Consumer beliefs about healthy foods and diets

That’s the title of a new article I just published in the journal PLoS ONE. This is an exploratory/descriptive study with the aim of probing consumer’s perceptions of the term “healthy” in relation to food. The study is motivated by the fact that the FDA regulates the use of the term on food packages, and is in the process of reconsidering the definition. Here are some of the key results:

Consumers were about evenly split on whether a food can be deemed healthy based solely on the foods’ nutritional content (52.1% believing as such) or whether there were other factors that affect whether a food is healthy (47.9% believing as such). Consumers were also about evenly split on whether an individual food can be considered healthy (believed by 47.9%) or whether this healthiness is instead a characteristic of one’s overall diet (believed by 52.1%). Ratings of individual food products revealed that “healthy” perceptions are comprised of at least three underlying latent dimensions related to animal origin, preservation, and freshness/processing. Focusing on individual macronutrients, perceived healthiness was generally decreasing in a food’s fat, sodium, and carbohydrate content and increasing in protein content. About 40% of consumers thought a healthy label implied they should increase consumption of the type of food bearing the label and about 15% thought the label meant they could eat all they wanted.

One part of the analysis focuses on parsing out the correlations between the healthiness rating consumers placed on different types of foods . Below are three dimensions of 15 food’s healthiness ratings as determined by factor analysis.


Here’s the portion of the text describing these results:

The first factor (explaining 54% of the total variance), shown on the vertical axis of the bottom panel of Fig 3 shows all animal products with high values and other non-animal products with lower values, suggesting consumers use animal origin as a primary factor in judging whether a food is healthy. A second factor (explaining 31% of the total variance), illustrated on the horizontal axis of the top panel of Fig 3, has canned and frozen fruits and vegetables with the highest values, bakery and cereal items, candy, and fresh fruits and vegetables with mid-to-low values, and animal products with the lowest values, which seems to suggest consumers use degree of preservation as another dimension of healthiness. Finally, the third factor (explaining 22% of total variance), illustrated on the vertical axis of the top panel and the horizontal axis of the bottom panel of Fig 3, indicates freshness or degree of processing is another dimension to healthiness evaluations. These results indicate that healthiness is not a single unifying construct, but rather consumers evaluate healthiness along a number of different dimensions or factors. A food, such as beef or fish, can be seen as scoring high in some dimensions of healthy but low in another.

There’s a lot more in the article.

What Food Policies do Consumers Like and Dislike?

I have a new working paper with Vincenzina Caputo in which we elicit consumers’ preferences for 13 different food policies. Here’s our main motivation (references removed for readability).

A variety of food policies have been proposed, and in some cases enacted, in an effort to improve public health, environmental outcomes, or food security. Proposed actions include a spectrum of policies ranging from fiscal incentives/disincentives, bans, labelling programs, and passive policies such as subsides and investments in education. What food policy proposals do consumers prefer? While there have been numerous studies aimed at calculating the welfare effects of individual food policies it is difficult to easily ascertain the relative preferability of numerous policy options, even those that have the same objective (e.g., “fat taxes” and nutritional education both aim to improve public health).

We conducted a nationwide survey of 1,056 U.S. consumers who were asked to indicate the relative desirability of the following food policies.


Rather than use a traditional approach, where respondents are not required to make trade-offs between policies (e.g., people can approve of all policies or rank all policies as “very important”), we used the “best worst scaling” approach that requires respondents to make trade-offs. The approach requires respondents to answer a series of questions like the one below, where for each question, they have to indicate their most and least preferred policies.


The results are analyzed using a choice model that allows for preference heterogeneity. The main outcomes are below, reported as “preference shares” - i.e., the percent of people predicted to choose each policy as most preferable. Results indicate the highest levels of support for investments in agricultural research and requirements of food and agricultural literacy standards in public education. Fat, calorie, and soda taxes are the least popular. These preference shares provide a measure of intensity of preference in a population. Funding for agricultural research is 14%/8% = 1.75 times more preferable than symbolic nutritional labeling.


While the above results are useful in providing intensity of relative preferences, they do not indicate whether people would actually vote in favor of a policy. The table below shows the results of that question; the results largely align with the best-worse scaling approach. Fewer than one-third of respondents are in favor of these three tax policies.


