2. Robust political economy and comparative institutions analysis

The concept of RPE posits that the comparative evaluation of political–economic institutions must recognise the significance of human imperfections. An institution may be judged robust if it is able to operate effectively in non-ideal circumstances (Boettke and Leeson, Reference Boettke and Lesson2004; Boettke et al., Reference Boettke, Coyne and Leeson2007; Leeson and Subrick, Reference Leeson and Subrick2006). If people are perfect or at least perfectible, then questions of comparative institutional analysis cease to arise because beneficial outcomes are likely irrespective of the framework in place. ‘Ideal theories’ that assume away human imperfections are of limited utility because the necessity of choosing between institutions arises owing to the constraints presented by non-ideal conditions.

The RPE framework identifies two human imperfections that should be accounted for in the task of institutional evaluation. The first is limited knowledge or bounded rationality. People are constrained in their cognitive capacities and must make decisions on the basis of theories that may prove mistaken (Gigenrenzer and Selten, Reference Gigenrenzer and Selten2002; Hayek, Reference Hayek1948a). Decision-making occurs under chronic uncertainty where information is often contradictory. This uncertainty is not reducible to a problem of imperfect information that can be addressed by searching for additional data known in principle to be available. The latter understanding assumes that mistakes may be anticipated by decision-makers but may not be worth avoiding given the cost of searching out additional data. The problem of uncertainty, however, extends beyond issues that can be analysed in probabilistic terms to encompass ‘radical ignorance’. It is not simply that actors may not know which possibility out of a given set will occur, but that the set of possible outcomes is unbounded and thus unknowable. Robust institutions, therefore, are those which allow people to adapt to and learn from un-anticipated mistakes and unforeseen opportunities (O'Driscoll and Rizzo, Reference O'Driscoll and Rizzo1996).

The second human imperfection that the RPE framework draws attention to is the problem of opportunism. Actors may follow their own self-interested goals – material or non-material – rather than some notion of the ‘public interest’. Self-interested behaviour itself should not be seen as evidence of ‘imperfection’ but the possibility that people may act opportunistically in pursuit of their interests to the disadvantage of others must be taken into account. Incentives matter and, as a consequence, institutions must be judged on their capacity to channel potentially self-interested motivations in ways that generate mutually beneficial outcomes (Ostrom, Reference Ostrom2006).

It should be emphasised here that the existence of limited knowledge is fundamental to the RPE framework whereas the problem of opportunism is not. If the assumption of opportunistic behaviour is dropped, the need for comparative institutional evaluation remains because even other-regarding actors require a process that enables them to cope with the possibility of error. If the assumption of limited knowledge is dropped, however, then the case for comparative institutions analysis is made redundant because generalised omniscience implies that people can monitor and discipline narrowly opportunistic actors irrespective of the institutional context.

The RPE perspective as set out above implies recasting how economists and especially those within the neo-classical tradition evaluate institutional regimes. At issue is the frequent failure of general equilibrium theories that emphasise the superiority of ‘free markets’ or more ‘interventionist’ regimes, to explain how the private and public actors concerned are able to acquire the levels of information needed to achieve efficient outcomes. Unless such an account can be offered then the value of these approaches is compromised because the normative conclusions amount to no more than proof of what is in fact assumed (Boettke, Reference Boettke1997; Hayek, Reference Hayek1948a: 45).

To question the implications of general equilibrium analysis, however, is not to imply that the concept has no use. From an RPE perspective, what is required is not abandonment of equilibrium as an idea but of those approaches that treat equilibrium states either as a description of the real world or as a description of how the world can and should be made to function (Boettke, Reference Boettke1997). Departures from ideal typical conditions focus attention on how different institutions cope with the actual conditions of ‘real-world’ human interaction. The task of an RPE, therefore, is not to highlight the failure of institutions to attain idealised states of ‘perfect’ efficiency but to provide explanations for the degree of coordination witnessed under different institutional regimes as they grapple with non-ideal constraints.Footnote ¹

Within this context, one of the most important applications of the RPE framework has seen an attempt by classical liberal political economists to specify the relative robustness of private, market-based forms of resource allocation. Post-war neo-classical economics concluded that given general equilibrium assumptions of perfect information, perfect competition and zero transaction costs there was little to choose in efficiency terms between a free enterprise system and a regime where most investment and production decisions are made by the state. It further suggested that in those circumstances where market-based systems take place under conditions of imperfect information, imperfect competition and positive transaction costs that targeted government intervention could improve on the market outcome (Boettke, Reference Boettke1997).

