spreadsheets seem to start from raw commodities, but most commodities are not traded in raw form. Jekanowski replied that analysts, who work on different commodities, go to great lengths to stay up to date on the most recent and accurate import/export codes for trade. They try to capture all forms of the product.

Erik Dohlman (ERS) said he has worked with trade codes. For beef and other animal products, most exports would be classified and captured as either frozen or chilled, with few meat exports cooked or in other processed forms. In response to de Gorter’s earlier question about farm loss, he posited that collecting data on farm loss would be challenging, because they would change every year by commodity. For example, with the California drought, the 2014 almond loss will be very different than a normal year. An analyst would have to go through the entire time series commodity by commodity to estimate farm loss accurately.

Bagherzadeh noted that in other parts of the world, retailers return food that is not consumed to producers, so they shift the loss to another part of the supply chain. She asked whether this is something that ERS would observe. Buzby replied that ERS is not capturing food that might be sent from the supermarket and donated to charity for a tax write-off, but it is another avenue that could be explored.

Alison Kretser (ILSI) asked Buzby if reporting of loss in terms of calories links back to what is reported concerning caloric consumption in the U.S. population from consumption survey data. She also asked whether the LAFA data were integrated with data from the National Health and Nutrition Examination Survey (NHANES). Buzby responded the data are not linked at this time. The estimates for calories available for consumption are provided on the LAFA spreadsheets for individual commodities. On the right-hand side of each spreadsheet, the amount of commodity available for consumption is converted to calories available for consumption per day.

Kretser also asked whether the 34 percent loss at the consumer level for carrots varies by commodity and whether it changes over time. Buzby replied that when she inherited the original FADS from Linda Kantor over a decade ago, the LAFA data series was static across time and commodities.15 For example, for all fresh vegetables, consumer-level loss was 30

________________

15See Kantor (1998).

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

percent. Since then, ERS has adopted different loss adjustments for each individual commodity at the consumer level using data from Muth et al. (2011) and where data are available.

Susan Krebs-Smith asked if the new estimates for consumer-level loss were applied for all years or for recent years. Buzby replied that ERS adopted the new consumer-level loss estimates for the entire span of the LAFA data series, namely from 1970 to the most recent year available, although some loss estimates have not been updated. For example, ERS has not had an opportunity to update many of the estimates at the retail level, such as for all canned vegetables. This is because the Perishables Group provided ERS with updated loss estimates only for fresh fruits and vegetables, meat, poultry, and seafood at the retail level.

Krebs-Smith said she views the FA and LAFA data in the way Jekanowski introduced them, as a measure of food entering retail distribution channels. She noted that ERS adds caveats to say that the data are not direct measures of food consumption. In her view, the LAFA data provide a good indicator of foods entering retail distribution channels. She said it would be good if there were measures along the food supply chain for the amount of foods, characterization of those foods that manufacturers are producing, the amount that enter retail outlets, the amount that go through food service outlets, and so on. She noted other surveys capture food that comes into the house and food intake. Ideally, she said, the different measures would be aligned so comparisons could improve the understanding of all series. For example, she observed that consumption estimates from NHANES may be biased because of a tendency to underreport. The LAFA data could be used to get a sense of an upper bound on underreporting.

She reminded the audience that it is suspected that the LAFA data do not include all losses. However, when LAFA data, even those from Kantor (1998) with static loss percentages, are compared with consumption data, they align very well for many food groups. Such comparisons provide a system of checks and balances, but, she noted, one series is not supposed to be a proxy for the other. The different data systems are measuring different things, and, she said, it helps to keep those things separate. Buzby noted that Muth et al. (2011) looked at consumer-level loss in many ways, including what people purchased using Nielsen Homescan data minus what people said they consumed in NHANES.

Mary Muth asked Jekanowski about the origins of the data ERS receives from NASS, asking how ERS knows quantities that are used are fresh versus in processed forms, such as canned and frozen. Jekanowski relied ERS generally uses data from NASS directly without adjustment. For many vegetables or fruits, ERS gets estimates from NASS of acres planted for fresh versus acres planted for frozen or canned products. The

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

product of acreage by yield gives an estimate for production of fresh, frozen, or canned. For most other crop commodities, such as grains, everything gets processed.

Responding to Krebs-Smith, Jekanowski stated that he thinks farm and processing sector losses are minor because farmers and processers have an incentive to minimize losses. They want to maximize the yield and minimize losses for every product. This might not be true in cases where weather events cause products to not be harvested or where products were lost from the food supply for other reasons.

Jensen asked about ways, now that Census has discontinued the CIR, that ERS or NASS could engage industry to collect more data. She suggested some of the larger manufacturers have collected information about losses because of the recent interest in sustainability. They have been recapturing ingredients that might be used in processing or diverting product into food for people that used to go into food for animals. Jekanowski said ERS has not considered engaging industry from a loss perspective, but it would be interesting to consider. He said there may be an opportunity in the future to add new questions now that NASS will be managing the surveys that collect information from industry.

