News & Publications



Breaking Habits: The Effect of the French Vending Machine Ban on School Snacking and Sugar Intakes | JPAM Featured Article

January 12, 2018 10:02 AM

By Sara Capacci, University of Bologna,  Mario Mazzocchi , University of Bologna, and Bhavani Shankar, SOAS (South Asia Institute) in London

Article Introduction

There is consistent evidence showing that food and drinks sold from vending machines (VMs) have on average a higher energy-density, and access to VMs in schools has been associated with unhealthy diets and excess weight in children, especially in middle schools. However, there are few studies assessing the impact of regulations banning VMs from schools. From September 2005, the French government has banned VMs from all middle and secondary schools. This paper evaluates the impact of the ban on the nutrient intakes and school snack frequency of schoolchildren, and considers intakes from out-of-school meals to check for compensation effects.

The identification strategy is based on the discontinuity observed at the age of moving from primary to middle school and on the availability of data from two national nutrition surveys (INCA1 and INCA2) before and after the ban.  Even before the ban, VMs were not allowed in primary schools, but they were available in middle schools. Hence, through a combination of difference-in-difference and a regression discontinuity design, the authors estimate the policy impact as the change in the discontinuity after the ban.  Their estimates, corroborated by a set of robustness and validation checks, indicate that the ban has induced a significant but small reduction (10 grams, or a third of an ounce) in the sugar intake of those children taking a morning break at school. The frequency of these morning snacks has also decreased.  However, the application of the model to consider out-of-school meal occasions is suggestive of compensation effects, and the ban had no impact on the overall diet quality of French schoolchildren. VM bans and other school restrictions should be part of a broader policy package to address compensating behaviors outside the regulated environment.

This article preview is from the Winter 2018 issue of the Journal of Policy Analysis and Management (JPAM). APPAM invites authors from each issue to asnwer a few questions about their research to further promote the quality work in the highly-ranked research journal. Check out this and other JPAM articles online.

Author Interview

What spurred your interest in this research on vending machine (VM) bans in schools, or in school meal programs more broadly?

We were involved in an European research project named EATWELL, aimed at mapping nutrition programs in Europe and reviewing existing evidence on their effectiveness. That’s how we came across the French vending machine ban. Rigorous ex-post evaluations of these programs were almost absent, or hindered by improper assumptions. The French ban proved to be an ideal case study for being properly evaluated using quasi-experimental methods, given the availability of national nutrition data at the individual level, together with information on individual meals.

Considering the overall limited impact of the VM ban as a standalone policy, what are your thoughts regarding the health impact verses the financial benefit of VMs on school property? 

The school environment is a particular one, regardless of the direct outcomes of a policy like the vending machine ban. It makes little sense to promote better school meals, include elements of nutrition education in curricula, but make less healthy foods easily available. The fact is that these policies should never be “standalone”, but part of an integrated policy plan.

Obviously, policy-makers cannot ignore how schools benefit from VM contracts. I think this explains why this policy has been implemented in a European country, where most of funding for education is public and the health care costs associated with unhealthy nutrition are also public. The US system is different, and rests more on individual responsibility and free choice, but we cannot assume that children and teenagers are fully rational in their lifestyle choices, they highly discount future health consequences.

We should also consider that it’s not the VM as a “retail mean” which is bad, and there should be room for contracts which impose healthy nutrition standards to contractors. These contracts are likely to be less profitable than the current one, but policy-makers should be ready to replace, at least partially, private funding with public funding. It’s an investment for the future, leading to savings in public and private healthcare costs.

Sugar consumption dropped for the morning snack itself, and your study also found a significant reduction of sugar intake when considering the afternoon snack only. However, potential compensation effects undermine that reduction in sugar intake throughout the day. Does this suggest that school food policies could be more impactful if focused more on the quality of the main meal(s) school children consume on the premises? Or do you attribute the possible compensation effect wholly to food consumed off-property?

Our evidence was mainly on compensation outside the school environment, and our data does not us allow to determine whether lunches are consumed at school or not with certainty. France is already implementing a school meal policy, although compliance wasn’t perfect at the time of our study. There is certainly room for improvement for school meals, but the key is to trigger behavioral change that continue outside school, by changing social norms.

According to your paper and sited research, VM use increases with school age – 41 percent for high school students, 28.5 percent for middle school students, 16 percent for primary school students. How can this study inform further research on the impact of the VM ban on older students? 

We think that an interesting element emerging from our research and other studies is that regulating the school environment does work within the school premises, but out-of-school children with their own pocket money are more likely to compensate. As age increases, children become more and more independent in their lifestyle choice, and research should address that, and explore the budget allocation of teenagers. Since budget constraints are likely more relevant, it is unsurprising that they go for cheaper calories. To our knowledge there are little studies and data about price elasticity and substitutions of older students. A better knowledge on these aspects might open the way to more effective strategies even in situations where a VM ban is less feasible, e.g. in non-school sport premises, movie theatres, etc. At any rate, one of the key conclusion of our study is that any policy should be part of a wider strategy including an education element. Our findings on compensation effects suggest that future research should also target different environments and lifestyle choices.