From the perspective of classical liberal political economy, however, these conclusions said more about the questionable value of the general equilibrium perspective as a normative benchmark than they did about the relative robustness of ‘real-world’ market and state-directed systems. The classical liberal case for the superiority of private markets rests not on claims that market actors are fully rational, possess complete information, or that transaction costs are zero, but on the suggestion that market processes may be better placed than state-directed alternatives to cope with a world where these conditions are absent (for example, Boettke, Reference Boettke1997; Leeson and Subrick, Reference Leeson and Subrick2006).

Following Hayek (Reference Hayek1948b), a market system is judged more robust in a context of limited knowledge because it reduces the need for producers and consumers to have direct knowledge of all of the facts that impinge on the scarcity of resources. Dispersed individuals and organisations make bids for resources reflecting preferences, the availability of substitutes, and alternative production techniques much of which is tacit and known only to the actors concerned. As they engage in trade, however, ‘bits’ of this information are communicated indirectly to others via the generation of price signals that help to inform subsequent adaptations made by these actors. The case for market-generated prices made here is not that they convey ‘perfect information’, but that they convey more socially useful knowledge than would be possible were a single planning authority to be responsible for resource allocation. Such an authority would need to rely on direct commands to allocate resources but no command structure could ever grasp as much about the constantly changing conditions of supply and demand as a network of exchanges reflecting the knowledge and actions of multiple private traders (see Friedman, Reference Friedman2006, for a recent example of this position).

A second implication of taking limited knowledge as a starting point is the recognition that ‘real-world’ decision-makers whether public or private do not know what goods to produce, at what prices to sell and which input combinations to use. The task of coordinating economic plans occurs under uncertainty where information is often contradictory. Private markets, however, may be better placed to facilitate a process of learning and adaptation in light of this uncertainty. Profit and loss accounting combined with freedom of entry and exit allows a wider range of competing production and consumption ideas to be evaluated than would be the case were there only one decision-making authority or were such an authority to determine how much and what kind of decentralised experimentation to permit. Mistakes will be made but the possibility of such mistakes being systemic in character may be increased in an institutional context that allows the opinion of any one authority or majority to be enforced by law (Friedman, Reference Friedman2006).

The classical liberal perspective exemplified in the writings of the Virginia school of public choice has also deployed the RPE framework to question the conclusions inferred by neo-classical theories that focus on incentive compatibility issues in their analysis of market imperfections. According to Stiglitz (Reference Stiglitz1994) and his followers, the existence of information asymmetries, principal agent problems, externalities and positive transaction costs in real-world markets generates outcomes that could be improved by governments designing and enforcing appropriate rules and regulations. From the Virginia school perspective, however, comparative institutions analysis suggests that these very problems may be magnified in the real world of democratic politics and bureaucratic administration. Though principal agent problems and information asymmetries are pervasive in markets, where consumers or stock-holders retain the right to ‘exit’ individually from relationships with suppliers or managers, they have incentives to acquire more information about possible alternatives because their personal decision to exit is decisive in determining which service they will receive. By contrast, when individual voters cannot, save for leaving the relevant country, exit from relationships with politicians and regulators, they are forced into a series of collective action problems where decisions to acquire more or less information have only an infinitesimal chance of affecting their personal outcome (Boettke et al., Reference Boettke, Coyne and Leeson2007).

