The goal of this dissertation is to investigate the enabling role that agent based simulation plays in business and policy. The aforementioned issue has been addressed in this dissertation through three distinct, but related essays. The first essay is a literature review of different research applications of agent based simulation in various business disciplines, such as finance, economics, information systems, management, marketing and accounting. Various agent based simulation tools to develop computational models are discussed. The second essay uses an agent-based simulation approach to study important properties of the widely used most popular news recommender systems (NRS). This essay highlights the major limitations of most popular NRS in terms of: (i) susceptibility towards manipulation and (ii) unduly penalizing the article which may have "just" missed making the cutoff in most popular list. A probabilistic variant of recommendation has been introduced as an alternative to most popular list. Classical results from urn models are used to derive theoretical results for special cases, and to study specific properties of the probabilistic recommender. In addition to simulations, various statistical methodologies are used, such as regression based methodologies as part of a broader decision analysis tool. The third essay views firms as agents in building regression based empirical models to investigate the impact of outsourcing on firms. Using an economy wide panel data of outsourcing expenses of firms, the third essay first investigates the value addition by the IT backgrounds of project owners in managing IT related projects. Then it investigates the impact of peer-pressure on a firm's outsourcing behavior.