Consider three cases. In case 1, the state requires that citizens shun anyone who is found to be engaging in activity x. Those who do not shun people who are found x-ing are themselves to be shunned. There are enough people who, for whatever reason, obey this requirement that it results in the reliable shunning of anyone who is found x-ing or who does not shun those who are found x-ing. In case 2, prominent members of the community who exercise a wide degree of authority over it require that citizens shun anyone who is found x-ing. Again, those who do not shun when required to do so are themelves shunned, and, also again, there are enough people who are prepaped, for whatever reason, to obey this command that those who x are shunned and those who do not shun those who x are themselves shunned. In case 3, everyone has decided, without being encouraged either by the state or other authority figures, to shun anyone who is found x-ing, and to shun anyone who does not shun those who x.
In all three cases, then, the effects on those who x and those who refuse to shun those who x are identical: whether or not the coercion in each case is the same, those who x and those who refuse to shun those who x are surely similarly and probably equally coerced. There is a plausible case that all three cases are matters of similar concern, if not necessarily similar remedial action, for theories of justice. If x were a particular religious or sexual practice, likely to be of non-neglible following and of some significance to its follwers, we could reasonably describe all three as falling within what Rawls calls the basic structure, and considers the proper subject of social justice: whether or not all three sets of practices would count as major institutions, all three "define [individual's] rights and duties and influence their life-prospects, what they can expect to be and how well they can expect to do".
What seems to place all three cases in the basic structure is that, when there is a significant interest in x-ing, the social practices that they all involve themselves involve quite serious punishment of those who x or 'collaborate' in x-ing. Indeed, the institutions which Rawls thought of as part of the basic structure all seem to have this feature, since he clearly intends the basic structure to capture a state's coercive legal structures qua sets of coercive rules within which a life has to be lived. Ignoring cases 2 and 3, despite them being identical in their effects to case 1, in a pluralistic society seems reasonable enough, since the chances of them ever happening, pace Mill and Tocqueville, are fairly minimal, as well as, I think, considerably more morally complicated.
Consider another case, though, raised by those who want to stretch or otherwise redefine the Rawlsian concern with the basic structure. Case 4 is more complicated. In case 4, it has been noticed that lots of people x-ing has effects which disadvantage other people, and so everyone is encouraged, through various formal and informal social institutions, to not x. Clearly, in case 4, the norm of not x-ing is part of the basic structure, assuming people have a significant interest in being able to x. Consider, further, though, yet another case. In case 5, although lots of people x-ing does disadvantage some other people, either this has not been noticed or, if it has, no decision to encourage people to avoid x-ing has been taken. The norm of it being acceptable to x in case 5 does not seem to be part of the basic structure. No-one who does not x suffers as a result of not x-ing in the way that someone who did not respect property rights or refused to sign any contracts in our society would suffer.
If that is the case, though, the egalitarian ethos that G.A. Cohen has argued for, which would require that people did not self-interest maximise in the economic marketplace - on the grounds that those who happen to have socially useful skills doing so reduces the take for those who happen not to have such skills, which, if you believe in economic equality and some kind of Pareto condition, is bad - cannot be required as a matter of justice, since its absence is not part of the basic structure. Case 5 is, after all, a case in which doing x, where large numbers of people doing x is disadvantageous to some other group, is permitted, and in case 5, the absence of such an ethos is not part of the basic structure as I have suggested it ought to be defined.