How grains and oilseeds flow through the U.S. food economy

The U.S. Bureau of Economic Analysis (BEA) recently (in December 2013) announced the once-every-five-years release of benchmark national input-output accounts, showing how resources flow from one industry to the next in the U.S. economy.

For people interested in the economics of the food system, some graduate students and colleagues and I last year developed a tool in Tufts' Visual Understanding Environment (VUE) for interactively illustrating such input-output flows. A working paper (#44) describes the tool. A previous blog post shows an example. The visualization tag at this blog collects other posts about interesting food policy data illustrations.

In this video, I use the new BEA benchmark input-output data to describe how grain and oilseeds flow through the food economy. Before making the video, I rounded the numbers to the nearest billion dollars and deleted some negligible small resource flows, so serious students of these data will want to refer to the original files from BEA. Because the numbers likely will be illegible in the video player embedded in the blog post, I've included a link to the original video, which you can download and play with higher resolution on your computer's own video program (such as Windows Media Player or the Mac equivalent).

Greenhouse gas emissions flows

I recently had cause to remember and appreciate this 2008 graphic from the World Resources Institute (WRI).  Ordinarily, there is confusion between various statistics one reads about economic sectors (such as transportation, energy, agriculture), about economic activities and end uses (such as heating residential buildings, heating commercial buildings), and about gasses (such as carbon dioxide and methane).  The graphic still doesn't answer one of my questions -- I am trying to reconcile environmental accounts that (a) place food distribution with the corresponding manufacturing and distribution sectors or (b) attribute all of these costs to food itself.  Nonetheless, it is a good data visualization.

USDA ERS releases "Charting the Essentials"

USDA's Economic Research Service this month published a delightful set of exhibits for food and agricultural statistics, titled "Charting the Essentials."

I especially like the maps.  For example, maps using the widely known definitions of "metro" and "non-metro" counties have the disadvantage of obscuring important rural areas within counties that are defined as "metropolitan."  As a remedy, the new ERS chart series provides this nice map of urban and rural areas, with much smaller resolution.

Similarly, it is interesting to see the contrast between the comparatively concentrated geographic location of crop production in the Midwest and California and animal agriculture production all over the country.  These maps use dot displays for the density of production value.

Using the Visual Understanding Environment software from Tufts University to illustrate food industry input-output flows

This new visualization tool allows you to explore resource flows between industries.

For example, you can see how much meat and poultry flows into the restaurant food industry, and then how much restaurant food flows to the final consumer (all measured in billions of dollars per year). You can create your own diagram showing the industries and flows that you select, in any order you choose.

This project extends the capability of Tufts University’s Visual Understanding Environment (VUE).  I worked on this with Rebecca Nemec, Graham Jeffries, Mike Korcynski, and Jonelle Lonergan.  Our working paper (.pdf) gives instructions for using several practice data sets, or for downloading your own data from the federal government's Bureau of Economic Analysis (BEA).  Accompanying data files and a processing program are available on my department's working paper series page.

The best way to understand the capabilities of this visualization tool is to watch this video, also available full-size on Vimeo.

Visualizing Input-Output Data Using VUE from Tufts University - Online on Vimeo.

Wine industry visualizations

The food industry visualizations of Michigan State University professor Phil Howard focus on relationships among businesses or sectors within the food system.

It is interesting how visualizations can carry different implicit messages even when they seem at first to address the same topic.

For example, the first of Howard's recent (December 2012) interactive visualizations of wine industry brands emphasizes the great diversity of options in the marketplace ...


... while the second of these visualizations emphasizes the comparatively heavy concentration at the corporate level.


For more about the food system more generally, see Howard's 2012 working paper with Harvey James Jr. and Mary Hendrickson. More to come on this topic of food system visualizations.

IOM report considers time constraints as part of assessing SNAP benefit adequacy

The Institute of Medicine, part of the National Academies, today released a consensus report on the adequacy of Supplemental Nutrition Assistance Program (SNAP) benefits.

The committee that prepared the report was led by Julie Caswell of UMASS Amherst.  Members included Mary Muth, Philip Gleason, Jim Ziliak, Barbara Laraia, Diane Schanzenbach, and others.  The committee had invited me to Washington for a workshop in spring 2012 to give a presentation (.pdf) about the economics of the SNAP benefit formula, the procedure that the federal government uses to assign a benefit amount to each participant household based on its income and other characteristics.

