Relationships between bounded rationality and transaction cost theories are discussed and their connections with stochastic theories of industrial evolution are considered. While these theories have their limitations, they are useful but have been ignored in many public policy prescriptions, especially those involving markets. For example, as discussed, these theories have failed, on the whole, to influence competition policy and the design of more efficient systems for public administration (contracting out, labour contracts for public employment, adoption of the user-pays principle and use of performance budgeting and accounting), as well as in policies to remove market frictions. The result may be less efficient systems than otherwise achievable and diminished long-term economic performance. It is also pointed out that market transaction costs are sometimes important contributors to improving the performance of market systems by increasing their stability. The presence of frictions and diversity of behaviours can sometimes improve the long-run performance of competitive systems.