Journal Issue: Childhood Obesity Volume 16 Number 1 Spring 2006
How Markets May Have Contributed to the Rise in Childhood Overweight
Several strands of research investigate how markets may contribute to increased calorie consumption, sedentary lifestyles, or overweight and how changes in those markets may have contributed to the recent rise in childhood overweight. The problem for researchers is not figuring out what could have caused the rise in childhood obesity; the problem is that too many things could have caused it. James Hill and several colleagues calculate that the rise in obesity in the United States could have been caused by a daily surplus of just 15 calories for the median person, with 90 percent of the population increasing their intake by 50 or fewer calories a day.5 To put this in perspective, the rise in weight for the median person could have been caused by consuming an extra three tablespoons of skim milk or walking 120 fewer yards each day. It will likely be impossible to determine which changes are responsible for such a small increase in daily calorie surplus, but I will consider several possible contributors.
Changes in the Cost of Food and Food Preparation
The most obvious contributor to the increasing calorie surplus is falling food prices. Between January 1980 and January 2005, the real price of food fell 13 percent.6 One study attributes 40 percent of the recent rise in weight to lower food prices.7
Changes in the “opportunity cost” of time spent cooking may also have affected eating patterns. Strictly speaking, the opportunity cost of a person's time is the value of that time devoted to the next best available alternative, but in practice it is often measured by the wage rate. All else being equal, if someone's wage rate rises, he or she will likely spend less time cooking and will instead use prepackaged foods that require less preparation time or will eat food prepared by others, such as restaurant meals or “takeout.” It is also possible that he or she would simply consume less food as a result of spending less time cooking.
How much wages have changed over the past twenty-five to thirty years varies by the wage earners' educational attainment. Real wages for high school dropouts have fallen, those for high school graduates have remained roughly constant, and those for college graduates have risen considerably.8 These wage dynamics imply that the time cost of cooking rose for college graduates and fell for high school dropouts. To my knowledge, data on time spent cooking have not been sorted by education, but for the U.S. population as a whole, the number of minutes spent each day preparing meals fell from forty-four in 1965 to thirty-two in 1999.9
While changing wage rates affected the opportunity cost of time spent cooking, technological change may have reduced the time required to prepare some foods. David Cutler, Edward Glaeser, and Jesse Shapiro argue that innovations in food processing, preservation, and packaging made it possible for food to be mass prepared far from the place of consumption and to be consumed with less time cost. These innovations contributed to a shift away from home-cooked meals toward processed food, thus increasing obesity. In support of their argument, the researchers show that consumption of mass-produced foods increased the most, that people most able to take advantage of these technological changes had the greatest increases in weight, and that obesity is greatest in countries where people have the greatest access to processed food.10
Changes in the share of women who work also affected time spent cooking. Over the past three decades, the labor force participation rate of women with children younger than age eighteen rose from 47 to 72 percent, with the largest increase among mothers with children younger than age three.11 The increased work time may have resulted in the increased use of prepackaged foods or of food consumed away from home. One study calculates that during the past three decades the increase in mothers' average weekly hours at work explains 12 to 35 percent of the increase in childhood obesity in families of high socioeconomic status.12
During this same period Congress reformed the nation's welfare system, in the process giving poor single mothers an economic incentive to spend less time cooking. With some exceptions, the 1996 welfare reform law required single mothers to work in the labor force to receive cash welfare benefits.13 Thus even though falling real wages lowered the opportunity cost of cooking for high school dropouts, poor single mothers had good reason to spend more time in the labor force and less time in household production.
To better establish the link among changing costs of time, time spent cooking, eating out, and childhood obesity, researchers must answer several questions. First, they must find out how different groups of people responded to the changing costs of time. For example, do college graduates and welfare-eligible women spend less time cooking and high school dropouts spend more time cooking? Second, they must determine whether changes in cooking time led college graduates and welfare-eligible women to use more prepackaged foods or restaurant food and led high school dropouts to use less. And, third, they must prove the link between the consumption of prepackaged and restaurant foods and childhood obesity.