There are a number of significant demographic correlates with policy preferences. Some are not surprising. For example, Nutrition Assistance (or SNAP) is more desirable to lower income vs. higher income households and Democrats vs. Republicans. As another example, soda taxes are less desirable among lower income households.

Funding for agricultural research was generally supported across all demographic categories except for age: older individuals were more supportive of funding for agricultural research than younger individuals.

My State is Better than Yours: Competition between State Food Branding Programs

The journal Agribusiness just released a new paper I co-authored with Clint Neill and Rodney Holcomb. The work was motivated by the observation that every state in the U.S. has an agricultural marketing program aimed at promoting foods from their state. Examples include the “Taste NY” and “Pride of New York” programs as well as “Go Texan” and “California Grown.”

Our questions were two fold: 1) How much do consumers value products labeled with their state’s logo relative to other states’ products, and 2) what are the implications for state marketing programs?

We surveyed 6,900 consumers in an eight‐state contiguous region. For our application, we chose milk, and asked people which of several milk products with different state logos (and a regional or national brand) they preferred at different prices.

Not surprisingly, we find that consumers prefer products with their own state’s logo. For example, Texans’ average willingness-to-pay (WTP) for Texas milk is $4.14/gallon, but Texans’ value for milk from bordering states, New Mexico, Oklahoma, and Arkansas only averaged $1.82, $2.65, and $2.72/gallon, respectively. There are a number of interesting patterns. Here’s an excerpt from the text:

While each state’s consumers tend to prefer their own label and have a distinct order of preference for other states, the asymmetry between states is less clear. For example, Oklahoma consumers are willing to pay $2.84 for the Texas label but Texas consumers are only willing to pay $2.65 for the Oklahoma label, so there is an asymmetry of $2.84−$2.65 = $0.19. Thus, Oklahomans value the Texas label $0.19 more than Texans value the Oklahoma label.

Table 5 shows this type of asymmetry for all combinations. Interestingly, every other state’s consumers value the Colorado label more than Colorado consumers value other states’ labels. Alternatively, New Mexico consumers value all other state brands more than the other states’ consumers value the New Mexico label.

While it is perhaps obvious that people in a state will tend to prefer their own products, it is also important to note that people have some value from agricultural products from other states (and, in fact, some small share of people prefer products from another state). The result is that state branding programs “steal” consumers from other states (the effect is a bit like the prisoner’s dilemma problem). The state branding program looks great if your the only state that has the program, but if all states have their own programs, the effects partially serve to cancel each other out. Here’s what we write about this so-called “beggar thy neighbor” effects:

In the case of market shares, we were able to illustrate the large decreases as a group of producers from one state starts with having no state branded competitors to competing against several other brands within a region. Producers, ideally, would have a higher return if they were the only ones with a state label, but the optimal strategy for all agents in the region is to utilize a state label. Thus, the potential beggar‐thy‐neighbor scenario is possibly a Nash equilibrium. Furthermore, states who market their brand outside their borders are shown to have increased total market share

For example, below is a graph showing what happens to demand for milk with a “Made in Oklahoma” label when no other states label their product (the green line with triangle markers) relative to what happens to demand for “Make in Oklahoma” milk when other states introduce their own labels (the red and blue lines). As the figure below shows, the market share more than halves when one state’s label has to compete with all the others in a region.


One potential solution (at least from the producer’s perspective) we discuss is for groups of states to band together and use a regional label.

Income and (Ir)rational food choice

That’s the title of a new paper I have forthcoming in the Journal of Economic Behavior and Organization.

In short, I find the more one spends on food, the less consistent are their choices. In the economic way of thinking, inconsistency is typically associated with irrationality. First saying I prefer A to B, but then later saying B is preferred to A is an inconsistency, which is often referred to as a preference reversal. It’s hard to square such preference reversals with any model of rational choice.

Why might preference reversals increase with a consumer’s income? Here’s a bit from the paper (omitting references):

This paper sought to determine the relationship between consumers’ incomes and food expenditures on the one hand and preference consistency on the other. Previous literature has suggested at least two channels through which increasing income or expenditure might have deleterious effects on preference stability. The first operates through increasing demand for novelty and variety as incomes rise and the second operates via the relative incentive to behave rationally as the stakes fall.

In an empirical application involving almost 540,000 food choices made by almost 60,000 people, I find that 47% of respondents committed at least one preference reversal. How do preference inconsistencies relate to income and food spending?