What challenges, if any, did you find when conduction this research? How can further study overcome these challenges? 

The main challenge regards the data. We used secondary data from two waves of a national nutrition survey which was not collected purposely for evaluating the vending machine ban. Therefore, we had to circumvent several shortcomings in data. For example, the lack of information on respondents’ date of birth has introduced a considerable amount of uncertainty to the relation between our age measure and the school level, which led us to choose an evaluation technique that accounts for this “fuzziness”. We challenged the assumptions of our modelling approach, and we were pleased to find robust results and consistent evidence.

Moreover, a huge work of data cleaning has been necessary. We had to discard holiday periods, which in France differ by region and harmonize potential differences in the school week.

So, the main challenge was to evaluate a very specific policy using secondary data. The availability of primary data collected for evaluation purposes would have made our evaluation easier and stronger.  

The use of appropriate methods might mitigate data issues, but our hope is that future studies might exploit better data.  

How does this study impact or add to the existing research that can inform policy in U.S. schools? What would be the ideal next step for your research findings? How would you like to see your findings implemented? 

To our knowledge, VM bans implemented in the US are partial (e.g. soft drinks) and mostly limited to school districts. Thus, evidence on a nationwide ban like the French one can be precious in terms of its generality, although the difference in culture and food habits definitely plays a role. We believe that our research has emphasized the need for comprehensive, consistent and rigorous school policies when the goal is to improve schoolchildren nutrition. These should be accompanied by adequate communication and education, and ideally one would want to see an intervention study to test how a multi-level policy might impact not only on actual behavior, but also on attitudes and knowledge. Behavioral change cannot be simply triggered by tough regulations, but should rest on awareness. In other areas (e.g. school fruit schemes) there is even evidence that school interventions might have positive spillover on parents and relatives, thus with economies of scale, but – again – this requires active participation and motivation by children. Our hope is that these multi-level interventions, addressing school food supply while providing adequate communication and education, could be evaluated by collecting good quality data on out-of-school behaviors.

About the Authors

Sara_Capacci Sara Capacci is Senior Research Assistant in Economic Statistics at the University of Bologna. Her research focuses on the evaluation of public policies, mostly in the health and nutrition field. She is particularly interested in quasi-experimental methods to assess the effect of interventions in observational settings. She participated in many European research projects focusing on the evaluation of public policies aimed at improving nutrition quality. Further research interests include demand modelling and consumers’ acceptance of public interventions.
Mario_Mazzocchi_picture Mario Mazzocchi is an applied statistician with a research focus on the evaluation of economic policies, especially food policies, and the modelling of consumer demand. He is currently Professor in Economic Statistics at the Department of Statistical Sciences of the University of Bologna. He is also a consultant to FAO on nutrition policies, and has been a member of the European Commission group of experts on the evaluation of the EU School Fruit Scheme. He is co-editor-in-chief of the international journal Food Policy. He has led research teams of the University of Bologna in various international projects. Among these EATWELL (Interventions to Promote Healthy Eating Habits: Evaluation and Recommendations) and DEDIPAC (Determinants of Diet and Physical Activity), and the forthcoming project PEN (Effectiveness of existing policies for lifestyle interventions - Policy Evaluation Network). His publication record includes two books, with Oxford University Press (Fat Economics) and Sage Publications (Statistics for Marketing and Consumer Research).

Bhavani Shankar is an applied economist working on research areas at the intersection of agriculture, food, nutrition and health, in developing as well as developed countries. He is currently Professor of International Food, Agriculture and Health at the Centre for Development, Environment and Policy at SOAS (South Asia Institute) in London. His current research interests include the analysis of economic drivers of over and under nutrition, food and nutrition policy evaluation and the role of agriculture and food systems in enabling better nutrition and health. His ongoing research projects include LANSA (Leveraging Agriculture for Nutrition in South Asia), POSHE (Palm Oil: Sustainability, Health and Economics) and SHEFS (Sustainable and Healthy Food Systems). He is currently an Academic Editor for PLOS One, and served as Managing Editor of Food Policy during 2011-2015.

Submit Your Research to JPAM

JPAM seeks contributions that span a broad range of policy analysis and management topics with an emphasis on research that conveys methodologically sophisticated findings to policy analysts and other experts in the field. Both domestic and international contributions in public management are welcome, as well as a broad range of policies related to social well-being, health, education, science, environment, and public finance. JPAM strives for quality, relevance, and originality. An interdisciplinary perspective is welcome as are articles that employ the tools of a single discipline.

Find information for prospective authors and submit manuscripts via Editorial Express.

Back to news