Perth, Western Australia (pop. 1.6m) derives 60% of its public water supply from the Gnangara groundwater system (GGS). Horticulture, domestic self-supply, and municipal parks are other major consumers of GGS groundwater. The system supports important wetlands and groundwater-dependent ecosystems. Underlying approximately 2200km 2 of the Swan Coastal Plain, the GGS comprises several aquifer levels with partial interconnectivity. Supplies of GGS groundwater are under unprecedented stress, due to reduced recharge and increases in extraction. Stored reserves in the superficial aquifer fell by 700GL between 1979 and 2008. Over a similar period, annual extraction for public supply increased by more than 350% from the system overall. Some management areas are over-allocated by as much as 69%.One potential policy response is a trading scheme for groundwater use. There has been only limited trading between GGS irrigators. Design and implementation of a robust groundwater trading scheme faces hydrological and/or hydro-economic challenges, among others. Groundwater trading involves transfers of the right to extract water. The resulting potential for spatial (and temporal) redistribution of the impacts of extraction requires management. Impacts at the respective selling and buying locations may differ in scale and nature. Negative externalities from groundwater trading may be uncertain as well as not monetarily compensable.An ideal groundwater trading scheme would ensure that marginal costs from trades do not exceed marginal benefits, incorporating future effects and impacts on third-parties. If this condition could be met, all transactions would result in constant or improved overall welfare. This paper examines issues that could reduce public welfare if groundwater trading is not subject to well-designed governance arrangements that are appropriate to meeting the above condition. It also outlines some opportunities to address key risks within the design of a groundwater trading scheme. We present a number of challenges, focusing on those with hydrological bases and/or information requirements. These include the appropriate hydrological definition of the boundaries of a trading area, the establishment and defining of sustainable yield and consumptive pool, and the estimation of effects of extractions on ecosystems and human users. We suggest several possible design tools. A combination of sustainable extraction limits, trading rules, management areas, and/or exchange rates may enable a trading scheme to address the above goals

The development of a market in groundwater usage rights can be inhibited by constraints arising from the institutional context. Such impediments may reduce the potential gains from trade and may generate high transaction costs for prospective traders. We analyse the regulations and policies influencing groundwater transfers in a case-study area-the Gnangara groundwater system around Perth, Western Australia-and identify significant impediments to a groundwater market. Property rights are found to be conditional, temporary, and vulnerable to amendment. Regulatory approval is required for all transfers. Facilitating infrastructure is lacking, and price information is unavailable. Management area boundaries reflect land ownership and use rather than hydrogeological realities; the limitation of transfers to within these boundaries eliminates much of the potential for gains from trade. Over-allocation and weak monitoring also impede the development of a market. The current management system is likely to obscure unmet demand for water-rights transfers between users and usage-types.

The Gnangara groundwater system in Western Australia occupies some 2,200km 2 , supports multiple ecological systems and human uses, and is under unprecedented stress due to reduced rainfall and over-extraction. The basin is currently managed according to command and control principles by the state's Department of Water. This paper examines some of Ostrom's "situational variables" for the analysis of institutional choice -the selfprovision of institutional arrangements in common-pool resources situations -as they relate to the Gnangara case. The paper approaches the topic of collective action not as a niche concept which may be fitted only to certain specific cases, but as a basic and natural mode of human co-operation and interaction when faced with inter-dependent interests and in the absence of militating factors. We therefore conduct the analysis from the perspective of identifying elements of the current management approach -as well as of the shared norms, expectations, and attitudes of the appropriators -which could be altered to allow collective governance to develop, at least at some scale within the overall management regime. We use data from a set of water licence documents obtained from the Department of Water, among other data sources. A number of factors are identified as inhibiting the development of collective action at present. Current arrangements are topdown in nature, with all rules, monitoring, and enforcement supplied by the state-level management agency. Current norms and expectations among the appropriators appear to be competitive rather than co-operative, and discount rates appear to be high. In view of the size of the resource, and the large number and heterogeneity of appropriators, we conclude that the use of 'nested' organisational units -beginning at the smaller scale -will be a key component of efforts to develop the requisite social and institutional capital.Further, we conclude that there are several historical and other factors in this case whose net effect is to prejudice the unassisted development of collective action institutions by appropriator efforts alone, and that significant external support will be required from government agencies. This study highlights some important aspects of the regulatory apparatus in place, their likely effects upon the resource appropriators in terms of attitudes and behaviours, and the resulting impacts on the common-pool resource upon which wildlife, ecosystems, and the appropriators all depend.