Cheap Food, Societal Norms And the Economics of Obesity

WSJ: Econoblog
August 25, 2006

During the past two decades, obesity rates in the U.S. have jumped dramatically. An estimated 30% of adults over the age of 20 -- more than 60 million people -- are obese, defined as those with a body-mass index of 30 or higher. (Read more on obesity definitions from the Centers for Disease Control.1)

Are economic reasons under-girding our ever-widening waistlines? The Online Journal asked academics Carol Graham, of the Brookings Institution, and Darius Lakdawalla, of the Rand Corp., to discuss their different economic takes on our burgeoning national girth.

* * *

Darius Lakdawalla writes: It's no secret that Americans have been getting fatter over the last several decades. But in fact, weight has been rising for more than 150 years, as shown by the economic historians Dora Costa and Richard Steckel3. From the Civil War to the 1990s, the weight of a 6-foot-tall American male increased by about 30 pounds on average.

These historical trends are not hard to understand. As we have gotten wealthier and more technologically advanced, food has gotten cheaper and work more sedentary. Both these factors have contributed to rising weight over the time-frame of centuries, and the recent rise in obesity has likewise been fueled by reductions in the price of food.

Since 1976, food has fallen in price by more than 12% compared to other goods. My colleague Tomas Philipson and I have shown that this reduction in price can explain at least half the recent growth in obesity4. Shin-Yi Chou, Michael Grossman and Henry Saffer reached similar conclusions about the importance of price. In addition to its overall price, they stressed the increasing availability of food service establishments5.

While it is not entirely clear whether restaurants make people heavier, or heavier people attract more restaurants, there is no question that eating is cheaper and easier than it used to be.

As if that were not enough, the most calorie-dense foods have seen the biggest price reductions. David Cutler, Edward Glaeser, and Jesse Shapiro have shown that technological advances have especially lowered the price of processed and snack foods6 -- like french fries and vending machine treats -- which are particularly high in calories.

The evidence above suggests that obesity is a by-product of prosperity and technological advance. In much the same way that traffic fatalities accompanied the boom in automobiles, roads and freeways, we now face the side effects of our efficient food production system and knowledge-based economy. I do not think it would make us better off to "roll back" advances in agriculture and food processing, or the shift away from physically demanding work. Instead, our goal ought to be to manage the consequences of obesity, by addressing its health impacts and making people accountable for bearing the costs and enjoying the benefits of their own weight choices.

* * *


Carol Graham writes: There is no doubt that cheap food and sedentary lifestyles play a major role in the increase of obesity. But an unanswered puzzle is the concentration of incidence in lower income groups. Obesity is largely a problem of the poor. If it's merely a story of cheap and readily available food, why then aren't the rich the fattest?

In recent research at Brookings7 with Andrew Felton, we use well-being (happiness) surveys to help explain the variance in obesity incidence across socioeconomic cohorts in the U.S., with a focus on the role of norms and expectations. Our hypothesis is that there is less incentive to prevent or reverse obesity if it is the norm to be in obese in one's racial or socioeconomic group and/or there are low expectations of rising into professional groups where obesity uncommon. It used to be the norm for the rich to be fat, to show off their prosperity8.

Nowadays, our president brags about how fit he is9. Isn't it interesting to see how norms change? And in developing countries, weight norms still suggest that obesity is a sign of prosperity. Our research finds that in Russia, obesity is concentrated among the rich, rather than the poor. In the U.S. we find that the poor suffer the highest well being costs (gauged by reported depression) from being obese, but these costs are concentrated among poor whites rather than among blacks or Hispanics. While obesity among the latter two groups is higher, they do not seem to suffer the same well-being costs as poor whites. We think that differences in norms -- and related stigma or lack of stigma -- may explain this. Respondents who depart from normal weight levels for their rank/profession also suffer higher well being costs than the average, and stigma is higher for those in higher status professions.

Finally, we find that obesity is negatively correlated with upward income mobility. We do not yet know, however, if this is due to lower expectations and effort by obese individuals or because of greater discrimination in the job market. Regardless, the results suggest that tailoring public health messages to marshal the attention of different cohorts in the U.S. -- and taking into account different norms and expectations -- might make them more effective.