In one notable conclusion of today's report, the committee suggested that the federal government should account for the difficulty low-income families face in finding sufficient time to cook at home.  The implication is that the government could consider increasing SNAP benefits to provide low-income households with sufficient resources to purchase more convenient and easy-to-prepare foods:

Time—USDA-FNS should recognize the cost-time trade-offs involved in procuring and preparing a nutritious diet. The dollar value of the Thrifty Food Plan (TFP), with its strong reliance on preparation of meals from basic ingredients, does not account for time constraints faced by most households at all income levels, particularly those with a single working head of household, which necessitate purchasing value-added or prepared foods with a higher cost. USDA-FNS should examine the impact of accounting for cost-time trade-offs, for example, by:

• applying a time adjustment multiplier to the cost of the TFP or reviewing options for adjustments to the current cost of the plan, and

• adjusting the earned income deduction to reflect more accurately time pressures for participants who are working. 

One question is whether encouraging greater food spending on convenient foods would have any health or environmental consequences.  For example, convenience foods tend on average to have more sodium and more packaging than their traditional counterparts.

Clearly, an increased benefit to allow for convenience foods would be a special boon for highly time-constrained households -- think for example of the dreadfully tight time budget for a low-income working single parent.  Yet, many other low-income SNAP participants are retired, or live in households that include a non-working adult.

Some low-income households are able to -- and even want to -- cook at home.  For these households, it might be counter-productive to increase the maximum benefit to a level based on the price of expensive convenience foods, while simultaneously forbidding the households to economize on food and use the savings for non-food needs such as housing and transportation.

While you're thinking this over, I thought it would be helpful to conclude with some real data from the American Time Use Survey for 2011.  Most of the leisure time segment at all education levels is for television and other screen-time recreation. Source: American Time Use Survey. Note: the amount of work time appears smaller than you may expect, because the data include both weekdays and weekends, averaged for working and non-working persons aged 15 and older.

Agricultural producer support declining over time

We hear all sorts of generalizations about U.S. farm policy.

Some say U.S. farm programs are too stingy and should provide more help to farmers, especially small farmers. Others say U.S. farm programs are a boondoggle that just makes rich farmers richer. Still others say farm programs make consumers fat by encouraging too much cheap food.

Instead of generalizing, it is important to think quantitatively.

One good data source is the Producer Support Estimates (PSE) from the Organisation of Economic Cooperation and Development (a club for the world's upper-income countries). I use this data source in several chapters of my forthcoming book from Routledge/Earthscan called Food Policy in the United States: An Introduction.

The PSE data measure diverse agricultural programs and policies in a consistent way across countries and over time.  One problem with the PSE is that it can seem a little complex.  To provide an orientation, Rebecca Nemec and I created the following data gadget.  Nemec is a graduate student at the Friedman School at Tufts and the teaching assistant for my class on U.S. Food Policy.  The top panel shows broad categories of support for agricultural producers.  The bottom panel shows more detail about each broad category in turn.   

Just click on each colored broad category in the top panel to see the corresponding detail in the bottom panel.

Working from top to bottom, we learn about trends in several major categories of producer support.

• Price supports and deficiency payments help farmers in years when prices are low.  OECD worries about these programs because they distort international trade and hurt farmers overseas.  Michael Pollan criticizes deficiency payments for making corn too cheap.  Notice that in recent years -- with greater scarcity and higher prices -- these distorting policies have fallen to almost nothing under current policy.
• Conservation programs have been growing in recent years, and also do not respond to price fluctuations as wildly as deficiency payments do.
• The other payments category includes direct payments, which pay farmers regardless of the current price.  These direct payments may end under some current farm bill proposals.  They do not distort agricultural markets very much, but it is unpopular to pay farmers when they are prospering during high-price years.
• Market Price Support represents the economic impact of the trade barriers that protect some producers, especially for milk and sugar, from imports.  Although they do not have a budget cost, these supports benefit farmers at the expense of consumers.  As with deficiency payments, the impact of these trade barriers has declined to almost nothing in recent high-price years.

An especially clever feature is that PSE data allow sensible comparisons across two seemingly different types of policies:

• payments to farmers at the taxpayers' expense (the first three broad categories), and
• trade policies that support farmers at the consumers' expense (the fourth broad category).

To speak coherently about U.S. agricultural policy, one must make important distinctions across several types of programs and be aware of rapid changes in program impacts from one year to the next.

There are a couple limitations that I should mention.  First, the OECD data may have some limitations of their own.  Second, while I did the best I could to classify programs from the OECD data into sensible categories, I did make some judgement calls about these program classifications.