Jekanowski discussed the definition of a loss. If a retailer sends its spoiled meat to a rendering facility and it is recycled into feed, is that really loss, or is it just an alternative use of the product? He noted that the product is not entering a landfill, but it is not being used for human consumption. He stated that it would be useful to sort out some of the terminology.

Schmidhuber noted that according to FAO data, there are very few losses at the farm level in developed countries. However, there are farm losses in developing countries. In a developed country, losses typically occur at the retail and household level, which means policy conclusions in developed countries are entirely different from those in developing country situations. He stated that the real externality in a developed country is that prices are too low and the externality of a too-deep resource footprint (water, land, biodiversity, and greenhouse gasses) is too high. In developing countries, he said, the situation is reversed. The losses take place at the farm or transportation level because of inadequate transportation and storage facilities. The policy implication is hunger can be alleviated by investing in loss reduction, but he said it would not make sense in a developed country to reduce waste in order to fight hunger. However, such a policy is considered, not only in the popular press, but also in other studies.

Schmidhuber said he shares the ERS concern about classifications, particularly because of the massive increase in processed products. As described later in the workshop (see Chapter 4), FAO has tried to address

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×

this through a proposal adopted by the UN Statistical Committee. FAO has worked on mapping the Harmonized System of codes for trade and the Central Product Classification (CPC) for production. At the level of trade disaggregation that is made available by Comtrade, a United Nations’ international trade statistics database, there was a one-to-one match. He characterized this as a huge benefit because there is no need for split factors that are required when the result of matching is many-to-many or one-to-many. He noted that the challenge is that not all countries report in CPC, and very few report in the expanded version of CPC. It would be a huge advantage for statistical systems if all countries reported in the expanded version of CPC, he said.

Bagherzadeh said farm-level loss is not small. Instead, she asserted, agencies decide to start from the farm because it is simpler. She noted times when it can make sense economically not to harvest and leave the result on farm. For example, many retail industries in developed countries have contracts with developing countries for fresh fruits and vegetables. However, standards differ by country, and not all loss data are collected. She said some losses in a developing country occur because of the specifications of products wanted by developed countries.

Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 12
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 13
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 14
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 15
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 16
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 17
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 18
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 19
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 20
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 21
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 22
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 23
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 24
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 25
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 26
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 27
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 28
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 29
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 30
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 31
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 32
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 33
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 34
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 35
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 36
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 37
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 38
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 39
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 40
Suggested Citation:"2 The Food Availability System and Food Loss Estimates: Current Methods, Data, and Uses." Institute of Medicine and National Research Council. 2015. Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary. Washington, DC: The National Academies Press. doi: 10.17226/18978.
×
Page 41
Next: 3 Historical and Current Uses of the Data for Economic Modeling and Reporting of Statistical Trends »
Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss: A Workshop Summary Get This Book
×
Buy Paperback | $48.00 Buy Ebook | $38.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The United States Department of Agriculture's (USDA's) Economic Research Service's (ERS) Food Availability Data System includes three distinct but related data series on food and nutrient availability for consumption. The data serve as popular proxies for actual consumption at the national level for over 200 commodities (e.g., fresh spinach, beef, and eggs). The core Food Availability (FA) data series provides data on the amount of food available, per capita, for human consumption in the United States with data back to 1909 for many commodities. The Loss-Adjusted Food Availability (LAFA) data series is derived from the FA data series by adjusting for food spoilage, plate waste, and other losses to more closely approximate 4 actual intake. The LAFA data provide daily estimates of the per capita availability amounts adjusted for loss (e.g., in pounds, ounces, grams, and gallons as appropriate), calories, and food pattern equivalents (i.e., "servings") of the five major food groups (fruit, vegetables, grains, meat, and dairy) available for consumption plus the amounts of added sugars and sweeteners and added fats and oils available for consumption. This fiscal year, as part of its initiative to systematically review all of its major data series, ERS decided to review the FADS data system. One of the goals of this review is to advance the knowledge and understanding of the measurement and technical aspects of the data supporting FADS so the data can be maintained and improved.

Data and Research to Improve the U.S. Food Availability System and Estimates of Food Loss is the summary of a workshop convened by the Committee on National Statistics of the National Research Council and the Food and Nutrition Board of the Institute of Medicine to advance knowledge and understanding of the measurement and technical aspects of the data supporting the LAFA data series so that these data series and subsequent food availability and food loss estimates can be maintained and improved. The workshop considered such issues as the effects of termination of selected Census Bureau and USDA data series on estimates for affected food groups and commodities; the potential for using other data sources, such as scanner data, to improve estimates of food availability; and possible ways to improve the data on food loss at the farm and retail levels and at restaurants. This report considers knowledge gaps, data sources that may be available or could be generated to fill gaps, what can be learned from other countries and international organizations, ways to ensure consistency of treatment of commodities across series, and the most promising opportunities for new data for the various food availability series.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!