Export land model analysis of food production and consumption—Concepts, Database and Definitions.

The Idea

I got the idea that I could quantify the total production and consumption food energies for different countries by putting together two different pieces of information: 

The first piece came from my research I did for the post: The relationship between hunger and petroleum consumption-Part 4.  That is when I became aware of the Food and Agriculture Organization (FAO) and its enormous Statistical Database, FAOSTAT.  FAOSTAT compiles annual records of food production and consumption from 1960 for almost every country in world.  I used the FAO’s compilation of “Food Supply” in my analysis of food supply calories available for human consumption, per capita, versus petroleum consumption, per capita, as previously described here.  FAOSTAT, however, compiles far more detailed statistics than this.  The database includes the production, import, export and consumption quantities of a hundred or different food types, in units of tonnes, for each country, for each year.  The information provided is so very finely grained in detail that it makes it hard to use it to see general trends without some investing some heavy time to re-organize and consolidate the FAOSTAT spread sheets, and so, I set it aside...until now. 

The second piece was finding a few months ago a recent 2011 paper entitled, “Food security and fossil energy dependence: an international comparison of the use of fossil energy in agriculture (1991-2003)” by Arizpe, N., Giampietro, M., Ramos-Martin, J., (“Arizpe”) in, Critical Reviews in Plant Sciences, Vol. 30: 45-63 2011 http://dx.doi.org/10.1080/07352689.2011.554352 (also available here). Arizpe did a survey of several fossil fuel supported energy inputs into agricultural production for 21 countries, for two time periods: 1991 and 2003. 

What really interested me, however, were the first two data columns in Arizpe’s Table 1:

To generate these two columns, Arizpe took the FAOSTAT production and consumptions quantities, for the yearly food balance sheets for these 21 countries, and converted the quantities into units of peta Joules per year (PJ/year).  

Arizpe doesn’t describe how this was done, only providing this brief comment:

The conversion factors used to assess the amount of embodied fossil energy slightly different from those used in the original study of Giampietro et al., 1999, since some data have been updated. For this reason, the original data set used in the CRPS paper of Giampietro et al. (1999) has been recalculated using this set of conversion factors to obtain a better comparability of the two assessments presented in this paper referring to 1991 and 2003.

Nevertheless, this was enough to tell me that “it could be done.”

Definitions—part 1

Jewel’s name stems from Lenedra, whose middle name is Jewel, and from her Grandfather, Yule (spelled "Juel" in his homeland), a Swiss native who settled in Homer, Alaska in 1940. Jewel Biography

NO, NO, NO, not that Jewel—we are talking about Joules, the unit of energy named after James Prescott Joule:

the energy expended (or work done) in applying a force of one newton through a distance of one metre (1 newton metre or N·m), or in passing an electric current of one ampere through a resistance of one ohm for one second

FAOSAT, when reporting food consumption in units of energy, and not tonnes, uses energy units of “large calories” or kilocalories (C, also abbreviated “kcal” in the FAOSTAT tables):

1 calorie = 4.1868 Joules. When the term calorie is used to express amount of energy provided by food or expended during body activities, the term kcalorie or large Calorie is actually meant. 1 joule is approximately equal to 0.238845896628 cal (calorie) (small calories, lower case c) or 2.390 ×10⁻⁴ kilocalorie, Calories (food energy, upper case C)

from FAOSTAT FAQs

So, 1 Joule equals 2.388 x 10^-4 kcals and peta means 10¹⁵, and therefore 1 peta J equals 2.388 x 10¹¹ kcal.

Also, as noted above, FAOSTAT typically reports food in units of 1000 tonnes per item, or groups of similar items.  A tonne here refers to metric tonnes which is 1,000 kg or 2,205 lb.  So 1000 tonnes is equal to 10⁶ kg.

Export Land Model (ELM) Analysis concept

Okay, enough about Jewel Joules; the important point to keep in mind here is that expressing food in terms of its energy content allows all of the different food types produced or consumed to be compared and combined in a meaningful way—their energy content.  That is, I can sum up any country’s yearly total food energy produced, imported, exported, and consumed and compare this between years and between countries.