In general, U.S. support for farmers has been declining in recent years, mainly because of high food prices that result from greater scarcity on world markets.  Though some people are more optimistic, I think population and environmental constraints may generally keep prices fairly high in the future.

This means that governmental support for U.S. farmers can be smaller over time, unless legislators replace existing programs with new and poorly designed alternatives.  For example, I worry about new and potentially expensive crop insurance programs that have been proposed in draft farm bills.

Powered by Tableau

State-level data on children's poverty and nutrition programs

The Massachusetts Budget and Policy Center this week released new resources about breakfast and lunch participation in Massachusetts schools.  A chart pack (.pdf) illustrates data describing the extent of take-up of nutrition benefits, and a summary graphic (.pdf) traces a wide variety of nutrition assistance programs from the federal funding sources to the state and local implementation level.

More generally, the Kids Count data center from the Annie E. Casey Foundation has a wide variety of state-level data resources for all states.  For example, here is an interactive map showing children's poverty levels by regions within Massachusetts (you can mouse over selected counties to see specific statistics).

The SNAP (food stamp) explosion

Every couple years, we update this dynamic interactive graphic showing how Supplemental Nutrition Assistance Program (SNAP) participation changes in response to economic conditions and federal and state policies.  (Thanks to graduate student Dan Hatfield for data processing in this new 2010 edition, and to Hanqi Luo and Joseph Llobrera for data processing in earlier versions).

With this Google Gadget, you can track a particular state of interest, or watch all states move together. Of course, the real drama is in the final two years.

A question (not just an answer): How much does a nutritious diet cost?

How much does a nutritious diet cost?

Some say that the high price of healthy food is making us obese and unhealthy. Others wonder how that could be so, because (even with recent inflation) food of all sorts has been comparatively cheap in the United States for many years, due to government policy and technological change in the food system.

The leading source of disagreement about the cost of an adequate diet is different definitions of "adequate," not different price estimates. Your estimate of the minimal necessary cost depends on your opinion on questions like the following:

• whether a high level of meat and dairy is necessary for an adequate diet,
• whether your vision of healthy food includes foods marketed as healthy (organic yogurt, low-fat cereal) or simple basic staples (whole grain rice, cabbage, carrots),
• whether diets should be judged by their adherence to USDA's Pyramid recommendations,
• whether diets should be judged by their adherence to the National Academies' nutrient recommendations, and
• whether you think low-income people can cook at home, or whether instead convenience and restaurant foods are central to your definition of adequacy.

More subtly, your estimate of minimal cost depends on your opinion about whether people can change their diets in order to meet cost and nutrition goals, or whether it is inevitable that any realistic diet closely resembles the current average diet.

Reasonable answers about the cost of a nutritious diet, corresponding to different definitions of nutritious, range from even less expensive than the federal government's Thrifty Food Plan to much more expensive.

No wonder this issue generates a lot of argument! Most people on all sides of this issue leave these key assumptions implicit and unstated. Yet, these assumptions strongly influence conclusions about minimal costs.

In a recent article in the Journal of Consumer Affairs (free abstract, pay site for full article), "Using the Thrifty Food Plan to Assess the Cost of a Nutritious Diet," Joseph Llobrera and I use USDA's Thrifty Food Plan (TFP) framework to clarify the relationship between assumptions and cost estimates for nutritious diets. Let me know by email if your library does not have the journal. There is a related seminar on the Friedman School website. If you would like to play around with these models yourself, see our Thrifty Food Plan calculator. In both the seminar and the calculator, I should have emphasized more strongly that all of the dollars are in 2001 dollars per adult in the household, not adjusted for inflation (if you didn't know this, the amounts would seem unrealistically low).

For some readers, the whole computation will seem beside the point. They may reason that is clearly wrong to set the TFP cost target too low, but harmless to set it too high, so why not just pick the highest estimate? For a number of reasons, I think better food assistance policy comes from trying to choose the right estimate for a minimal cost target, rather than padding the estimate too much.

In the article, we find that the USDA's Thrifty Food Plan cost level can purchase a nutritious diet if (1) you think nutrient constraints (adequate protein, for example) are more important than food category constraints (plenty of meat), or (2) if you think it is reasonable to expect people to drastically change their current consumption pattern. If, instead, you think substantial meat and dairy amounts are essential to an adequate diet and you defer to the current consumption pattern of low-income consumers, you will probably prefer a more generous TFP cost target.

Update: Slate's Daily Bread food business blog has a thoughtful post about this article (gently needling the online presentation as "a little geeky" -- ha!).