Skip over navigation

Journal Issue: Opportunity in America Volume 16 Number 2 Fall 2006

Intergenerational Social Mobility: The United States in Comparative Perspective
Emily Beller Michael Hout

Measuring Intergenerational Social Mobility

Important differences in the concepts of occupational and income mobility can help to explain how it is possible that mobility in one domain might be greater than mobility in another. People's incomes vary significantly even if their jobs share the same occupational category. Analyses of occupational mobility and analyses of income mobility provide different pictures of people's prospects, because they ask different questions. Intergenerational persistence in occupational status is not a good proxy for persistence in income, and vice versa; a person who is upwardly mobile occupationally does not necessarily enjoy a higher relative income than his or her parents (and vice versa).5

In addition, analysts investigating occupational and income mobility face different limitations and use different methodologies. On the one hand, occupation is easier to measure than income because people remember their parents' occupations reliably and with a high degree of accuracy, whereas dollar amounts are much harder to recall, and most people plainly do not know their parents' incomes. Inflation erodes the value of the dollar over time, too, further complicating the task of evaluating parents' incomes, even if they are known. On the other hand, occupations can be hard to rank, whereas income is straightforwardly scored in dollars (or the relevant local currency). In addition, researchers interested in occupational mobility often want to measure the component of mobility that is independent of growth, whereas income mobility researchers do not typically distinguish between the two.

Researchers interested in occupational mobility must first come to grips with the problem of how to rank occupations, getting beyond the qualitative detail of specific job descriptions to arrive at useful categories or scores. Some solve the problem by grouping occupations into relatively large classes. Others rank them on a scale from 0 to 100.6 In the first approach, researchers gather occupations into several broad classes, such as professionals (for example, doctors and lawyers), skilled trade workers (for example, electricians and carpenters), or the self-employed, and then create a matrix that allows them to compare each person's occupation with his or her father's occupation. While this approach reveals details of which occupations are linked across the generations and which are not, its results are hard to summarize unless the categories are clearly ranked.

Ranking allows the straightforward estimation of an overall intergenerational correlation between the ranking of a person's occupation and that of his or her father. A correlation of 0 implies that a person's occupational rank is completely independent of that of his or her parents, and therefore that there is perfect mobility between ranks across generations. A correlation of 1 implies that ranks do not change from generation to generation. The correlation that a researcher calculates for a real society places that society on the continuum from perfect mobility to complete rigidity.

In principle, one could use an intergenerational income correlation to measure income mobility as well as occupational mobility, but in practice researchers (usually economists) typically measure income mobility slightly differently. They look at the strength (persistence) of the relationship between parents' and children's income in percentage terms; that is, they ask how much (what percentage) of the income difference between families in one generation persists into the next generation. This estimate is called the intergenerational elasticity. If the elasticity is 0.4, for example, they would conclude that a 10 percent difference in parents' income would lead to a 4 percent difference in offspring's incomes. The advantage of using the intergenerational elasticity, from the researcher's point of view, is that it can capture the amplifying effects on mobility of rising income inequality, or the dampening effects of falling income inequality (the formula for the intergenerational correlation discards this useful information). On the low extreme, an elasticity of 0 describes a society in which family economic background is not at all related to adult income, whereas an elasticity of 1 describes a society in which each person ends up in exactly the same economic position as her or his parents (just like the correlation). But unlike the correlation, the elasticity is unbounded, so one could, in principle, discover that two people who started life in families 10 percent apart ended up 15 percent apart (if the elasticity was 1.5). Mobility is the complement of the elasticity— a low intergenerational elasticity translates to a high mobility rate, and a high elasticity translates to a low mobility rate.