A goal of food nutrition labeling policies is to help consumers make informed choices regarding the healthiness of the food they purchase. Yet, findings regarding the effectiveness of label use to decrease negative health outcomes, such as overweight and obesity, are mixed. As most studies focus on estimating mean effects of labels, little is known on whether labels have any effect at the tails of the body weight distribution, which, given the social gradient of obesity, often includes at-risk groups. Additionally, label use, overweight and obesity vary across population subgroups. Thus, the relationship between using food nutrition labels and body weight may be characterized by marked heterogeneity, which the current literature has failed to address. This study explores the non-linearity of the relationship between reading ingredients on the food label and Body Mass Index (BMI) using an unconditional quantile regression estimator and one year of data on adult Italians from the Multipurpose Household Survey. We study this relationship across the BMI distribution and for different groups of respondents, divided by gender, income above and below the sample average, education level, perceived hardships to access food, and regular practice of sport. The results indicate that reading ingredient labels has a negative association with BMI, mostly at higher BMI quartiles (overweight and obese), although a relationship at the highest quartile is only found in a few subsamples. Females, and individuals with a higher risk of being overweight and obese such as low-income, low educated, or those who do not practice sport seem to garner the highest benefit from reading ingredients in the food label. The paper concludes with policy implications, tapping into the recent debate regarding the revision of food labeling in the European Union.
- Food labeling
- Unconditional quantile regression