3. Elinor Ostrom, robust political economy and common-pool management

It should be emphasised at this juncture that though the RPE framework has been used to build a case for the superiority of predominantly private-enterprise systems, there is no necessary connection between this form of analysis and these particular normative conclusions. The way is open for both theoretical and empirical analyses that can explain how alternative institutions might cope more effectively with bounded rationality and incentive compatibility in specific problem situations. It is within this context that Elinor Ostrom's work should be viewed. Ostrom herself does not use the term ‘robust political economy’ but in Governing the Commons and more recent publications what she describes as the ‘institutional analysis and development framework’ incorporates key elements from the RPE perspective to argue for the superiority of governance structures that transcend the private enterprise versus state ownership dichotomy in the context of common-pool resource management.Footnote ²

Prior to Ostrom, most work on common-pool resources had assumed that all such assets suffered from the same dynamic, reflecting a basic incentive compatibility constraint. In the absence of exclusion devices enabling people to profit personally from resource conservation, it was argued that ‘free riding’ behaviour would predominate and thus lead to resource depletion. The only way to internalise externalities and to avoid the ‘tragedy of the commons’ would be to have an external body impose a management structure on the resource (Hardin, Reference Hardin1968). This could take the form of either private ownership where exclusive rights to extract fish, timber, water etc. are assigned to individual owners or it could involve the government taking responsibility for the asset either by direct ownership or by the imposition of rules and regulations on private actors. The debate over the relative effectiveness of these alternatives then proceeded along familiar lines. Some argued that in a context of positive transaction costs, private ownership was inefficient in comparison to state-centric alternatives (Ehrenfield, Reference Ehrenfield1972). Others, however, using RPE reasoning maintained that while positive transaction costs were an obstacle to fully efficient markets, a combination of Hayekian knowledge problems and public choice issues of bureaucratic and special interest capture in the state sector provided good reason to believe that private ownership though ‘imperfect’ was the preferable option (for example, Baden and Stroup, Reference Baden and Stroup1981; Deacon and Johnson, Reference Deacon and Johnson1985).

Ostrom's work represents a direct challenge to the contours of this debate. At an empirical level, she demonstrates that it is not always the case that common-pool resources will be overexploited. There are many instances where communities of resource users have developed and enforced effective rules without external regulation and Ostrom explains when such arrangements are likely to arise and the primary obstacles to their formation. Specifically, where there are potential boundaries to a resource, where there are procedures for resolving disputes, and where the community concerned has sufficient decision-making autonomy to create, monitor and enforce its own rules and to exclude outsiders, then the tragedy of the commons can be avoided without external imposition of private or public ownership (Ostrom, Reference Ostrom1990: chapter 3).

Second, Ostrom shows that many solutions that have worked to address common-pool problems defy neat categorisation in terms of state versus market dichotomies. Some of the most successful models of commons management such as those found in the Swiss Alps are ‘mixed regimes’ involving elements of individual property, common property and public regulation. Regimes of this character sometimes involve a specific form of public regulation that Ostrom describes as a ‘facilitator state’ (1990: 137). Such a state does not take over responsibility for managing assets directly but is involved in supplying institutions of dispute resolution that enable local level actors to overcome resource-based conflicts.

Finally, Ostrom uses a combination of bounded rationality and incentive compatibility arguments to suggest that the mixed forms of governance which she describes as ‘common property regimes’ offer a more robust solution to commons problems than centrally imposed versions of private or state ownership.

To understand the specific way that Ostrom applies RPE reasoning, it is important to recognise that most common-pool resources are governed by a set of rules but that the origin of these varies across cases (Ostrom, Reference Ostrom1990: 50–53). In some circumstances, rules are developed and enforced endogenously by the resource users themselves, whereas in others they are imposed exogenously by an external governing body. Ostrom uses an RPE framework to argue for the frequent superiority of endogenous rule formation and to draw attention to the inter-relationships that exist between different layers of rule making.

