A social policy that worked? Learning lessons from the past.

It's easy to get cynical about social policy interventions that are meant to level the social playing field or temper the worst extremes of the raw market outcome. Too often we are told that what trickles down to the truly disadvantaged is a small fraction of what was actually intended and that the savvy middle-classes scoop up the lion's share of the benefits. More university places? Great say the not so bright scions of the stockbroker belt who, despite an expensive private school education, only manage a mediocre set of A level grades. Let the 3 year party begin! Well, maybe. But not all the advantages of reform are necessarily captured by the middle-classes. It depends on how the reforms are implemented.

Consider the case of England and Wales in the early Twentieth Century. Then the field of battle was not access to tertiary education but access to secondary education. Secondary education was almost entirely in private (though mostly not for profit) hands and supply was severely limited. Secondary schools were not however averse to making a deal with the State and many agreed in return for significant grants from the public purse to make 25% or more of their places free to children who had attended state elementary schools. Of course some of these free places were captured by the canny middle classes who sent their kids to good local elementary schools and then crammed them for the scholarship tests. But not all, and here is the evidence (click on the image for a larger picture):

Figure 1. Trends in odds-ratios and relative-risks under three models for the proportion with „secondary“ education. Men with father’s occupational score  ± 1 standard deviation from the mean. England and Wales, 1949.

The key to interpreting this graph is that each line involves a contrast between men from different social backgrounds. I've chosen to display this by comparing men whose fathers had "occupational status" scores +/- 1 standard deviation around the mean. Purists will  complain that I have broken one of my own golden rules by not giving any indication on the graph of the absolute level of "risk" (in this case the probability of achieving a secondary school education) but it suffices to know that roughly 80% of the population had no secondary schooling. 

Whether one looks at relative risk or odds ratios it is clear that inequality of access with regard to father's status declined. What is really telling is the comparison of the dotted and dashed counterfactual relative-risk line and the solid black "trend in free places" relative risk line. The first of these shows what would have happened if the odds ratios describing the association between social background and secondary schooling would have remained  at the the same levels as those observed for the birth-cohort 1880-89 ie men whose schooling mostly took place in the last decades of the Nineteenth Century. Inequality (as measured by relative risk) would still have declined but it declined even more than it otherwise would have because the association between social background and schooling  also declined.

A nice feature of the data I'm using here (David Glass's 1949 Social Mobility Survey) is that it allows you to have two views of the same time period. The men who appear in the data as sons in Figure 1 appear as fathers in Figure 2 which displays the trends for their children (in this case boys though the pattern for girls is essentially identical).

Figure 2. Trends in proportions (with 95% confidence intervals), odds-ratios and relative-risks for the proportion with „secondary“ education. Men with father’s occupational score  ± 1 standard deviation from the mean. England and Wales, 1949.

 Figure 2 tells essentially the same story as Figure 1 with the sole difference being that the level of the estimated proportions differs slightly because father's "occupational status" is measured in a slightly different way in the two graphs.

So what is new? Actually not much. I can't claim to be the first to notice this, in fact Anthony Heath and Peter Clifford pointed it out quite a long time ago in one of their JRSS articles. All I can claim is that after digging around there is some evidence that favours an interpretation of the trend towards greater social equality in terms of the operation of the free-places scheme. Firstly the differences in the odds-ratios between birth cohorts are more or less linear in the proportion of the cohort getting a free place. In itself this isn't terribly strong evidence as lots of other things may well have been changing in the same direction. More convincing is that if you break secondary schooling down into free-places and non-free places, then the trend completely disappears. In other words the trend is only there when you aggregate over paying or not paying for your education and is in fact a compositional effect. The proportion of people who got a secondary education increased and this was mainly driven by an increase in the proportion with free places. People with free places tended to come disproportionately from the less well-off sectors of society.

So here is one social reform that, at least partly, worked as it was intended to work. I wonder if there are any lessons here for current policies about access to higher education? If you want to avoid the middle-classes grabbing the benefits of reform you have to put the extra resources into places they are unlikely to find appealing (the equivalent of the state elementary schools).  So, how about  reserving 90% of places in universities for children who attended state secondary schools. Smacks too much of social engineering? OK, if we are going to have higher tuition fees why not take a levy on all those who enter state subsidized higher education after attending a private secondary school. We already charge non-EU students higher tuition fees. Why not charge the privately educated  more than the cost of their tuition and use the surplus to make sure that talented kids from poor backgrounds are not excluded from the best universities? That would pose an interesting decision problem for middle-class parents. They realised a long time ago what, under the present system configuration, an incredible bargain private-schooling can be when compared to the cost of buying a house in the catchment area of a good state comprehensive.