Beating the odds (2): a new index of intergenerational social mobility

The twin starting points for this paper are recent discussions in economic theory concerning the normative foundations for indices of intergenerational mobility, and the development of a human capital score intended (in conjunction with a parallel measure of individual wealth) to act as an indicator of social class. The paper specifies an intergenerational Gini-type mobility index, based on the mean expected earnings of the children of fathers in each class. It makes a series of estimates of mobility in Britain from BHPS lifetime employment history files. The first of these uses the Goldthorpe threecategory class schema for both origins and destinations; the resulting estimates suffer from (1) the systematic exclusion of those (particularly women) outside employment, and (2) the effects of the changes in the sizes of the class categories. The second estimation is a hybrid using Goldthorpe (seven category) origins and human capital quintile destination categories; this deals with problem (1) but not problem (2), The third estimation uses human capital quintiles for both origin and destination categories, and avoids both problems. But it is dominated by the fourth approach, a continuous version of the Gini index using scalar variables (the fathers’ human capital scores and the children’s expected earnings). The women’s intergenerational Gini score now shows substantial increases from the 1935-44 birth cohort to the 1955-64 birth cohort; the men’s score shows small increases over the same period. The conclusion is that in terms of this index, intergenerational mobility in Britain has become more unequal overall. This paper is part of the “Social Position and Life Chances” (SPLC) project which aims to formulate and estimate new measures of social class appropriate for understanding life chances in modern Britain. An initial outline of the project is set out in ISER Working Paper 2001—20; the human capital measure used here is discussed in ISER Working Paper 2002—2, and the associated measure of wealth developed for the project, and its relationship to human capital through the life-course, is discussed in ISER Working Paper 2002—16. Regression modelling approaches to the intergenerational mobility of human capital are discussed in ISER Working Paper 2002—17. The SPLC project is part of the Research Programme of the ESRC Research Centre on Micro-social Change. NON-TECHNICAL SUMMARY Intergenerational social mobility is normally considered by British sociologists as a matter simply of transition between parents’ categorical positions and those of their children. But it is now becoming clear that it may be legitimate and indeed appropriate to consider in addition, any changes that may have taken place in the conditions of life of those in each of the positions over the period. There are in effect two alternative general normative bases for the construction of indicators of social mobility. We may take the traditional view that it is sufficient just to consider the transition probabilities. Or we may consider that an adequate view of mobility requires in addition a consideration of the “rewards” attached to the positions. The first task, in this paper, is to construct a measure of intergenerational mobility that considers both transition probabilities and rewards. A very simple solution to this problem is an “intergenerational Gini score”. There are essentially two steps. First, calculate for each parental class, the appropriate “children’s opportunity set”, which is the sum of the products of the transition probabilities from the parental class to each of the destination classes, and the “expected reward” (eg expected annual income) of a member of each destination class. Second, calculate a categorical Gini score, ordering the parents’ classes according to the size of their respective childrens’ opportunity sets. The larger the resulting Gini score, the more unequal are the life chances of the children of the various categorical classes. Estimating mobility in the UK in these terms using the Goldthorpe Class categories shows little evidence of change in intergenerational mobility patterns—ie life chances—for men or women in successive birth cohorts in mid-century Britain. This analysis however suffers from the exclusion of women without recent employment experience from the class schema, and from confounding effects of historical changes in the relative sizes of the class categories. As an alternative to the class categories, we can use a continuous human capital measure intended (alongside a parallel measure of individual wealth) to fulfil some of the same functions. The human capital score has been estimated for each year of the adult life of all members of the British Household Panel Study (using the retrospective work history data file). It is possible also (though with some reservations) to estimate the human capital of the BHPS respondents’ parents when those respondents were aged about 14. If, rather than class categories, human capital quintiles are used for the mobility analysis, we avoid both of the previously noted problems. We find, in this case, a rise in the women’s intergenerational Gini coefficients for the most recent birth cohorts. Finally we move to the more satisfactory continuous version of the intergenerational Gini index. Individuals’ expected (labour) incomes (at ages 34-36) are ordered by their fathers’ human capital scores. On this basis the women’s Gini indices show a substantial increase from the 1935-44 birth cohort to the 1955-64, probably reflecting inequality of access to the labour market of women from different social backgrounds; the men’s shows a small increase. Britain’s intergenerational mobility patterns have, overall, been becoming more unequal over this period.