Though Ostrom has developed this reasoning further than others, it should be pointed out that she was by no means the first to adopt this perspective. Some of the early proponents of what has come to be known as ‘free market environmentalism’ did not in fact favour the top-down ‘imposition’ of ‘privatisation’ but argued instead for a ‘bottom-up’ approach, emphasising the co-evolution of customary and common law procedures in defining property rights (for example, Anderson and Hill, Reference Anderson and Hill1975; Reference Anderson and Hill1983; Anderson and Leal, Reference Anderson and Leal1991; and more recently Hasnas, Reference Hasnas2009). In the desire to separate her analysis from those operating within the markets versus states dichotomy, Ostrom (Reference Ostrom1990: 12) fails to distinguish sufficiently between those private property advocates supporting a centrally imposed form of privatisation from those favouring a removal of centralised obstacles to endogenous rule development that would enable private property and other institutional devices to evolve. This is a significant omission because the ‘free market environmentalists’ were among the first to point out that the external imposition of individual private property rights may prove more costly than the maintenance of common property regimes (Anderson and Hill, Reference Anderson and Hill1983). It should also be emphasised here that Ostrom herself conceives of ‘common property regimes’ as an endogenous form of group-based ‘privatisation’ which may evolve where individual private property may not be practicable. Thus

This clarification aside, Ostrom's support for endogenous solutions to conservation dilemmas can be understood as an application of RPE analysis where free-riding dynamics are combined with knowledge problems involved in discovering the rules best placed to overcome such opportunistic behaviour. It is important, therefore, to elucidate the mechanisms Ostrom suggests may address these issues both to appreciate the strength of her arguments and to highlight the questions she leaves unanswered or fails to answer effectively, when decentralised governance may not be a viable option.

4. Robust political economy and the limits of the Ostromian framework

The analysis thus far has emphasised the RPE arguments in Ostrom's case for reliance on endogenous rule formation to manage common-pool resources. Ostrom offers a powerful rejoinder to those arguing for the imposition of private or public property without recognising the ‘real-world’ constraints that afflict such ‘solutions’. Ostrom was, however, keen to emphasise that her approach does not suggest that decentralised forms of governance always offer the best alternative and was explicit in rejecting what she called the ‘panacea trap’ (Ostrom, Reference Ostrom2012: 69). Thus

What she suggests both in analysing the evidence and in explication of various ‘action situations’ is that there are many contexts involving small- and medium-scale resources where bottom-up rules could deliver more robust institutional arrangements but where their potential has been severely limited by top-down regulation. By implication, however, there is a further category of ‘action situations’ where the scope for endogenous solutions may be limited by objective characteristics of the relevant assets. This has led some theorists to speculate that what Ostrom has provided is a theory of ‘optimal decentralisation’. In other words, that there are some problems best left to decentralised action – and more of these than was previously thought – but that there remain others where more centralised solutions may be required (Aldrich, Reference Aldrich2010). The latter seems most likely to be the case when there are no clear boundaries to a resource and where there are large-scale and mobile populations of resource users leading to a situation of prohibitive transaction costs for endogenous arrangements. Regional and international air pollution and the problem of anthropogenic climate change seem to be the most obvious candidates in this regard.

It is not, however, at all obvious from Ostrom what ‘avoiding the panacea trap’ might imply in such cases. On the one hand, there is certainly no implication that decentralised units are always best placed to engage in resource management decisions. In the context of the debates on metropolitan governance, for example, both Elinor and Vincent Ostrom are at pains to emphasise that for some service functions centralisation may be appropriate. They note in the context of policing services that while neighbourhood policing is best conducted by small-scale units, for larger scale problems such as city wide traffic communications and dealing with criminal networks there may be advantages from ‘contracting up’ services to higher jurisdictions (Ostrom, Reference Ostrom and McGinnis1999; Ostrom and Parks, Reference Ostrom, Parks and McGinnis1999). Similarly, in the context of some larger scale common-pool resources such as the management of river catchments, the most successful governance structures have often arisen in ‘nested arrangements’ where relatively small groups of water users at the level of individual water basins have created associations to manage intra-basin issues but where these have ‘contracted up’ to form higher level associations to address inter-basin dilemmas (Ostrom, Reference Ostrom1990: chapter 5).