* * *


Darius Lakdawalla writes: Without a doubt, obesity rates and body weights are higher among the poor than the rich, as they have been for decades. However, weight has been rising just as rapidly for the rich as for the poor. The "equal opportunity" spread of obesity points to causes that affect all groups in society, causes like cheaper food and more sedentary activities. Data from the CDC's National Health Interview Survey10 -- one of the primary sources of information about trends in American health -- illustrate how the growth in weight has cut across class lines in America. From 1976 to 2001, a 5-foot-10 high-school dropout would have gained 16 pounds, but a college graduate of that height actually gained slightly more. Moreover, during this period, the rate of obesity doubled among high school dropouts (growing from 12% to 25%), but it tripled among college graduates (rising from 5% to 17%). Today, more than one in six college graduates is obese.

This makes it very difficult to argue that obesity is a problem of just the poor, or of any one particular group in society, because obesity growth has cut across lines of age, sex, race, annual income, and just about any other characteristic that can be measured. As Tomas Philipson and I have argued11, rich people can afford more food, but they are also more inclined to demand better health. When it comes to body weight, these two effects of wealth oppose each other.

As such, it is not clear whether the rich will weigh more or less, and as Dr. Graham notes, the difference between rich and poor body weights varies from one country to another. The presence of different weight norms only adds to the complexity of this gap, which is certainly interesting and important. However, the uniform growth in weight points to changes that are unrelated to rich-poor differences, but instead related to broader and more sweeping societal transformations that affect all of us.

* * *


Carol Graham writes: I agree with Dr. Lakdawalla that there has been an increase in weight across groups over the past three decades, and I think that his argument does much to explain it. Still, there are stark and persistent differences in obesity across socioeconomic and racial groups. For all racial and ethnic groups combined, women of lower socioeconomic status are approximately 50% more likely to be obese than those with higher socioeconomic status. Men are about equally likely to be obese whether they are in a low or high socioeconomic group. And why are black men the thinnest group and black women12 the most obese? Income and manual labor can't explain that much of it. The fattest cities13, meanwhile, all have 2002 per-capita incomes of less than $30,00014, while the thinnest cities are all greater than $30,000. Plus, the thin cities are artsy, athletic cities -- Denver, Santa Fe, Burlington, etc. -- while the fattest cities are all down-and-out rust belt or poor southern cities. It's not very likely that Denver has more people working manual jobs than San Antonio does.

There must be something else going on. Dr. Lakdawalla notes that there are certain things driving the health behavior of the rich that mediate the effects of cheap food, sedentary life styles, and so on. That is precisely what our work tries to get at. What ARE those things? We focus on the role of norms in lifestyle and appearance, across socioeconomic and racial groups and across professions, as well as on the role of expectations. The idea here is that if you are in a socioeconomic group or profession where it normal to be obese or overweight, and you do not have high expectations of future mobility, then the incentives to make investments in healthy food and exercise that richer people seem to are much lower.

Changes in social norms about smoking15, for example, seem to have made a difference in reducing incidence smoking. Thus, if it's the norm in your social group to be obese, then that is one less reason to do something to either prevent or reverse a very difficult problem. Also, norms tend to be passed on from parents to children. On the other hand, if you have high expectations and want to move up the professional socioeconomic ladder (and the research of several scholars16 finds that discrimination against overweight and obese individuals is higher in higher status professions), then there is at least more incentive to make those kinds of investments. Our own research finds that a) the well-being costs -- or put another way, the stigma -- from being obese are much higher in high-status jobs than in low-status occupations and that b) obese individuals are less likely to have upward mobility than non-obese individuals. While we do not know if it is low expectations or discrimination in the labor force that drive the lower levels of mobility for obese individuals, both could well be at play.

There is a well-known study of the Pima Indians in Mexico and those in the southwestern U.S. (abstract17). They have the same genetic composition, but those in Mexico are engaged in farming and other agricultural activities, while those in the U.S. have one of the highest obesity incidences in the world. Surely sedentary lifestyles and cheap food play a role, but it is also quite likely that life on reservations, with low expectations for future mobility and high levels of depression/alcoholism also play a role.

