Given the credence nature of food quality and food safety attributes, consumers cannot easily verify whether food is high quality or is safe to eat, thus they rely on abstract systems of regulation and quality signals such as brands to make informed consumption choices. In fact, trust is recognized as a rational strategy that reduces consumers’ uncertainty when purchasing goods with credence attributes. While trust in food is a topical issue in an era of increasingly complex food systems, how trust and more precisely brand trust affects consumers’ perceptions of food is a relatively new research area in food economics. This thesis aims to answer questions such as what drives trust in the actors within the Canadian food system (i.e. government, farmers, food manufacturers, and food retailers) and in food brands, and the relationship between that trust and consumer confidence in food quality and food safety. Previous studies on institutional and system trust have been carried out primarily in the sociology, marketing and political sciences disciplines, while a few studies in food economics have investigated the influence of institutional trust and reported the degree of public trust in market actors. This study extends previous research on consumer trust in the context of food by developing a conceptual model in which trust in the food system and brand trust are expected to evolve to public confidence in credence attributes and lead to consumer commitment to food brands. Inspired from a comprehensive synthesis of the literature on consumer trust, the theoretical background suggests that consumer confidence in food attributes is jointly determined by trust in the food system (system-based trust) and brand trust (product-based trust), and it is moderated by consumer characteristics (personal-based trust) – namely: risk, past consumption experience and ethically-motivated behaviour. As well, consumers are assumed to perceive an actor or a brand as trustworthy through the influence of four postulated dimensions of trust: perceived competence, credibility, benevolence and reputation. A Structural Equation Modelling (SEM) approach is used for the empirical analysis. Data were gathered through an online survey of consumers conducted across Canada in July 2012 focusing on fresh chicken and packaged green salad products. The results show that individually none of the postulated antecedents of trust (i.e. competence, credibility, benevolence and reputation) was a significant driver for trust, especially for packaged salad. Unlike previous research on institutional trust and brand trust that uses a number of separate dimensions to measure trust predominantly in non-food contexts, this thesis finds that trust in the food system and food products could be measured differently by taking into account the interactive effects of perceived competence, credibility, reputation and benevolence on public trust. As such, the findings suggest that Canadian consumers tend to trust brands of chicken and packaged salad when these products are perceived as high quality, are backed by credible information, have a good reputation and, at the same time, enhance consumers’ welfare. In fact, a brand cannot be perceived as high quality and safe to eat (brand competence) without containing transparent information signalling its quality and safety (brand credibility). Additionally, results reveal some apparent product-specific effects: brand trust matters in fostering consumer confidence in chicken but not for salad. Furthermore, trust in the food system as a whole appears to be more influential in leading to public confidence in credence attributes than trust in food brands in the case of chicken. As such, it appears that trust in actors within the food supply chain is more important than relying on individual products. As well, the analysis shows that while psychographic variables (risk aversion, past consumption experience) and some demographics (e.g. gender, education) moderate the relationship between trust in the food system and brands for chicken, this is not the case for salad. In terms of marketing implications, the results suggest that while brands are useful signalling mechanisms, trust in these brands is not the main driver for consumer confidence in credence attributes. Furthermore, the much stronger effect of system trust implies that decision-makers would benefit by investing in building trust relationships with the public. Transparent communication and credible assurances about the practices and the intentions within the food system could be a way to gain and maintain public trust and, ultimately, consumer confidence in food quality and food safety.

Cette thèse a pour objectif d'étudier les programmes de labellisation environnementale. Dans un premier temps nous différencions deux types de programmes de labellisation, les programmes de type ISO I qui donne une information discrète (les produits sont labellisés ou non), et les programmes de type ISO III qui eux révèlent l'information de manière continue (les cartes énergétiques). Nous montrons que les programmes de type de ISO III conduisent à des problèmes d'auto-sélection au sein de la ligne du produit du monopole alors que les programmes de type ISO I permettent à celui-ci de parfaitement discriminer entre ces produits. Ensuite dans un second chapitre, réalisé en collaboration avec J.W. Maxwell et R. Harbaugh, nous faisons l'hypothèse que le consommateur n'a qu'une idée incertaine de la difficulté du standard qu'un produit doit atteindre pour être labellisé. Nous montrons que la prise en compte de cette hypothèse diminue les incitations des entreprises à se faire certifier et l'efficacité des programmes de labellisation. Nous prouvons aussi que les entreprises choisissent de manière stratégique entre les différents labels disponibles sur le marché. Enfin dans un dernier chapitre réalisé en collaboration avec J. Rosaz et F. Poinas, nous choisissons d'utiliser l'économie expérimentale afin d'étudier le comportement du consommateur ou plus précisément la révision des croyances d'un sujet lorsque celui-ci reçoit un signal imparfait. Le signal envoyé au sujet consiste en une restriction des états possibles de la nature et s'apparente ainsi au label observé par le consommateur sur le marché. Nous montrons qu'en moyenne ces signaux permettent aux sujets d'améliorer leur estimation, cependant certains signaux trop incertains peuvent avoir des effets pervers. / This thesis has for goal to study environmental labeling programs. In the first chapter, I use a self-selection model to study environmental labelling program in a context of multiproduct monopoly. I show that the manner by which information is conveyed affects the firm strategies. I prove that programs, which disclose continuous information such as report cards, can lead to self-selection issues within the monopoly product line, whereas those which give discrete information such as the EU flower allow the monopoly perfectly discriminating among consumers. In the second chapter, joint with J. Maxwell and R. Harbaugh, we relax the assumption that consumers know the exact standard that the product has to meet to be labelled. The uncertainty over the standard leads the consumers to jointly estimate the product quality and the standard difficulty. This undermines the ability of voluntary certification to reduce information asymmetry. In the third chapter, joint with J. Rosaz and F. Poinas, we analyze belief updating behaviors with imperfect signals in an experimental setting. The novelty of the paper is to consider a signal that restricts the set of the possible states of the nature. One of the objectives of this chapter is to study the consumer updating behaviors when they observe a label on a product but with the methodologies developed in the Bayesian updating literature. We show that this kind of signals increases the correctness of the subject estimations. However these signals can also let the subjects very uncertain about their estimation which can conduct to adverse effects.