The theory of Agency, specifically that developed by Jesen and Meckling (1976), will be the subject of examination. Agency theory has been the subject of extensive research since its introduction in modern form by Jensen and Meckling (1976). The generality of the theory of Agency appears unquestionable and it has been widely adopted. Surprisingly, however, the model correctly predicts particular phenomena under investigation in only the simplest of instances, and even in the simplest of instances there are cases where the simple agency model has limited success. Possible reasons for this failure may lie in the assumed universalist foundation and in the common formulation regarding agent behaviour, that all agents are self-interested rationalists seeking to maximise their own utility to the disregard of their principal's interest. While the hypothesis of self-interested rationalism may be apt in some contexts it may be misleading or inadequate in others. This is especially so when the narrow interpretations of self-interested rationalism are used. Human beings are more complex in their totality than can be represented in any parsimonious model. This is particularly a problem when model predictions are not empirically supported. Aspects omitted in a model may be a source of the misfit between prediction and observation. An extended conceptualisation and reformulation of agent behaviour is presented. An approach is developed that addresses the context of agent behaviour, the socio-environment within which the agent interacts. The context particularly refers to the institutional affiliations and interactions that influence agent behaviour through their belief structure (i.e., their Belief-Desire-Intention, BDI, model of rational action). Through the use of an institutional framework contextual analysis is incorporated into the theory of agency and ultimately agent behaviour. This agent is termed a socio-environmental rationalist agent (SERA) which is contrasted with the self-interested rationalist (SIR) agent in the existing agency literature. This research utilises an object-oriented approach to develop a simulation of the extended conceptualisation and reformulation of agent behaviour. Simulations investigate agent behaviours and outcomes at the micro (specifically through individualised SERA and SIR formulations) and macro (specifically through a multi-agent SERA community formulation in the context of the EU financial accounting harmonisation process) levels. Netlogo is the simulation tool through which this is attained. The simulation demonstrates how alternative formulations of rationality lead to different outcomes and these differences are evident at both levels. Importantly the extended model has outputs that are more in tune with current empirical evidence. The analysis thus demonstrates the plausibility of the extended conceptualisation and reformulation and the need to incorporate the context of behaviour more fully within the analysis of the principal-agent relationship. Through this extended examination of agent behaviour further theoretical and practical insights regarding the understanding of agent behaviour, the principal-agent problem and relationship, multi-agent communities, and of business and society in general may be attained. This dissertation provides one step in advancing our fundamental understanding of the principal-agent problem. The scope and power of agency analysis can be substantially extended using the approach and methods outlined, particularly beyond that present in existing Agency research.