Trade policy liberalisation requires institutional change, in the sense of a change in the rules of the game. The question is whether these changes produce "superior institutions" judged in terms of a reduction of transactions costs; improved coordination; stronger strategic commitment to investing in needed specific assets; and allocative efficiency. In conventional approaches to the analysis of liberalisation, changed institutional arrangements are studied, but they tend to be considered in the category of "practical details": important but not especially intellectually interesting. In contrast, this paper argues for a parallel approach to the study of the effects of liberalisation on the rural poor, in which institutional matters are central. A broad range of institutional issues is considered, informed by a theoretical framework provided by the various strands within institutional economics. The framework set out and discussed leads to the contention that smallholder agriculture in poor countries needs coordinated market economy (CME) type institutions if it is to develop, at least at the earlier stages. Ideally, these would be based on deliberative institutions, working horizontally inside a sector, and also vertically along the supply chain. It is argued that the way forward is likely to involve a rethinking of the role of the state (at sub-national, national and international - aid donor - levels) and of the roles of producer organisations and other stakeholder (including trader) associations. The aim must be to find a way in which the state and other powerful actors can initiate deliberative processes and take a lead in encouraging appropriate asset specific investments, while at the same time planning to fade into the background as initial success is achieved. These conclusions challenge conventional analysis of trade policy liberalisation in poor countries and also challenge institutional specialists to provide insights, ideally quantifiable, into the consequences of those liberalisation policies which drive changes in such features as "non-standard institutional arrangements"; non-market coordination; and the roles of government.