This research project is concerned with dynamic processes in which agents learn from or adapt to the actions of their neighbors, and the spatial patterns of conformity that result. Processes of this type cover a wide variety of phenomena, including the diffusion of new technologies, the transmission of social norms, and sorting by class, race, or ethnicity. The research extends our theoretical understanding of such processes, and where possible compares predictions of theory with evidence. Three topics are investigated. The first is the relationship between the macroscopic behavior of these processes and the micro-response rules governing individual behavior. The hypothesis is that certain long-run properties, such as the relative likelihood of various equilibrium and disequilibrium patterns, are robust under various specifications of individuals' adaptive behaviors. This would enhance the usefulness of such models in applications where individual response rules are heterogeneous and difficult to estimate statistically. A second topic of theoretical investigation is the inertia of local interaction processes --the expected time it takes the system to reach (or come close to) long-run conditions from an arbitrary initial state. Extending previous work in this area, it will be shown how inertia is related to geometric properties of the social network. Of particular importance is the degree to which agents interact in small, close knit groups instead of large, disparate ones. The third part of the research tests certain qualitative predictions of these theories against empirical evidence. The primary application is an empirical study of principal-agent contracts agriculture, using an extensive data base covering several thousand farms in Illinois from 1977-1995. The level of detail in these data presents a unique opportunity to test various hypotheses suggested by theory. One hypothesis is that contract terms tend to exhibit a local conformity effect, that is, they cluster around certain values that can be construed as local norms. Once established, these norms shape the expectations of the contracting parties, and thus tend to perpetuate themselves through the force of precedent. This model of norm formation leads to specific predictions about spatial and temporal variations in contract terms that can be tested empirically. A second hypothesis suggested by the theory is that such norms can cause measurable distortions from neoclassical price equilibrium, that is, they create a stickiness in expectations that impedes the equilibration of factor returns. The overall goal of the research is to identify situations where dynamic models of local interaction lead to predictions that differ from those of neoclassical equilibrium models, and to put some of these predictions to the test empirically.
