Posts Tagged statistics

Should Americans be spending more on food?

Michael Pollan echoes a popular foodie talking point:

For the majority of Americans, spending more for better food is less a matter of ability than priority. We spend a smaller percentage of our income on food than any other industrialized society; surely if we decided that the quality of our food mattered, we could afford to spend a few more dollars on it a week–and eat a little less of it. (187)

This amounts to a textbook case of statistical abuse. When Pollan refers to the small percentage of income Americans spend on food, that doesn’t say anything about the majority of Americans at all. It is, instead, a statement about our collective income (as compared to the incomes of other societies).

To understand this distinction, consider somebody who makes a million dollars a year. To spend 9.9% of that income (a figure Pollan cites in the next paragraph), this person would need to buy $99,000 worth of food in a year. Even assuming this person supports a large family and buys the kind of food that Michael Pollan says we should be eating, that’s a huge amount of food! Thus, people with large incomes will tend to spend a small percentage of their income on food. Moreover, because these people make so much money, their spending patterns are weighted more heavily in the figure than a comparable number of people at the lower end of the income scale*. The point here is that when we look at the percentage of income spent on food, that number is pushed downward by income inequality. Since the United States has a relatively high level of income inequality, this distortion is particularly pronounced in our case.

To look much deeper into the statistics, it would be helpful to be able to look at some actual data. Unfortunately, Pollan doesn’t give a citation on this claim, but when foodies compare food spending among countries, if a source is provided, it’s the USDA’s data on “Expenditures on food and alcoholic beverages that were consumed at home, by selected countries.” This looks at food spending as a percentage of household final consumption expenditures (which is certainly not the same as income). I don’t know for sure that this is Pollan’s source, but since he’s given his endorsement to a similar use of that data before, I’ll say a few things about it here. One weakness of the data shows up right in the title: it’s referring only to meals eaten at home. That means it’s not even clear that it includes the same number of meals in different countries. Looking at the data for 2006, for instance, if (and this is strictly hypothetical because I haven’t seen any data on the number of meals eaten at home) Americans devoted 6.6% of their spending to 14 meals eaten at home per week at home and the British devoted 8.6% of their spending to 18 meals per week, the British and the Americans would actually be putting about the same percentage of their spending toward each meal. Again, this is hypothetical, but if foodies are right when they say that Americans eat too many meals in restaurants or at their desks at work, this could be important.

Another problem with the data lies with the way that household final consumption expenditures are defined. Americans tend to have lower taxes than Europeans and pay for some things out of their pockets that are paid for with taxpayer money in Europe. For example, things like health care, education, and transportation are more heavily subsidized by taxes in Europe than in the United States. This means that those expenses are largely included in household final consumption expenditures for the US, but less so for the European countries. This, in turn, serves to decrease the percentage of household final consumption expenditures which Americans spend on food.

None of this is to say that Americans are spending the ideal amount on food. Nor do I mean to defend everything on which Americans spend money. My point is simply that the comparison of the US to other countries doesn’t support Pollan’s argument very well because the data reflect many differences among the various countries. I don’t doubt that there are significant externalities in the production of food or that food would cost more if these were reflected in the price. However, I don’t see much to be gained by throwing around misleading statistics to overstate the extent to which people are able to pay those costs.

*To see why, consider a fictitious country with only two people, one making $10,000 and spending $3,500 on food and the other making $150,000 and spending $9,000 on food (these are approximately at the averages of the top and bottom income quintiles for the United States for recent years). One person is spending 35% of income on food and the other is spending 6% of income on food. The population as a whole spends $12,500 of its $159,000 income on food. That’s only 7.9%, much closer to the wealthier resident’s figure.


Comments (9)

A radical simplification of our diet?

Michael Pollan argues in In Defense of Food that our diets are less diverse than they once were. He explains,

Today [corn, soy, wheat, and rice] account for two thirds of the calories we eat. When you consider that humankind has historically consumed some eighty thousand edible species, and that three thousand of these have been in widespread use, this represents a radical simplification of the human diet. (117)

This is a rather egregious abuse of the numbers. Yes, three thousand species is many more than four species, but these numbers are measuring very different things.

I’ll start with the smallest problem. As Pollan notes, much of the corn and soy is actually fed to animals to produce meats and other animal products for human consumption. Yet for some reason, he doesn’t bother to include the animal species in his count here.

A bigger issue is that when he refers to the number of species that have been historically eaten, he seems to be including species that have been eaten by various cultures across different time periods. This means, for instance, that it includes both the American bison eaten by the Native Americans of the Great Plains and the cassava that has long been a staple in South America. Of course, geographical considerations dictate that it’s highly unlikely that many people have regularly eaten both of these species, so the “eighty thousand edible species” is misleading because it tells us nothing about what any particular group of people have historically eaten.

It’s also deeply misleading to use the four species that provide most of our calories as an indicator of the diversity of our diet. The fact is that some edible species don’t have many calories. If you wanted to get more than a third of your calories from the leafy vegetables that Pollan argues we should be eating more of, you’d better be prepared to wolf down more than three pounds of greens per day. And we could easily add a few dozen species to the list just by visiting the spice aisle of a grocery store, even if there are very few calories to be found there.

So is the American diet unusually homogeneous? I don’t have the time or resources to make that determination, but the evidence certainly isn’t in the data that Pollan cites. However, there is evidence that we are not alone in our dependence on a small number of crops. For example, the United Nations Food and Agriculture Organization estimated that about thirty years ago rice alone contributed 68% of calories to South Asian diets.

Leave a Comment

%d bloggers like this: