À la suite des réformes successives de la Politique Agricole Commune (PAC), les soutiens publics par des prix ont diminué au profit de soutiens directs aux revenus agricoles. Cela a exposé les agriculteurs français à une grande volatilité des prix, reconnectés avec les prix mondiaux.Cette thèse mesure l'évolution de la productivité de l'agriculture française dans un modèle dynamique stochastique en intégrant la récente augmentation de la volatilité des prix. Nous étudions le lien dynamique entre le risque de prix, les décisions des agriculteurs et la productivité dans le cadre de l'estimation structurelle. La revue de la littérature présentée dans le chapitre 2 décrit la productivité comme un résidu et souligne les problèmes de mesure des données du capital et le problème de l’endogénéité dans l’estimation primale.Le chapitre 3 compare les méthodes numériques permettant de résoudre et d'estimer les modèles d'équilibre général dynamique stochastique (DSGE) ou de type DSGE, dans lesquels le capital et la productivité sont des variables d'état. Le chapitre 4 estime la productivité dans un modèle dynamique stochastique en utilisant l'approche d'entropie maximale généralisée (GME). Nous trouvons que la croissance de la productivité de l’agriculture française a diminué après la réforme de la PAC, à cause de l'augmentation de la volatilité des prix. En effet, le risque de prix impacte la productivité négativement à travers les choix de production, de consommation, d’investissement et d’emprunt des agriculteurs. Le chapitre 5 simule les impacts de marché des instruments de la P / The EU has adopted many reforms of the Common Agricultural Policy (CAP) in the past decades. Price support has decreased, and decoupled payments have been introduced. As a result, European agricultural prices have become more volatile, in line with world prices.This dissertation measures the evolution of the productivity of French agriculture in a dynamic stochastic farm decision model in the new economic context with increased price volatility. On this basis, it studies the dynamic link between price risk, farmer decisions, and productivity in the structural estimation framework. The literature review in Chapter 2 describes productivity as a residual and emphasizes the measurement issues from the unobserved capital data series and the endogeneity problem in primal estimation.Chapter 3 compares the numerical methods to solve and estimate nonlinear dynamic stochastic general equilibrium (DSGE) or DSGE-like models, in which capital and productivity are latent state variables. Chapter 4 estimates productivity in a dynamic stochastic decision model based on the generalized maximum entropy (GME) approach. We show that the productivity growth in French agriculture has slowed down and become more volatile following the rise in price volatility. Overall, price risk has an impact on productivity in the way that when exposed to high risks, farmers change their production, consumption, investment and financial borrowing decisions, which in turn affects the realized productivity negatively. Chapter 5 simulates the market impacts of the CAP instruments in a dynamic GTAP-AGR CGE mode

As mentioned in the introduction, the objective of this work has been to get a more realistic understanding of economic decision making processes by adopting an interdisciplinary approach which takes into consideration at the same time economic and psychological issues. The research in particular has been focused on the psychological concept of categorization, which in the standard economic theory has received until now no attention, and on its implications for decision making. The three experimental studies conducted in this work provide empirical evidence that individuals don not behave according to the perfect rationality and maximization assumptions which underly the SEUT, but rather as bounded rational satisfiers who try to simplify the decision problems they face through the process of categorization. The results of the first experimental study, on bilateral integrative negotiation, show that most of the people categorize a continuum of outcomes in two categories (satisfying/not satisfying), and treat all the options within each category as equivalent. This process of categorization leads the negotiators to make suboptimal agreements and to what I call the ?Zone of Agreement Bias? (ZAB). The experimental study on committees? decision making with logrolling provides evidence of how the categorization of outcomes in satisfying/not satisfying can affect the process of coalition formation in multi-issue decisions. In the first experiment, involving 3-issues and 3-parties decisions under majority rule, the categorization of outcomes leads most of the individuals to form suboptimal coalitions and make Pareto-dominated agreements. The second experiment, aimed at comparing the suboptimizing effect of categorization under majority and unanimity rule, shows that the unanimity rule can lead to a much higher rate of optimal agreements than the majority rule. The third experiment, involving 4-issues and 4-parties decisions provides evidence that the results of experiments 1 and 2 hold even when the level of complexity of the decision problem increases.

As mentioned in the introduction, the objective of this work has been to get a more realistic understanding of economic decision making processes by adopting an interdisciplinary approach which takes into consideration at the same time economic and psychological issues. The research in particular has been focused on the psychological concept of categorization, which in the standard economic theory has received until now no attention, and on its implications for decision making. The three experimental studies conducted in this work provide empirical evidence that individuals don not behave according to the perfect rationality and maximization assumptions which underly the SEUT, but rather as bounded rational satisfiers who try to simplify the decision problems they face through the process of categorization. The results of the first experimental study, on bilateral integrative negotiation, show that most of the people categorize a continuum of outcomes in two categories (satisfying/not satisfying), and treat all the options within each category as equivalent. This process of categorization leads the negotiators to make suboptimal agreements and to what I call the ?Zone of Agreement Bias? (ZAB). The experimental study on committees? decision making with logrolling provides evidence of how the categorization of outcomes in satisfying/not satisfying can affect the process of coalition formation in multi-issue decisions. In the first experiment, involving 3-issues and 3-parties decisions under majority rule, the categorization of outcomes leads most of the individuals to form suboptimal coalitions and make Pareto-dominated agreements. The second experiment, aimed at comparing the suboptimizing effect of categorization under majority and unanimity rule, shows that the unanimity rule can lead to a much higher rate of optimal agreements than the majority rule. The third experiment, involving 4-issues and 4-parties decisions provides evidence that the results of experiments 1 and 2 hold even when the level of complexity of the decision problem increases.