This thesis investigates the problem of market mechanism design for supply chain management and e-marketplace development. Instead of viewing a market as an isolated entity, we consider each market to be a part of a supply chain and study the effects of interactions between different markets in the supply chain. We focus on three important issues in relation to market mechanism design and investigate them using a variety of approaches, such as game-theoretic, strategic and experimental methods. Capacity allocation has been a research topic in management science and operations research for several years, dealing with imbalance between supply and demand in a market. In recent years, this topic, known as resource allocation, has become an active research area in computer science. The first issue we consider is how to design market mechanisms for capacity allocation. We deal with capacity allocation problems in a supply chain model rather than in a single market. Based on this model, we examine the game-theoretic properties of allocation mechanisms, such as efficiency, profit maximising, and truth-telling, with respect to quantity competition and price competition in a related market of the supply chain. We prove that a few typical allocation mechanisms that have been generally used in industry are sensitive to supply chain settings. For instance, proportional allocation is no longer a Pareto optimal allocation mechanism in our supply chain model in contrast to the results in the existing researches that deal with monopolistic downstream markets. There are two reasons for this. The first reason is that competition in our supply chain model makes retailers submit greater orders than the quantities that maximises the total retailer profit. The second reason is that proportional allocation does not strictly prioritise allocations to the best performing retailers. As a result, this allocation mechanism does not maximise the total retailer profit. In order to achieve Pareto optimality, we propose max-max allocation that strictly prioritises the greater orders. We prove that this allocation mechanism satisfies Pareto optimality in our competitive supply chain model. However, under this allocation mechanism, we show that retailers inflate orders. This phenomenon is known as bullwhip effect in supply chains which leads inaccurate transmission of order information. In order to prevent order inflation, we design a new allocation mechanism, capped-allocation, which assigns maximum allocation quantity prior to order submission. In addition to analyse how supply chain settings influence properties of allocation mechanisms, we undertake equilibrium analysis that shows how allocation mechanisms influence market behaviour in supply chains. We prove that order quantities are greater under proportional allocation than the ones under uniform allocation. Under price competition, proportional allocation leads higher market price than uniform allocation. A key reason is that strict imbalance of allocation leads higher market prices and proportional allocation tends to allocate heterogeneously. The second issue we investigate is design of market mechanisms for online markets. Typical market participants in online markets are loosely connected even though online market owners require their business partners such as sellers. In order to bind these market participants, we introduce an approach to the modelling of online markets as supply chains, in which a coordination mechanism is applied to the market between the online market owner and the sellers. We examine a set of coordination mechanisms based on fixed-fee contracts, revenue-sharing contracts, and profit-sharing contracts in relation to different marketing strategies, such as advertisement. We prove that the fixed-fee contract achieves coordination while the revenue-sharing contract does not achieve coordination if we do not consider the effect of advertisement. These results are opposed to the existing researches in traditional intermediaries. A key reason is that online market owners do not deal with transactional costs, in contrast to the transactional intermediaries. Therefore, revenue-sharing may charge more than the coordinated case. We design a new online market contact based on the idea of profit-sharing and prove that it achieves coordination between online market owners and sellers with advertisement. This contract shares both revenue and cost between these parties. Finally, we explore the problem of market mechanism design using autonomous trading agents. We introduce a formal representation of market policies, such as accepting policies, charging policies, pricing policies and matching policies, based on double auction mechanisms. By utilising the Market Design Game platform for the Trading Agent Competition (TAC), we analyse how specific market policies influence market behaviour of specific types of autonomous trading agents using experimental methods. We design a range of market policies and test them with three types of trading agents: random, learning and human-like bidding agents. We find that market performance, such as the profit of market makers, allocation efficiency and transaction volume, improves with time. The experiments also show that the periodic clearing policy gives market makers greater profits than the continuous clearing for all three types of trading agents we use. However, for learning and human-like bidding agents, the continuous clearing policy is preferable to the periodic clearing policy because it can improve allocation efficiency and transaction volume. Based on this experimental approach, we design and implement our market mechanism. This mechanism has been tested and proved as a robust, efficient and effective market mechanism as a winner of the tenth Trading Agent Competition in Market Design.