Another possible contributor to obesity is the price of energy-dense foods, such as those containing fats and sugars, relative to that of less energy-dense foods, such as fresh fruits and vegetables. Adam Drewnowski and S. E. Specter calculate that, on a per-calorie basis, energy-dense foods are cheap whereas foods low in energy density are much more expensive.14 However, a Department of Agriculture (USDA) study suggests that less energy-dense foods can still be quite cheap in absolute terms; it calculates that a person can satisfy the USDA's recommendation of three daily servings of fruit and four servings of vegetables for just 64 cents a day. The study also concludes that 63 percent of fruits and 57 percent of vegetables were cheapest in their fresh form.15
Further research must determine whether the relative price of energy-dense foods has fallen in the past several decades. A quick comparison of the various consumer price indexes indicates that between January 1989 and January 2005, the real price of fruits and vegetables rose 74.6 percent while that of fats and oils fell 26.5 percent and that of sugars and sweets fell 33.1 percent. Thus energy-dense foods have become considerably cheaper, relative to less energy-dense foods, in the past fifteen years.16
Changes in Where Americans Eat Their Meals
Perhaps because of the increasing opportunity cost of time for college graduates or the movement of women into the labor force, Americans are eating more meals away from home today than they did thirty years ago. Between 1977 and 1995, the share of total calories consumed away from home rose from 18 to 34 percent, the share of meals consumed away from home rose from 16 to 29 percent, and the total share of food dollars spent away from home rose from 26 to 39 percent.17 From 1994 through 1996, children consumed 32 percent of all their calories away from home: 10 percent in fast-food restaurants, 9 percent in schools, 4 percent in restaurants, and 9 percent from all other sources, such as vending machines and other people's houses.18
The move toward food away from home came just as this food itself (not the time costs of preparing it) was becoming more expensive relative to food at home. Between January 1980 and January 2005, the real price of food at home fell 16.2 percent while the real price of food away from home fell only 5.1 percent.19 Eating out became more common even as it was becoming relatively more expensive.
The distinction between food at home and food away from home is important because consumers typically have less information about the calorie content of foods they eat away from home. Relative to food at home, food away from home has on average lower fiber and calcium density, similar sodium density, and higher cholesterol density.20 Nutritional trends both for food at home and for food away from home are promising: in both, the densities of fat, saturated fat, and cholesterol have declined, and the density of fiber has increased slightly, though away-from-home foods have improved less than food at home.21 Further research must explore whether the move toward consuming more food away from home and eating less at home has caused an increase in childhood obesity.
Changes in Portion Sizes
Portion sizes of certain foods have increased since the 1970s. One study of portion sizes looks at package labels and manufacturers' information and concludes that the portion sizes of “virtually all” the packaged foods and beverages it examines have increased during the past three decades.22 Such a finding is important because several experiments have documented that when people are served larger portions, they consume more calories.23 The portion size effect is first detectable in children five years old.24
The increase in portion sizes, combined with people's tendency to eat more when served larger portions, implies that the amount of food consumed at one sitting has increased. Smiciklas-Wright and her colleagues use the Continuing Survey of Food Intakes by Individuals to study the quantities consumed at one sitting and find significant increases for about a third of the 107 foods they examine. The study does not specify whether these items were ones packed in larger portions by manufacturers. They find significant decreases in amounts consumed for six other foods.25
One limit of the research on portion sizes and calorie intake is that the increases in portion sizes in experiments do not match the increased portion sizes in the market. For example, some studies experimentally increased portions of macaroni and cheese, but the Smiciklas-Wright team found that the portions of macaroni and cheese that people reported consuming fell 17 percent between 1989–91 and 1994–96.
Whether people eat more when they are notified of the increase in portion size is unclear. In the market, some food manufacturers take pains to emphasize the increased size of their products—for example, the Big Gulp, Monster Thickburger, and “supersized” meals. Moreover, the increases in calorie consumption documented in these experiments (30 to 50 percent) are far larger than the small increase in daily calorie surplus that caused the rise in obesity. Because these experiments typically last no longer than a few meals or days, it is also unclear whether in the longer run people adapt to the larger portion sizes and return to consuming their normal amount. For example, suppose every day were Thanksgiving. You might eat large portions the first day, but you probably would not continue to eat “Thanksgiving-sized” meals every day. At some point you might return to your previous level of caloric consumption. Longer-term research is needed to determine the long-run effect of portion size, as well as what the effect is when consumers are notified of the larger portions.
