Developing Robust Methods for Evaluating Policies in the Markets for Food and Nutrition – NUTRITION

The proposed research will be organised around three inter-related projects: Project 1 will take as its starting point random coefficients models of demand and extend these by relaxing the restrictive distributional assumptions that are commonly applied; we will use household-level panel data on food purchases and develop models that allow for rich and flexible heterogeneity in household preferences; these models will also allow firms to strategically responses to potential policy reforms. Project 2 will test the validity of existing methods that are used for identifying individual consumption from household level expenditure data; and allow for more heterogeneity along observable dimensions. Project 3 will apply these structural models to the analysis of existing and potential policy interventions in the area of food and nutrition. These policies will include information campaigns, taxation on nutrients or foods categories, and other interventions in the food retailing industry.

Our work has allowed to start numerous research articles that we cannot summarize here and I'll summarize two of the most important ones below::

- Dubois, Pierre; Griffith, Rachel and Nevo, Aviv: “Do Prices and Attributes Explain International Differences in Food Purchases” à paraître à l’American Economic Review
Food purchases differ substantially across countries. We use detailed household level data from the US, France and the UK to (i) document these differences; (ii) estimate a demand system for food and nutrients, and (iii) simulate counterfactual choices if households faced prices and nutritional characteristics from other countries. We find that differences in prices and characteristics are important and can explain some difference (e.g., US-France difference in caloric intake), but generally cannot explain many of the compositional patterns by themselves. Instead, it seems an interaction between the economic environment and differences in preferences is needed to explain cross country differences.

- Dubois, Pierre; Griffith, Rachel, Arthur Lewbel, Martin O’Connell : “Evaluating the impacts of a government information campaign on food demand: a discrete-continuous choice model of demand”
Many countries have implemented “5-a-day« campaigns that encourage consumers to eat at least five portions of fruit and vegetables every day. We consider how such policies affect outcomes, considering both supply and demand effects. We develop and estimate a model of grocery demand that embeds individual information into overheads and Barten scales. This allows information to shift the demand curve that retail firms face in quite flexible ways. We estimate how demand curves did shift, exploiting variation in individual households' exposure to the campaign. We estimate what firms' optimal pricing responses to this shift in demand would be, under the assumption of Bertrand-Nash equilibrium.

We have different perspectives related to the on-going work.
One is to develop a model which is ongoing that allows to evaluate the effectiveness of regulating food advertising as a means of reducing consumption of food that are of low nutritional quality, or junk foods. France, the US and UK currently restrict adverting of junk foods. The impact of such a policy will depend on how brand advertising influences consumer demand, and on the strategic pricing response of oligopolistic firms. In this ongoing research, we use transaction level purchase data and detailed advertising data on the UK crisps market to estimate a structural model with dynamic demand and supply and evaluate the welfare effects of a ban. We simulate the impact of a ban on advertising, allowing for strategic pricing response of firms operating in the market.

1Dubois, Pierre; Griffith, Rachel and Nevo, Aviv: “Do Prices and Attributes Explain International Differences in Food Purchases” à paraître à l’American Economic Review
Conferences and seminars:
1. Boston College, Demand Estimation and Modelling Conference, December 12-14, 2013
2. EARIE Conference Evora, August 30-31, 2013
3. Barcelona Summer Forum, Applied IO, June 20, 2013
4. CSIO/IDEI Workshop, Toulouse, April 5-6, 2013
5. Workshop on Econometrics and I.O. of Food and Nutrition, Toulouse, December 17-18, 2012
6. Winter Meeting of UK Industrial Economics Network, London, December 14, 2012
7. French Econometrics Conference, Rennes, November 23, 2012
8. ESEM Malaga, August 27-31 2012
9. EARIE, Roma, September 2-4, 2012
10. INRA IDEI Conference, Toulouse School of Economics, June 21, 2012
11. American Economic Association Meeting, Chicago, January 6, 2012
12. Royal Economic Society, Cambridge, March 28, 2012

Seminars:
1. Paris School of Economics, March 12, 2012
2. Imperial College Business School, London, January 24, 2012
3. Research Institute of Industrial Economics - IFN, Stockholm, November 14, 2012
4. CREST seminar, Paris, September 20, 2012
5. University of Manchester, October 24, 2013
6. Duke University, October 29, 2013
7. North Carolina State University, October 30, 2013
8. London School of Economics, November 11, 2013

Diet-related diseases are on the rise in most Western countries. There is growing evidence that children's health and educational outcomes are related to diet, and that these have lasting effects into adulthood. Poor quality and excess consumption of food are believed to be among the most important causes of the rise in these diet-related diseases. Addressing these public health issues is a key policy concern. Because of this there is increased interest from both academics and policy makers in increasing the evidence base for policy formation, and in understanding how policies are likely to affect nutritional outcomes of individuals. The impact of policy on diet will depend crucially on how consumers adapt their food purchasing behaviour in response to the policy, on how firms in turn respond in terms of the prices they set and the foods they offer, and on how food purchased is shared between individuals within households.

There have been important methodological advances in modelling these behaviours, but the application of these to policy have relied on several restrictive distributional assumptions, which are known to heavily influence the results. Our aim is to develop a better set of tools that allow us to relax some of these restrictive assumptions, and that will help us to understand better how potential policy interventions will affect consumer and firm behaviours, and so to understand better the subsequent impact on nutritional outcomes.

The research teams at IFS and Toulouse have different expertise, and an aim of this project will be to forge closely links. We believe this will lead directly to higher quality research, and that it will be beneficial to both research teams in that we will share experiences and skills, and it will help develop research capacity in both institutions.

The research will organised around three closely inter-related projects:

Project 1 will take as its starting standard models of demand in oligopolist markets and extend them by relaxing the restrictive distributional assumptions that are commonly applied to allow for rich and flexible heterogeneity in household preferences and for firm's responses to potential policy reforms.

Project 2 will extend existing methods that are used for identifying individual consumption from household level expenditure data by bringing additional data to bear on the problem, and so allowing for more heterogeneity along observable dimensions.

Project 3 will apply these structural models to the analysis of policy in the area of food and nutrition. The policies we consider will include those that provide consumers with more information on the nutritional characteristics of specific foods, policies that tax certain types of food or inputs to the food manufacturing process, and policies that seek to regulate or incentivise the firms that produce or retail foods to provide healthier foods.

The main output of these projects will be academic papers that are aimed at publication in top international journals. The research will also result in work that is of direct policy interest, and will also be of interest to the broad general public.