International Differences in Food Purchases?" which appears in the March 2014 issue of the American Economic Review (104:3, pp. 832-867). The AER isn't freely available on-line, but many readers will have access through library subscriptions. For me, the paper is a useful exercise in showing how statistics about averages can still paint a picture of real people behave.
The authors have information from household surveys in which thousands of households in each country recorded all their purchases of food for consumption at home. (One limitation of this data is that it doesn't look at food consumed in restaurants.) As they write: "We have information on quantities, prices, and characteristics of the products purchased at the level of the individual food product, as defined by the barcode or what is called the Universal Product Code (UPC) in the United States. The characteristics include nutritional characteristics such as calories, proteins, fats, and carbohydrates, as shown on nutritional labels." Here's are some overall patterns, expressed in terms of average per adult per day:
For example, note that Americans spend less per person on food each day, and also consume more calories. This pattern is of course no surprise to economists, because food prices in the U.S. are on average lower, so a basic economic model would expect quantity demanded to be higher. Notice also that compared with the French, Americans on average consume less fat and less protein, but more carbohydrates.
With their detailed data set, Dubois, Griffith, and Nevo can also break down these differences further, into the expenditure categories shown below. They write: "The UK and US expenditure patterns are more similar, while the French numbers are different. The average French household spends less on processed food, such as drinks and prepared foods, and more on basic ingredients such as meats, dairy, fruits, and vegetables, both in dollar terms and as a fraction of overall expenditure. The average UK and US household spends less than French households on meats, and the United Kingdom spend more on grains, while the average US household spends less on dairy and more on drinks and prepared foods."
They can also look not at what is spent on categories of food, but also at the quantities of food purchased for home consumption. This table shows quantities consumed, and share of calories, calculated per adult for three-month period. Americans get a higher share of their (higher level of) calories from prepared foods and from drinks. Apparently, the French are much more likely to purchase water in the drinks category. The typical American consumes a smaller quantity of vegetables than the French or the British, and a smaller quantity of dairy. The Americans and French consume about the same quantity of meat, both more than the British. The authors write: "Generally, the French tend to purchase less processed food, such as drinks and prepared foods, and more basic ingredients such as meats, dairy, and vegetables. This is especially true compared to the US purchasing patterns. The UK and US purchasing patterns are more similar, but even here there are differences, with the average UK household consuming more vegetables, grains, and dairy, and the average US household consuming
more meat and drinks."
One final way to slice the data is to look at the nutritional content of these various categories. As noted a moment ago, "drinks" means something different depending on whether people are more likely to buy water or full-calorie sweetened beverages. In the chart below, for example, the typical 100 grams of drink purchased by the French has 27 calories from carbohydrates, while the typical 100 grams of drink purchased by an American has 69 calories from carbohydrates. They explain: "For example, the meat products that US households buy have on average much more fat and carbohydrate than the meat products that French households purchase, which are more protein intensive. Another example: we saw above that the higher fraction of calories from prepared foods in the United States is consistent with prepared foods in the United States being more calorie dense relative to UK prepared foods. The difference in calories from prepared foods seems to come from the differences in carbohydrates and fats. Drinks are also much more carbohydrate intense in the United States than in the
United Kingdom, and even more than in France."
The main focus of the paper is not just to report these patterns, but to estimate an economic model of demand for different products, which lets the authors tackle the question of what causes these kinds of differences across countries. Differences in food prices across countries? Differences in the characteristics of the available food? Differences in preferences and eating habits? Here are a few of their findings:
"Price differences mostly explain the large difference in caloric intake between the average French and US household. However, nutrient characteristics are important when comparing to the United Kingdom, and differences in preferences and eating habits are generally quite important, and in some cases can offset the influences of the economic environment. For example, we find that UK households have healthier purchasing patterns than US households despite the prices and product offering they face, not because of them. ...
The French have the highest relative preference for fats and proteins in dairy and meat. And the Americans have the highest preference for proteins in prepared food, and the lowest for fats in prepared foods. The ratio of the fats coefficients to the carbohydrates coefficient is the highest in France and the lowest in the United States, while the ratio of proteins to carbohydrates tends to be higher in the United States compared to France and the United Kingdom (this is mostly driven by the coefficient for the prepared category)."Of course, given the public health issues posed by obesity, an obvious question is the extent to which public policy might seek to shape eating habits. But for now, I would just emphasize that the common food choices even among high-income countries vary by more, and in some different ways, than I would have guessed.