Decisions: theories, experiments and applications.

Summary This thesis explores several intertwined dimensions of decision research. In particular, while focusing on decisions under risk and uncertainty, this work illustrates the transversality of approaches that characterize this field of investigation by bringing together experimental study, theoretical development and application. The continuous dialogue between these different approaches to decision research is essential to its development, as empirical investigations test and illuminate formal theories and further ensure their meaningful application in theoretical and empirical contexts, thus contributing to a virtuous dynamic of "creative destruction" in the field of study. For example, a large body of empirical evidence shows violations of expected utility theory. In response, decision theory and behavioral economics have provided a wide variety of non-expected utility theories. However, the existing evidence does not clearly distinguish between these theories. In the investigation of the sources of violations, particular attention is traditionally devoted to testing several variants of the independence axiom. Yet, in the domain of earnings, many of the best-known non-expected utility models adhere to the minimal behavioral restrictions traditionally known as Savage's (1954) P3 and P4 axioms, which allow for the separation of tastes, as captured by a utility function, and beliefs, often captured by a distorted probability. In the first chapter of this thesis, we derive a nonparametric procedure for testing the separability hypothesis of tastes and beliefs and apply it to the results of two experiments. Although P3 is rarely violated, our test reveals widespread and pronounced violations of P4, suggesting that the assumption of separability of tastes and beliefs may not hold in empirical settings. On the theoretical side, the question of the separation of tastes and beliefs has apparently been closed by a series of papers by Ghirardato and Marinacci on biseparable preferences. Inspired by the findings of the previous chapter, the second chapter of this thesis aims to reopen this question by investigating whether, or under what theoretical conditions, such a separation actually holds and what its implications are. In particular, we will provide separability conditions in terms of preference midpoints and the new concept of likelihood midpoints. In the third chapter, we use recent developments in decision making criteria to provide an extension of Brandenburger and Stuart's (2007) seminal biform games setup. While biform games provide the theoretical basis for formal work in value-based strategy, our framework helps reconcile observed behavior in applied settings with the theory. In particular, we apply the results of our theory to the choice to involve substitutes for complementors in business ecosystems, a central decision in competitive strategy. Our solution has several advantages. First, it subsumes the original biform game framework and seamlessly integrates recent related work that provides bounds on value capture. In addition, it addresses issues such as the possible non-uniqueness of solutions and the invariance of the structure of the competitive environment while retaining the role of competition in determining value capture. Finally, it allows for richer preference representations that, for example, can include subjective distortions of the objective chances of value capture.