While it is extremely difficult to precisely isolate the effects of norms and expectations versus those of cheap food and sedentary lifestyles, it seems very plausible that differences in the former set of factors play some role in explaining differences in incidence. Our research suggests that stigma against obesity is much lower in some racial, socioeconomic, and professional groups than others, and that accords with the higher obesity rates among those groups. It also suggests that obese people are less likely to experience mobility into higher status professions where obesity is rarer.

Merely accepting that there is some difference in norms and expectations has major implications for public health policy. Those with higher obesity norms and/or with lower expectations are much more likely to either ignore or be skeptical of public health messages about obesity.

While this makes the challenges for public health policy even greater, identifying the challenge is usually a first step toward designing appropriate policy.

* * *


Darius Lakdawalla writes:

To clarify our discussion, it is helpful to distinguish between two questions:

(1) Why has obesity grown so much over the past 30 years?

(2) Why do body weight and rates of obesity vary across the population at a single point in time?

Dr. Graham and I agree that society-wide changes in the price of food and the strenuousness of activity are necessary to understand the first question, of why obesity has grown so much. However, we differ in how best to explain variation in weight across the population. I find incentives to be a more compelling explanation, while she favors a norms-based explanation.

Dr. Graham points out that rich cities are thinner than poorer cities, and asks why the rich choose healthier weights. In its variation between rich and poor, weight is no different than other health conditions, where social norms are less likely to be relevant. Richer people live longer, have fewer diseases of all kinds18, and spend less on curative medical care. Surely we would not argue that it is more acceptable among the poor to die earlier, develop brain tumors, or be stricken by heart attacks.

Moreover, health behaviors where norms might be potentially important, such as obesity and smoking (which, as an aside, lowers obesity, but not enough to explain much of its remarkable growth19), explain very little of the vast difference in health between rich and poor. Since the rich are healthier in ways that norms cannot explain, one has to ask why they should be deployed in explaining the specific case of weight, but not the many other illnesses that exhibit the same rich-poor variation without social norm differences.

Moreover, incentives explain the variation between rich and poor at least as well as social norms. The seminal work of Michael Grossman, in 1972, argued that richer and more educated people have higher demands for health, because they stand to lose more20in the way of lifetime income if they die young. Dr. Graham's examples make this point as well. For instance, she cites recent work by Jay Bhattacharya and Kate Bundorf that shows "discrimination against overweight and obese individuals is higher in higher status professions." In fact, Bhattacharya and Bundorf argue that people with health insurance face a larger wage penalty if they are obese, because their employers end up paying for their higher medical costs. They rely on incentives, not social norms, in explaining the phenomena they observe. Incentives also have several important practical advantages over social norms. Norms-based theories always involve a "chicken-and-egg" type problem. Are people fatter because it is more acceptable to be fat, or did it become more acceptable when more people got fat? Finally, and perhaps most importantly, it is often difficult to act on a theory of social norms, because policymakers can change incentives much more easily than social norms.

* * *


Carol Graham writes: Dr. Lakdawalla and I actually agree on a role for incentives. That is precisely why Andrew Felton and I focused on mobility and discrimination in the workplace in our work. Our focus on expectations for future mobility is linked to the whole issue of incentives -- for investing in education and work, and in healthy behaviors. We think that norms and expectations interact with each other. Just because this is difficult to measure and there is a chicken or the egg problem in looking at the role of norms does not mean that their role should be dismissed or that they are irrelevant to policy.

I agree with Dr. Ladkawalla that the evidence is clear that rich people are healthier, and no doubt, demand better health care. Richer parents are also more likely to invest in their children's education. And there are strong links between education levels and health outcomes. One plausible explanation is that those with lower expectations for their own and their children's future mobility have higher discount rates.

In other words they are less willing to delay present consumption or income for future benefit. If the future is uncertain or bleak, why bother? In addition to our research, the discount rates hypothesis is explored in this paper (abstract21).

Our argument is that the whole issue of incentives is moderated by norms. Investing in healthy eating habits and exercise takes time and resources. If the economic incentives to do so are high, because there is professional discrimination against obesity and high economic rewards to attaining those professional opportunities, then there are clear incentives to make those investments. If, in contrast, there are not many opportunities for upward advancement and the norm in the lower-paying professions is overweight or obese, then the incentives to make those investments are very low. Professional norms are interacting with incentives here.