As soon as I saw Arizpe’s Table 1, I realized that I had the data and means to do an export land model analysis for food production and consumption, analogous to what I did for petroleum production and consumption in the USA and other countries.   Moreover, the FAOSTAT database goes back to 1961 for nearly every country in the world, which is even more extensive than the BP statistical review for petroleum.

“Export land model” (ELM), as best I can tell, was a phrase coined by Jeffery Brown to describe the phenomenon, originally observed by Matt Simmons, of decreasing exports of oil from a country as its oil production peaks and declines and as its domestic consumption  increases.  This can have the effect of a double-exponentially declining rate of exportable oil (for more background see e.g.,   An Export Land Model Analysis for the USA-Part 1).   A rapidly declining rate of exportable oil, in turn, can cause major problems both for the countries that transition from oil exporters to ex-oil exporters, and for the countries that relied upon those exports (see e.g., Survey of Oil Exports from the Middle East; Survey of Oil Exports from North Africa; and Relating Per Capita GDP to Petroleum Consumption and Exports for the MENA Countries).

Just as the net amount of oil available for export from country equals that country’s total production, minus its domestic consumption, so too should a country’s net food exports equal its food production minus domestic food consumption...plus food imports. 

That last term, “plus imports” is somewhat different than the situation for an ELM analysis for oil in that most oil exporting countries don’t simultaneously export and import large quantities of oil.  For instance, Canada is one of the few countries that I know of that both export and imports large quantities of oil or at the refinery products of oil (Canada—Petroleum Superpower or Super-slave?).  However, that is not the case for food, where most countries import and export food.  Fortunately, FAOSTAT presents both food imports and exports, so it is possible to calculate the net imports or net exports, in terms of food energy, for each country over time. 

Nevertheless, similar to an oil exporter who becomes a ex-exporter, I expect that when a net food exporting country becomes a ex-food exporter, this creates major economic and practical problems for both the ex-exporter and the countries that previously relied on the food that was being exported. 

To some extent, a country can deal with less oil by reducing its energy consumption, e.g., by not wasting petroleum of useless consumption, or by transitioning to another source of energy.   That is, if a country can’t import all of the oil it wants, then people can walk, bike work at home, take mass transit, by more efficient cars, heat their homes by other means etc... 

My hunch is that the margin for cutting back on food consumption, by reducing waste, is much smaller, and, there is no substitute for food.   That is, if a country can’t import all of the food it needs, then people die.

One practical problem with doing an ELM analysis for food, however, is that unlike the BP statistical review, the total food energy production and consumption data is not made available in a nice tidy spreadsheet.  Rather, as you are about to see, for each year, and for each country, there is separate spreadsheet with a column of +100 food items listed in units of 1000s of tonnes of food items.  The rest of this post and the next post is about how I can get from this raw data to where I want to be.

Let’s take Arizpe’s second country on the list, Australia, and work through it. 

A Worked Example—Australia

Alright, orientation is over, let’s get on with it.  If you can follow the calculations below, which are all long but simple arithmetic, then you will be able to understand how I derived my country-by-country data, in the posts to come.

Here’s the first portion of the food balance sheet for Australia for 2003, as downloaded from FAOSTAT (after adjusting some column widths in the sheet and getting rid of columns that provide repetitive un-useful or no information after the first row: country, country code, year, total population 19,904,000).  You can down load it yourself here at the Food Balance Sheet (FBS) portal  here; just select "Australia," "2003," "download format excel," "codes," and then hit the download button and save the file.

After spending an inordinate amount of time trying to make this nicely fit within WORD, I gave up and printed this out as a landscape page and then scanned the page as a .jpeg image.

Keep in mind that the full spread-sheet has 118 rows that presents major categories of plant and animal food items for Australia.

There is one column of data that I took out that does have useful internal information—the food item code—I’ll get back to that later on, but for the time being let’s just focus on what the food information provided in this table means.