On the other hand, however, what Ostrom is referring to in such cases is a process wherein the level of centralisation is itself determined by a decentralised process with lower level units merging to create a superior unit from ‘the bottom up’. In the context of metropolitan governance, for example, she notes:

Vincent Ostrom (Reference Ostrom and McGinnis1999) makes a similar case when suggesting that the advantage of polycentric governance structures where authority is divided between a range of voluntary associations and political jurisdictions arises from the ability of the different actors to ‘contract up’ or ‘contract down’ with other agents and to compare and contrast their performance with that of other jurisdictions.

These arguments appear analogous to the classical liberal/Coasian case for allowing the process of mergers and de-mergers in a competitive market to determine the amount of ‘central planning’ in the economic system in a decentralised way. Coase notes that decentralised exchange systems come with costs that are sometimes outweighed by the benefits of centralised management within large firms, but he was equally eager to emphasise that these efficiency advantages can only be discovered in a wider context of decentralised institutions where organisations of different sizes compete in an ongoing process. He notes:

None of this analysis appears to suggest that there might be circumstances where a more centralised institution imposed ‘from the top down’ and eliminating institutional competition could be trusted with resource management issues. There are few, if any, case studies Ostrom provides where action by centrally imposed regulatory or management institutions has produced positive outcomes. While some of her successful case studies do recognise an important role for higher level authorities, the role of these is typically confined to conflict resolution procedures and in some instances the supply of technical and scientific expertise in helping local actors to define resource boundaries.

From the perspective of RPE, a consistent application of Ostrom's theoretical concepts suggests that avoiding the ‘panacea trap’ implies recognising a further option beyond community management, markets and government – that of ‘doing nothing’. Beyond a certain point of centralisation, the ability to learn via a trial-and-error evolution may be so blunted that the scope for systemic failure may negate any benefits that central regulation might in principle bring. Similarly, the scale of the monitoring problem facing lower level actors, combined with the loss of exit that arises when jurisdictional units exceed a certain scale, may result in massive potential for rent seeking and the abuse of power. Ostrom's own analysis suggests that bounded rationality and incentive compatibility problems under excessively centralised regulation may result in worse outcomes than would otherwise prevail. This may be relevant in circumstances where owing to high transaction costs and the inability to organise regulation ‘from below’, governance structures are imposed ‘from the top down’. Crucially, however, it may also be relevant where decentralised units ‘contract up’ responsibilities to units of such a scale that the relevant decisions, however mistaken, become virtually irreversible. This may be a particular problem where the relevant units are political jurisdictions that can block exit options. Depending on the scope for such error and abuse, it may be that is not worth the risk of creating institutions beyond a certain scale, even if this means that the relevant externality remains unresolved. Yet, this is a possibility that Ostrom herself does not entertain.

In making a case for centralised regulation, it will not suffice to posit the potential seriousness of the consequences from failing to address the underlying issue because it is the possibility that centralised action may magnify these issues that is at stake. Consider anthropogenic climate change. This represents the most pressing common-pool problem with potentially massive implications for global living standards and quality of life. Yet, clear support for regulatory action on a global scale would require the validity of some highly implausible assumptions. The first is that global policy-makers have sufficient knowledge not only to manage climate patterns by appropriate adjustments to CO₂ emissions, but also that they know how to design a set of regulatory tools that will deliver these reductions – and to do so in a tolerably cost-effective manner. The second assumption is that those involved in the process of regulation will not abuse the enormous powers at their disposal through rent seeking either by way of bureaucratic expansionism and/or deal-making with powerful interest groups. Arguably, the most serious common-pool problem of all is that involved in holding to account centralised authority and nowhere would this be more problematic than in the case of global-level institutions, however democratic in character. Given the infinitesimal chance of any voter affecting the decisions of such an authority, there would be few incentives to monitor and even less capacity to affect its performance. Absent omniscience and full motivational compliance, then the possibility that the costs of climate change regulation will exceed any benefits must be very real.Footnote ⁴