Some other evidence which provides indirect support for the role of norms is the effects of junk food on childhood obesity. A recent paper by Patricia Anderson and Kristin Butcher22finds that a 10% increase in the availability of junk food in school cafeterias leads to a 1% increase in students' body mass index. But that effect is driven entirely by students who have an overweight parent. It is likely that genetic susceptibility to obesity is interacting with family-based norms -- about how much of the available junk food it is appropriate to consume -- in driving higher obesity rates among certain adolescents and not others.

Norms are more difficult to change than incentives. And, as Mary Burke of the Boston Fed23writes, changing norms takes time: "The lagged adjustment of weight norms helps to explain recent observations that obesity rates have continued to rise since the mid 1990s, despite an apparent leveling off of price declines." Still, norms are changeable via both public health education and peer pressure, as the smoking story indicates. And eventually, there can be cascade effects, as noted in this paper24by Cass Sunstein: "Should Government Change Social Norms?"

Finally, this week's New York Times magazine features25a woman working to reduce obesity in schoolchildren, and makes them recite a pledge: "I do solemnly swear… to be healthier, to eat low fat, to eat high fiber, to drink lots of water and get lots of exercise." She is actively trying to change social norms, although it's not easy. Some children are responding; however, after the schoolwide assembly to introduce the full program last September, one frantic parent called to report that her child was refusing to eat anything in the house that was not healthy. "I can't afford to throw everything away," the mother said. "Please tell her to eat."

* * *


Darius Lakdawalla writes: We are left with one remaining question: What role can social norms play in explaining why some groups are more obese than others? The show-stopper for social norms is the impossibility of validating a theory that includes norms at all.

Norms may change behavior, or behavioral changes may produce new norms. To be sure, this chicken-and-egg problem occurs throughout economics: Do more restaurants make people fatter, or do fatter people attract more restaurants? Physical scientists solve these problems with randomized trials: Begin with two identical groups, and treat one group differently; the resulting difference between groups must then reflect the difference in treatment and nothing else. The analogous approach in economics identifies policy changes that treat one "group" but not an otherwise similar group. For example, what happens to weight in two similar U.S. states -- one that reduces its sales taxes on food, and one that does not? This general approach, widely used by economists to study important policy questions, is well-described for a general audience by Levitt and Dubner, in their popular book Freakonomics26.

This is the best -- and perhaps only known -- way to sort out the chicken-and-egg problem, but it is impossible to use for social norms. We would need to find two very similar societies, and isolate a sudden change in the attitudes of one, but not the other. The very fact that norms change slowly makes them impossible to validate scientifically, as well as extremely unwieldy for policymakers. We can certainly theorize about social norms, but it seems quite difficult, if not impossible, to demonstrate their relevance in the real-world.

* * *


Carol Graham writes: I accept that it is very difficult to isolate the role of norms, particularly as their effects seem to interact with the role of other factors, such as expectations. That said, we know that norms can tip quite suddenly and induce very radical changes in social behavior. Recent Nobel Prize winner Tom Schelling's work on segregation is one example. He documented, in a very simple and straightforward way, how modest changes in seating assignments in army and other cafeterias could result in outcomes ranging from complete segregation to complete integration. My Brookings colleague Peyton Young's work on norms27-- based on data for agricultural planting practices -- shows how modest changes in norms can then have cascading effects. Josh Epstein and Rob Axtell, also Brookings colleagues, have used agent-based computational models to demonstrate long lags and then tipping points in the response to policy changes28, such as changing the retirement age. The U.S. retirement age was lowered to 62 almost a decade before a significant number of people began to retire at that age.

Dr. Lakdawalla references the very exciting new work in economics that mimics scientific research via the use of randomized trials. Why is it impossible to envision a randomized trial experiment where one takes a number of school cafeterias that are similar in their socio-demographic make-up and introduces a policy intervention designed to change norms -- such as those described in the New York Times article referenced above -- in some cafeterias and not in others? At the least it could help answer the question of whether norms can be changed via such interventions.