Changes in Farm Policies
Agriculture policy may contribute to obesity by promoting lower prices and greater production of certain commodities. From 1933 to 1995, price supports kept the prices of wheat, corn, soybeans, oats, and other commodities above their free-market prices in the United States; the government purchased excess supplies to bolster prices. The 1995 Farm Bill, however, reformed the system. Rather than subsidizing farmers by keeping prices artificially high, the government now gives farmers a production subsidy, or a payment based on their historic production. In other words, the new law completely decoupled subsidies to farmers from production, and consumers switched from paying above-market to below-market prices for agricultural commodities.26
Trade policies still keep some commodities' prices high. A system of quotas and tariffs, for example, keeps the U.S. price of sugar above the world price.27 Agriculture policy has lowered the prices of other sweeteners, however. In particular, farm policy has been criticized for subsidizing the production of corn and, thereby, of high-fructose corn syrup, which is now common in soft drinks, fruit juices, jelly, and other foods.28
The USDA estimates that production subsidies increased the land under cultivation by 4 million acres and lowered the price of wheat by seven cents a bushel, of corn by nine cents a bushel, and of soybeans by forty-nine cents a bushel.29 A second study figures that direct payments to farmers boost agricultural output 4.39 percent.30 And a third calculates that policy changes in the 1995 Farm Bill and thereafter increased production of grains and soybeans 4 percent and lowered prices 5 to 8 percent.31 Of course, consumers do not buy and consume bushels of wheat and soybeans directly; instead, they eat food manufactured from these crops. One study estimates that for every 1 percent decrease in commodity prices, food prices decline 0.25 percent.32 Another study determines that agricultural subsidies explain about 1 percent of the increase in obesity over the past two decades.33
When price supports were still a mainstay of U.S. farm policy, Washington established “checkoff” programs to increase demand for covered commodities, thereby lessening the excess supply that it was obliged to purchase. The checkoff programs required producers to contribute a fixed amount per commodity unit sold to a nonprofit organization that used the money to increase consumer demand for that commodity by researching new uses for it and advertising those uses to consumers. Commodity checkoff programs, which have persisted even after price supports were removed, now spend $1 billion a year to increase demand for their products.34
Substantial checkoff resources go for advertising, and they have a considerable impact. Checkoff-funded advertising campaigns include “Ahh, the Power of Cheese,” “Beef— It's What's for Dinner,” “Pork—the Other White Meat,” and “Got Milk?” Noel Blisard estimates that the generic advertising of milk totaled $29.8 million between October 1995 and September 1996, raising milk sales by 1.4 billion pounds, or 5.9 percent, while generic advertising of cheese funded by checkoff programs increased sales of cheese by 62.7 million pounds, or 2.8 percent.35 Although advertising funded by checkoff dollars increases consumer demand, collecting checkoff funds from producers is a form of tax, which raises producers' costs and lowers the quantities sold. On net, however, the checkoff programs increase commodities' sales.36
Whether checkoff-funded advertising encourages consumers to spend more money overall on food or simply redirects their fixed food budget to the advertised products is unclear. Brenda Boetel and Donald Liu find that beef and pork advertising is “beggar thy neighbor” in the sense that beef advertising decreases the demand for pork and vice versa. They conclude that beef and pork producers would be better off if they agreed to reduce their generic checkoff-funded advertising campaigns.37 Checkoff funds are also used to increase commodities' sales by developing new menu items for fast-food restaurants, such as the McRib pork sandwich for McDonald's and Insider Pizza (which used one pound of cheese per medium pizza) for Pizza Hut.38
Although analysts have linked various farm policies to the increased sales of certain crops and to the development of menu items for fast-food restaurants, documenting the effect these programs have on obesity itself requires more research.