Definitions—part 2

A valuable reference in understanding what the columns in this table mean is the FOOD BALANCE SHEETS A handbook,  FOOD AND AGRICULTURE ORGANIZATION OF THE UNITED NATIONS, Rome, 2001 (hereafter, “Handbook”; also available here). Here are some definitions and commentary from the Handbook (p.11-15) for the differ headings in the order they appear in the table, with some annotations on my part:

Production:

Unless otherwise indicated, production is reported at the farm level for primary crops (i.e., excluding harvesting losses for crops) and livestock items and in terns of live weight (i.e. the actual ex-water weight of the catch at the time of capture) for primary fish items. Production of processed commodities relates to the total output of the commodity at the manufacture level (i.e. it comprises output from domestic and imported raw materials of originating products).

Import:

... all movements of the commodity in question into the country as well as of commodities derived there from and not separately included in the food balance sheet. It, therefore, includes commercial trade, food aid granted on specific terms, donated quantities, and estimates of unrecorded trade. As a general rule, figures are reported in terms of net weight, i.e. excluding the weight of the container.

Stock variation:

... changes in stocks occurring during the reference period at all levels from production to the retail stage, i.e. it comprises changes in government stocks, in stocks with manufacturers, importers, exporters, other wholesale and retail merchants, transport and storage enterprises, and in stocks on farms.... Increases in stocks of a commodity reduce the availability for domestic utilization. They are therefore indicated by the - sign and decreases in stocks by the + sign since they increase the

available supply

Export: 

... all movements of the commodity in question out of the country during the reference period. The conditions specified for gross imports, under 3 above {i.e., Imports}, apply also to exports by analogy. A number of commodities are processed into food and feed items. Therefore, there is a need to identify the components of the processed material exported in order to arrive at a correct picture of supplies for food and feed in a given time reference period.

Domestic Supply:

Production + imports - exports + changes in stocks (decrease or increase) = supply for domestic utilization.

Feed:

... amounts of the commodity in question and of edible commodities derived there from not shown separately in the food balance sheet (e.g. dried cassava, but excluding by-products, such as bran and oilcakes) that are fed to livestock during the reference period, whether domestically produced or imported.

Seed:

.... all amounts of the commodity in question used during the reference period for reproductive purposes, such as seed, sugar cane planted, eggs for hatching and fish for bait, whether domestically produced or imported

Processing:

... amounts of the commodity in question used during the reference period for manufacture of processed commodities for which separate entries are provided in the food balance sheet either in the same or in another food group

Other Utilization:

... quantities of the commodity in question, consumed mainly by tourists, are included here... quantities of the commodity in question, ... as well as the amounts of the commodity in question used during the reference period for the manufacture for non-food purposes (e.g. oil for soap). Also statistical discrepancies are included here.

Food:

.... the amounts of the commodity in question and of any commodities derived there from not further pursued in the food balance sheet that are available for human consumption during the reference period.... The amount of food actually consumed may be lower than the quantity shown in the food balance sheet depending on the degree of losses of edible food and nutrients in the household, e.g. during storage, in preparation and cooking (which affect vitamins and minerals to a greater extent than they do calories, protein and fat), as plate-waste, or quantities fed to domestic anjmals and pets, or thrown away.

Food Supply:

Per caput supplies in terms of quantity are derived from the total supplies available for human consumption by dividing the quantities of the food element by the total population actually

partaking of the food supplies during the reference period...

For the purpose of calculating the caloric value and the protein and fat content of the per caput food supplies, the choice of the appropriate food composition factors is very important. .... The nutritive factors can be obtained directly from national food composition tables. These tables give the nutritional composition of food per 100 grams of edible portion..... The conversion is made by applying waste/refuse factors to the nutritive composition in term of edible portion. The resulting per caput total nutritive values are usually expressed on a daily basis. In the absence of food composition tables prepared by appropriate national institutions, use can be made of FAO's food composition factors as shown in the appendix.   

For calories, protein and fat, a grand total and its breakdown into components of vegetable and animal origin is shown at the beginning or the end of the food balance sheet.

That second last paragraph, about the use of the food composition tables is quite fundamental to what I want to do here, and quite a problem to overcome, as you will see.   

Worked Example—Back to Australia

Okay, with these definitions in hand, I want to take a look at that table for Australia again, and show by way of example, just where some of these numbers are coming from. 

In the worked example, I focus mainly on one food quantity, wheat.

Whoa! ... this looks like something a group of engineers on their lunch break would draw—lets just take it one step at a time.   