Results show that the likelihood of a reversal [or preference inconsistency] and the number of reversals are significantly increasing in expenditures on food at home and away from home, and to a somewhat lesser extent, total household income. The magnitudes of these effects are large; larger than that associated with any other demographic or study design variables explored. For example, that the odds of committing a preference reversal [or preference inconsistency] are about 1.8 times (2.5 times) higher for individuals who spend $160/week or more on food at home (away from home) compared to individuals who spend less than $20/week. Exploring responses to three different “trap questions” that measure respondent attentiveness indicates that results cannot be explained by higher income households generally being more careless in their responses to questionnaires.

To explore the extent to which income and preference stability is related to variety or novelty seeking, the relationship between preference reversals and food values is also explored. As hypothesized, of the 12 food values studied, the relationships with preference reversals are strongest for the food values of price and novelty. Consumers for whom food price is a more important food value tend to commit fewer preference reversals. By contrast, consumers who rate novelty as a more important food value are more likely to exhibit unstable preferences.

Why does it matter whether rationality falls as incomes and food spending rises? As I’ve argued previously, increasing affluence likely allows us to indulge “higher” needs related to self actualization and self expression. Here’s a last bit from the paper, which is more speculative, and hopefully will spur some additional research (again, omitting references for readability).

There is a view among many food and agricultural scientists that many new food products marketed to higher income consumers are “unscientific” insofar as they make absence claims about ingredients and processes scientists have deemed safe. The preference instability observed among consumers with greater food expenditures in this study need not necessarily relate to beliefs about food that diverge from scientific consensus. Nonetheless, rising incomes might better enable people to seek out and identify sources of information that conform to their beliefs and cultural identities. It has also been argued that consumers might directly obtain utility from holding certain beliefs, which might lead to information avoidance. Whether certain food and agricultural beliefs are normal or inferior goods, in this framework, is an open question.

Can You Call it Meat?

NPR recently ran a story, in which I was quoted, about the rise of state laws limiting the use of words like “beef”, “meat” and even “rice” on plant-base alternatives. The American Civil Liberties Union (ACLU) has just filed suit against the state of Arkansas over the state’s enactment of a law that would fine “plant-based and cell-based meat product, such as “veggie burgers” and “tofu dogs,” marketed or packaged with a “meat” label.”

What to make of all this? One one hand, these sorts of new laws originate from some of the same motivation of older “standards of identity” laws. These laws define how certain words can be used on food labels and in marketing. The stated purpose of the laws are to protect consumers and to prevent consumers from being misled. For example, in the past, some unscrupulous millers added wood shavings to flour. If consumers can’t tell before buying whether it’s the real or adulterated version, we can wind up a markets-for-lemons problem, which would drive the high quality products out of the market and leave consumers worse off.

However, here’s what I wrote about this a while back (I also included a few illustrative pictures of labels):

In the case of beef, I am a bit skeptical that consumers will be mislead by the start-up meat alternatives. Why? These aren’t generic products being sold by companies trying to water down or adulterate a product with cheaper inputs. These are branded products created by firms whose whole marketing strategy is to tell people their product is NOT beef. ... Even without the identity standards, it is not as if consumers are totally unprotected. If they are, in fact, misled, the legal system offers possible remedy. As witnessed by the numerous lawsuits over the use of the word “natural,” I suspect there are plenty of lawyers out there willing to help a consumer who can show they’ve experienced damages.

The counter response is that people might associate words like “beef”, "meat”, or “milk” with other product attributes such as nutritional content, which might (sometimes inappropriately) carry over to the plant- or lab-based products. Nutritional facts panels may serve to mitigate some of these concerns, but there is little doubt that labels create various taste and health halos that extend beyond the objective facts.

At the same time, words are needed to convey meaning to consumers beyond just animal content. Using the word ground “meat” tells me something about how the food is expected to be cooked and served and which condiments are appropriate. In this instance, using “meat” with “plant-based” is helpful to the consumer insofar as quickly conveying key information about how the product is to be cooked and consumed.

Thus, there are pros and cons and costs and benefits to these types of labeling laws. I’ve seen a few polls on what consumers think about these labeling laws. However, It would be useful to see more empirical research over whether consumers are, in fact, mislead or perhaps more informed by meat/milk labels on plant-based products.