The latter point is not to imply that there is never any possibility of more centralised actions delivering beneficial results. The Montreal Protocol, which has witnessed significant reductions in the use of chlorofluorocarbons (CFCs), for example, has been widely credited with halting the destruction of the ozone layer. This particular arrangement arose via multi-lateral bargaining between nation states that agreed to impose controls on CFC use within their borders. Success here may well have been due to the low costs involved for the millions of affected parties in switching to what were cheap and widely known substitutes that made the agreement effectively self-enforcing. In the context of climate change, however, massive uncertainties about the costs of behavioural change, combined with similar uncertainties about how regulatory structures might operate, raise the possibility that simply adapting to whatever climate changes may occur could be an equally if not more viable option. One cannot draw an unequivocal conclusion on this front, but at the very least there appear no robust theoretical grounds to believe that the risks from ‘doing nothing’ are likely to be any worse than those resulting from imposing a misguided cure.

Ostrom (Reference Ostrom2009) comes close to endorsing the view expressed above in her own account of the climate change question but, in the final analysis, is unwilling to endorse the ‘do-nothing’ option. She argues that while the global nature of the climate change problem may appear to suggest that global regulation is the only viable option, this is not so. On the one hand, Ostrom contends that in a context requiring immediate action to start the process of reducing emissions, it may take too long to create the necessary regulatory structures at the global level and to build institutional support for them (Ostrom, Reference Ostrom2009: 28). The more attention given to the global dimension of the problem the more local and national level actors will feel the irrelevance of their particular contribution to any solution – and this may increase the propensity for ‘free-riding’ on the necessary institution building. On the other hand, even if global regulatory structures can be put in place, there is a high probability that there will be mistakes in designing the regulations and this may result in systemic failures on an enormous scale. The regulations themselves are unlikely to be self-enforcing and will incur massive transaction and monitoring costs that will limit their effectiveness in the context of overwhelming incentives to free-ride or to engage in rent-seeking behaviour within the regulatory structure (Ostrom, Reference Ostrom2009: 31).

From Ostrom's point of view, while the very nature of the climate change problem precludes a fully decentralised approach, she maintains that though imperfect, polycentric alternatives emphasising not the global benefits of adopting emissions reducing measures but those benefiting actors at the local, regional and national levels are the ones more likely to be successfully enacted and enforced. Within this context, Ostrom notes that at many different scales of governance, measures are already being enacted that may contribute solutions to the climate change dilemma without waiting for global action. At the local scale, these include anti-pollution measures to improve air quality and local health outcomes, congestion charging schemes for road use which reduce journey time and contribute to lower emissions, and at a higher scale measures such as the European Union's carbon-trading regime (Ostrom, Reference Ostrom2009: 16–22).

From an RPE perspective, however, beyond the relatively localised measures such as road pricing, the levels of decentralisation Ostrom endorses in this case may themselves be too centralised to avoid large-scale errors resulting from bounded rationality and incentive compatibility constraints. When the units that are ‘contracting up’ responsibilities are political jurisdictions and states that can wield coercive powers in a manner that blocks effective exit options, then there are grounds to be very wary of such ‘solutions’.

One example here is the European Union emissions trading system. As a European wide scheme, this provides little or no scope for learning via inter-jurisdictional competition and presents a massive principal–agent problem for the hundreds of millions of citizens and energy consumers to monitor its design and enforcement. While it is too early to come to judgement on the eventual performance of the programme, initial results are not encouraging. Research reported in Helm (Reference Helm2010), for example, suggests that the European Union scheme has produced higher prices for European consumers, no reduction in CO₂ emissions but significant anti-competitive benefits to incumbent energy firms. If such results were to continue, then it would not be obvious why taking measures to reduce CO₂ emissions at the European Union level should be thought preferable either to leaving matters entirely at the discretion of individual nation states or indeed of not engaging in any kind of climate change policy beyond those measures that might deliver purely local level environmental goods.