Let’s start in the upper left corner, the red arrow is just pointing out that the population of Australia in 2003 (shown in one of the columns I deleted, since that was the only entry in the column) was 1.99x10⁷. 

Now let’s move straight down to the red arrow pointing to the total production of Cereals (excluding beer).  That total of 41485 thousand tonnes (i.e., 4.1485x10⁴x10³x10³ kg) is the sum of Wheat, Rice, Barley, Maize (i.e., Corn), Rye, Oats, Millet, Sorghum, and other cereals as circled in the vertical red bubble.  Similarly, Starchy Roots (Total) is the sum of Cassava, Potatoes and Yams produced in Australia.  And so it goes, down that column for the other food quantities products.  And, so it also goes, for the quantities of food imported, stock variations and food exported, across the next three columns.

Now let’s move back up to the red arrow pointing to the Domestic supply of wheat.  As shown in the connected bubble, the total of 5750 thousand tonnes is derived from (total production) +(imports)-(stock)-(exports). 

Immediately to the right, is another red arrow pointing to the Food available for human consumption, which is derived from:  (Domestic Supply) – (Feed) – (Seed) – (Processing) – (Other uses).

Immediately to right of Food available, is yet another red arrow pointing to Food Supply quantity 71.8 expressed as kg per capita per year, which is simply derived from: (Food) / (Population).

Finally to the right of Food Supply quantity we come to the Food Supply energy, expressed as kg per capita per year.  This is where the conversion factor comes in.  You have to know the kcal per unit weight used by the FAO, for Australian wheat, to convert the annual per capita wheat to per capita wheat energy available per day which equals 584 kcal per person per day. 

I don’t know what that conversion factor is and the FAO doesn’t report it (or at least I haven’t found it), but, I can derive it, as shown in the large red bubble leading away from Food Supply quantity and Food Supply energy.

Specifically, since 71.8 k/py of wheat is equal to 196.7 g/pd of wheat, and that in turn, equals 584 kcal/pd, the conversion factor is 2.97 kcal/gm of wheat.  That is, the FAO has assumed that 1 gram of Australian wheat, on average, contains 2.97 food kcals of food energy.

This same calculation could be performed to derive the conversion factor used for each of the food categories in the table (blue numbers on the far right hand side)—more on that in a moment.

Finally, let’s finish our tour around this table by going to the upper right hand corner, and looking at the Grand Total of the Food Supply energy availability, equal to 3134 kcal/pd.  The Grand Total equals: (Total Plant Food Supply Energy) + (Animal Food Supply Energy).  The Total Plant Energy in turn equals: (Total Wheat Supply Energy) + (Total Starchy Roots Supply Energy) +  .... for all the different plant food types.  A similar formula applied to Animal derived food.

As shown in the red bubble at the top of the table, one can use the Grand Total of the Food Supply Energy availability to calculate the total human food supply energy for all of Australia for 2003—which is 95.3 PJ. 

The problem is that, for my purposes, knowing the total amount of food energy for human consumption does not tell the full story.  This is somewhat analogous to only knowing the total amount of petroleum consumed for domestic automobile use, but not knowing the total amount produced, imported/exported or petroleum used for other purposes.  No, what I want is total food energy production, total food energy imports, total food energy exports and the total domestic food energy supply. 

However, to derived these quantities I need the energy conversion factors for all of the food categories that contribute to these different quantities of food.

The Case of the Missing Conversion Factors

This immediately raises a problem illustrated in the above marked up table for Australia.  It is apparent that there are several food categories that make large contributions to total production, imports, exports and domestic food supply, but which do not contribute to the human supply.  For instance, just under the heading of cereals, large quantities of Barley, Millet, Sorghum and other cereals contribute to the animal Feed category, but make zero (blue bubbles) contribution to the human food supply.

However, if a particular food doesn’t contribute to the human food supply, then the FAO doesn’t include this in its food energy supply calculation.  And, if there is no food energy supply reported, then I don’t have a conversion factor for that particular food type.  That is the case for Barley, Millet, Sorghum and several other food types in the full food balance table.

So how to derive the missing conversion factors?

The answer is involved, so I will have to devote an entire post to it.  See you next time.