Fairness and efficiency in university admission

My recent paper on the non-effect of fee abolition in Ireland emphasized the socio-economic gradient in secondary school attainment: low SES means low points and therefore a low probability of progressing to university. Fees are a side-issue, at best.
This problem is not unique to Ireland, of course: exactly the same pattern was described in a recent study for the UK. A recent book published by the Brookings Institution suggests ways of dealing with this issue in the US where there is a pronounced SES gradient in the SAT tests. One suggestion is for expanded use of affirmative action programs beyond their traditional emphasis on race. In essence, the idea is that universities take into account the fact that for some people doing well is harder and not because of their innate ability.
Access Programs here, like UCD's New ERA, do this although the scale is necessarily small, see here for an evaluation of its effectiveness we did at the Geary Institute. Some people might take the view that moving away from our simple points-based CAO application system would be messy and expensive. But, on the other hand, the present system is hardly efficient from a national point of view given that it means a huge chunk of our potential labour force is not getting the education that it needs to be at its most productive. Relative to the huge losses arising from not educating our population adequately (for evidence see here), the administrative costs are peanuts.
My own view is that making such adjustments at the point of progression i.e. adjusting the admission criteria to take account of SES, while worth pursuing, is not the best solution. We need to address the causes, not the symptoms of the problem. That is, we need to address the reasons for under-achievement at school level. Access programs do this too as do some government initiatives although their effects are largely unknown.