Two of my posts earlier this month (here and here) focused on the institutional implications of limited rationality. My claim was that robust institutions are those that minimise the consequence of inevitable human errors. In a nutshell, this is the Hayekian argument for the competitive ‘exit’ principle as facilitating a process of evolutionary learning under ‘bounded rationality’. Though this insight has often been recognised in economics (though not as much as one might hope), relatively little has been made of what bounded rationality might imply for questions of distributive or social justice. This is surprising for no less a figure than John Rawls recognised that when choosing distributive rules from behind the ‘veil of ignorance’ people should have access to ‘basic facts of social theory’ to enable them to engage in an impartial choice of just institutions. Impartiality for Rawlsians, of course, equates to support for a ‘basic structure of society’ that institutionalises the ‘difference principle’ (maximise the position of the worst off) and a distributive branch of government – i.e. a welfare state, that distributes resources accordingly.

If the ‘basic facts of social theory’ include recognition of limited human rationality and the need for institutions that promote learning, however, then it is worth reconsidering what the requirements for an impartial choice of social institutions might be. More specifically, if the choice of social institutions takes place under bounded rationality or genuine ignorance (save for ‘basic facts of social theory’), then it seems highly unlikely that people would opt with certainty for any particular distributive rule, such as Rawls’ difference principle. If people are genuinely ignorant, not only of their talents and social position, but also about the values and beliefs they will possess, their attitudes to risk and how these are likely to change in the light of experience and mistakes then it seems more plausible to suggest that they might opt for a ‘basic structure’ that would enable them to learn about, and to choose between competing distributive standards – i.e. one which allows for exit.

As well as ‘rigging’ the result of his own particular thought experiment by positing that actors behind the veil are risk averse in the extreme, Rawls rules out the possibility of people learning from the experiences of others by specifying that actors behind the veil know themselves to be choosing rules of distribution for a ‘closed society’ where exit occurs only by way of death. The justification for this move follows from Rawls’s assertion that given the option potentially selfish actors may threaten exit in order to exert bargaining power against others to secure ‘unfair’ terms of cooperation. People should, in other words, be thought of as choosing rules of distribution in conditions where their social position might be determined by their enemies. This was always rather an odd move on Rawls’ part since anyone thinking about how to design institutions when they might be assigned a place by their enemies might allow for an exit option precisely so they might escape the possibility of exploitation by opportunistic actors.

Even if one were to grant the benefit of doubt to Rawls in this instance, however, it would not be enough to save his theory from the ‘basic fact’ of bounded rationality. On the contrary, in the unlikely event that people do actually agree on a unitary distributive rule such as the difference principle they would have to reintroduce exit options internally in order to have any chance of achieving it. When people must make distributive choices under conditions of limited rationality then it cannot be assumed that knowledge of how to achieve the difference principle or any other ‘patterned principle’ of justice can be ‘given’ to the administrative arm of government.

Does redistribution raise the absolute standard of living for the poor? Should redistribution be conditional on behavioural changes and should it be supplied via cash payments or the provision of ‘free’ services such as education? Is a person best placed to help the disadvantaged by starting a new enterprise and employing poorer sections of the population? Would it be best for someone to take a high paying job and contribute part of their income to charity? If charitable activity is indeed an appropriate way to help the disadvantaged should this take the form of monetary contributions or spending time directly with the less well off? Rather than assuming that answers to these questions can be known by any single actor or group, the ‘basic structure of society’ must facilitate the discovery and communication of ‘who should give what to whom’ – i.e. it should allow for people to associate with one another on a range of different terms.

In short, without the possibility of exit and the learning opportunities it affords there is no hope of achieving justice understood in Rawlsian terms or anybody else’s. Moreover, in a world where ‘differences’, i.e. ‘inequalities’ in results between individuals and jurisdictions are needed to have any chance of learning what the ‘difference principle’ requires, the observed pattern of distribution at any point in time may be indistinguishable from a scenario in which there was no attempt to institutionalise any particular distributive rule in the first place.

Mark Pennington is the author of Robust Political Economy: Classical Liberalism and the Future of Public Policy.