Increased Food Advertising to Children
The food industry spent $7 billion on advertising in 1997, more than any other industry except automobiles. Two-thirds of that advertising was by food manufacturers, 28 percent by the food service industry (mostly fast-food restaurants), and 8 percent by food stores.39 It is estimated that the number of television commercials viewed each year by the average American child doubled from about 20,000 in 1970 to 40,000 around the year 2000.40 But although children are viewing more commercials, the length of the average commercial has fallen.41 In addition, Roland Sturm finds that the average time children spent watching TV fell 23 percent, or four hours a week, between 1981 and 1997.42
One study, which analyzes the food advertisements aired during children's Saturday morning television programming, concludes that if a child consumed only the advertised food, his diet would not be consistent with U.S. dietary recommendations.43 Gerard Hastings and several colleagues conclude that over time advertisements for fruits and vegetables have disappeared and have been replaced by ads for fast-food restaurants, breakfast cereals, soft drinks, and snacks.44
Two broad reviews of the experimental research on food advertising and youth diets conclude that there is very mixed evidence on whether television advertising causally affects children's diets.45
Because of several limitations of this research, including the use of small, nonrepresentative samples, making it difficult to draw inferences about larger populations from their findings, the American Academy of Pediatrics and the Institute of Medicine have called for more research on whether food advertising affects the diets of American youth.46
Differences in Local Availability of Food and Exercise Opportunities
The prevalence of childhood obesity shows clear racial disparities. In 1998, 21.5 percent of African American children and 21.8 percent of Hispanic children but only 12.3 percent of white children were overweight.47
Whether there are also economic disparities in the prevalence of childhood obesity within race and ethnic groups is less clear. R. P. Troiano and K. M. Flegal find no significant correlation between income and childhood obesity for African Americans or Hispanics. They find weak evidence that higher-income white adolescents are less likely to be obese, but they warn this evidence must be interpreted cautiously because of small sample sizes.48
Some researchers have argued that racial and socioeconomic disparities in weight may be due in part to differences in the availability of food and exercise opportunities. One study finds that supermarkets are three times more common in census tracts with home values in the highest quintile than in tracts with homes in the lowest quintile and four times more common in predominantly white than in predominantly black neighborhoods.49 Conversely, smaller grocery stores and convenience stores without gas stations are more common in lower-wealth and predominantly black neighborhoods. Another study finds that the probability of at least one supermarket located in an urban zip code is higher where income is high and the poverty rate is low. But it finds no link between that probability and the share of residents who had graduated from high school.50
Some researchers argue that these disparities are related to obesity because the proximity of such businesses is correlated with diet quality. Kimberly Morland, Steve Wing, and Ana Roux compare food consumption reported in questionnaires to food outlet availability in the census tract of residence and find that blacks living in census tracts with supermarkets are more likely to meet U.S. Dietary Guidelines for fruits and vegetables, total fat, and saturated fat. Findings for whites were generally not statistically significant.51
Joel Waldfogel finds that the population's educational and racial composition is correlated with restaurant density. Controlling for population size within a zip code (which may itself be correlated with both restaurant density and the educational and racial composition in an area), when the share of blacks and Hispanics is higher, there are fewer sit-down restaurants but more of certain fast-food restaurants.52
Analysts find similar patterns for active recreational facilities, such as public beaches, pools, youth centers, parks, YMCAs and YWCAs, dance studios, and athletic clubs. Penny Gordon-Larsen and several colleagues find that all major categories of such facilities are inequitably distributed across census block groups by socioeconomic status, minority population, and education. They find that the presence of just one such facility per census block group is associated with a 5 percent lower probability of overweight.53
These studies are observational, not based on randomized experiments, and thus include an unknown degree of what researchers term “selection bias.” Supermarkets and health clubs may open outlets in places where people are most interested in them—where they can earn the highest profits—and not in areas with low demand. Likewise, when people choose where to live, they may consider the retail options available nearby. People who cook meals from scratch may find it more attractive to live near a full-service supermarket, and exercise buffs may want to live near parks and gyms. Because of such self-selection, correlations between diet and supermarket proximity and between physical activity and proximity to athletic facilities may arise even without a causal relationship. For this reason, these research findings cannot be interpreted as causal or evidence of discrimination.
A related research area claims that new suburbs and developments contribute to obesity because they lack sidewalks and places to which people could walk or because they require long commutes.54 But children's physical activity appears uncorrelated with such neighborhood characteristics as the availability of local facilities and safety.55
Over the same period that the prevalence of obesity has been on the rise, the prevalence of adult smoking has fallen—from 33.2 percent in 1980 to 21.6 percent in 2003.56 The share of high school students who smoke has also declined, from 27.5 percent in 1991 to 21.9 percent in 2003.57 The trends in smoking and obesity may be related. A surgeon general's report reviewed fifteen medical studies and found that between 58 percent and 87 percent of those who quit smoking gained, on average, four pounds.58 One study finds adults' weight is positively correlated with the local price of cigarettes (the higher the price, the higher the weight); the correlation with weight is roughly the same as that of grocery prices. The study finds no correlation of weight with clean indoor air laws that restrict smoking.59 A second study, however, faults the first for not taking into account time trends in smoking and for focusing on cigarette prices instead of cigarette taxes; after making the necessary corrections, that study finds that higher cigarette taxes are associated with lower weight.60