Understanding the effects of norms on social behavior is far from an established or exact science. But dismissing the role of norms -- about professional appearance and aspirations, about investments in education and in healthy lifestyles, and about eating habits, and how these norms vary across various socioeconomic and professional groups -- risks missing an important part of the story of how and why obesity in the U.S. keeps going up in the midst of an increased barrage of public health warnings about its dangers.

* * *


Hyperlinks in this Article:
(1) http://www.cdc.gov/nccdphp/dnpa/obesity/defining.htm
(2) http://discussions.wsj.com/n/mb/message.asp?webtag=wsjvoices&nav=messages&msg=3983
(3) http://web.econ.ohio-state.edu/rsteckel/VITA/1997 Health and Welfare during Industrialization---1997.pdf
(4) http://harrisschool.uchicago.edu/About/publications/working-papers/pdf/wp_02_03.pdf
(5) https://wfs.gc.cuny.edu/PAnderson1/www/Papers/jhe obesity paper.pdf?uniq=-vxujr2
(6) http://www.economics.harvard.edu/hier/2003papers/HIER1994.pdf
(7) www.brookings.edu/views/papers/graham/2005graham_dict.pdf
(8) http://seattlepi.nwsource.com/national/190064_poorthin09.html
(9) http://www.usaweekend.com/00_issues/000730/000730interview_bush1.html
(10) http://www.cdc.gov/nchs/about/major/nhis/hisdesc.htm
(11) http://harrisschool.uchicago.edu/About/publications/working-papers/pdf/wp_02_03.pdf
(12) http://www.cdc.gov/nchs/data/hus/tables/2003/03hus068.pdf
(13) http://www.obesity.org/subs/fastfacts/cities.shtml
(14) http://www.bea.gov/bea/newsrel/MPINewsRelease.htm
(15) http://www.frisch.uio.no/sammendrag/265_eng.html
(16) http://iis-db.stanford.edu/pubs/20879/healthcare_costs_of_obesity.pdf
(17) http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=10702751&dopt=Abstract
(18) http://www.pophealth.wisc.edu/phs548/smith wallets.pdf#search=%22smith healthy bodies and thick wallets journal of economic perspectives%22
(19) http://www.nber.org/papers/W11483
(20) https://wfs.gc.cuny.edu/PAnderson1/www/Papers/conceptofhealthcap.pdf?uniq=-vxubit
(21) http://papers.nber.org/papers/w12352
(22) http://www.dartmouth.edu/~pmaweb/RaisnetsOct2005.pdf
(23) http://www.bos.frb.org/economic/ppdp/2006/ppdp065.htm
(24) http://www.aei.org/publications/pubID.18910,filter.all/pub_detail.asp
(25) http://www.nytimes.com/2006/08/20/magazine/20lunches.html
(26) http://www.freakonomics.com
(27) www.brookings.edu/es/dynamics/papers/custom/default.htm
(28) www.brookings.edu/es/dynamics/papers/csed_wp01.htm
(29) http://discussions.wsj.com/n/mb/message.asp?webtag=wsjvoices&nav=messages&msg=3983


Darius Lakdawalla is an economist at the RAND Corp., in Santa Monica, Calif. His research focuses on the economics health risks and the causes of contemporary trends in health and disease. Some of his previous work has explored reasons behind declining use of nursing homes by the elderly, acceleration in the rate of new HIV infections, and the growth in disabilities among younger Americans. Dr. Lakdawalla received his Ph.D. in economics from the University of Chicago. He is currently a Faculty Research Fellow at the National Bureau of Economic Research, in Cambridge, Mass., and an associate professor of economics at the Pardee RAND Graduate School of Public Policy.

Carol Graham is professor of public policy at the University of Maryland and a Senior Fellow in the Economic Studies Program at the Brookings Institution, where she also co-directs the Center on Social and Economic Dynamics. Her research focuses on poverty, inequality, and measurements of well being in both developed and developing economies. Some her previous studies have explored the role of low expectations and high inequality in explaining poverty traps and sought to explain high levels of public frustration with globalization. She is the author of "Happiness and Hardship: Opportunity and Insecurity in New Market Economies" with Stefano Pettinato among numerous other books and articles. She received her doctorate in political economy from Oxford University.