What is surprising about Ostrom's reluctance to consider ‘doing nothing’ and her support for measures such as European emissions trading is that her own framework offers little to suggest that state-driven schemes beyond those operating at a localised level will be robust. Ostrom herself is scathing about some of the European Union's other environmental policies – notably the Common Fisheries Policy (Ostrom, Reference Ostrom2012). Yet, this policy could be considered as a classic case where a mistaken decision to ‘contract up’ responsibilities to a higher level has placed both fishermen and the European Union citizenry in a scenario where they are unable to escape from what looks like a virtually irreversible decision. The case of fisheries is, of course, more telling from Ostrom's standpoint because it is an area that her own research suggests is open to more decentralised forms of governance. Nonetheless, what the case shows is that there are risks from centralisation whenever those involved in the process of institution building are political actors that can ‘lock’ their citizens into failing institutional structures – an analysis that is of relevance even when there are no decentralised alternatives.

Insofar as centralised state action has demonstrated benefits, Ostrom's research tends to confine this to conflict resolution measures, offering the background rule of law that can stabilise expectations for local actors as they craft their own solutions to collective action problems. A further ambiguity in the Ostromian framework here is the absence of principles to explicate what can prevent states that perform this necessary role from becoming a threat to predominantly bottom-up processes. There is, for example, little to suggest in Ostrom's work that democratic states are more likely to maintain the conditions supporting endogenous rule formation than non-democratic alternatives (Mansbridge, Reference Mansbridge2010: 591). Institutions of majority rule offer no guarantees that regional or national majorities will not vote for measures that impinge on the property regimes developed by individuals and communities at a more local scale. What matters in Ostrom's framework is maintaining an appropriate division of responsibility between different levels of decision-making – yet successful and unsuccessful attempts to maintain such division have occurred in a variety of democratic and non-democratic contexts. It seems, therefore, to be a matter of historical accident whether the surrounding state institutions work to support Ostromian structures or to undermine them.

Ostrom is, of course, hardly alone in lacking a fully developed theory of how to create and sustain the framework needed to support effective bottom-up institutions. Classical liberal political economy, for example, has a well-developed explanation of the potential robustness of private property, market-based institutions, as sketched in the opening section of this paper. What it lacks, however, is an account of how to create these institutions where they do not exist and where they do, how to prevent those who achieve entrepreneurial success from using state powers that may be necessary for law enforcement and the supply of collective goods to entrench their own position via anti-competitive rent seeking.

In attempting to address these difficulties, two potential avenues consistent with the RPE perspective suggest themselves. The first might be termed the ‘anarchist solution’. On this account, since there is no convincing explanation of how central state institutions can overcome bounded rationality and incentive compatibility in the provision of large-scale collective goods, attention should shift to avoiding all forms of such control. Working in a developing countries context, for example, Leeson (Reference Leeson2007) suggests that when central state institutions are vehicles for purely predatory purposes, developmental prospects may actually be enhanced by not having an apparatus of government in place. This conclusion reinforces that expounded by James Scott who suggests that for many in the developing world ‘escaping from government’ might be preferable to living under the auspices of an incompetent and/or predatory regime (Scott, Reference Scott2009).

A second avenue of response might be termed the ‘conservative solution’. According to this view, in contexts where people live under tolerably functioning state institutions – albeit owing to historical accidents – it would constitute an act of extreme rationalism to dismantle these institutions in the absence of any functioning alternative. Although contemporary liberal market democracies may be riddled with rent seeking and cronyism of various sorts, they function well enough to have produced reasonably high levels of economic growth, combined with some protection for common-pool resources. This perspective might point towards an incremental removal of regulatory controls in areas where bottom-up rules have the potential to evolve. At the same time, however, it would be wary of creating new national or international structures of collective action that may generate systemic risks – even in the absence of bottom-up alternatives.

Both of the above perspectives seem consistent with an RPE approach that recognises the significance of ‘non-ideal’ conditions. Ostrom has provided a compelling explanation of common-pool problems and of the institutional processes that may be best placed to address them. What she offers, however, is nonetheless a partial solution. Until it can be explained how the dangers that arise whenever large-scale institutions of collective action are put in place can be minimised, then one of the most important, though probably unwelcome, lessons to take from Ostrom's work is that ‘no solution’ is sometimes the best available option.