What Do We Know and What Should We Do About Social Mobility?. Lee Elliot Major
of mothers with labour earnings in many data sources. This has changed during recent decades as more females have joined the labour force, and richer data have become available. There is a separate literature focusing on many important issues of gender inequality we do not cover. However, where possible we report mobility trends and relevant statistics for women as well as men.
Similarly, the literature has for the most part suffered from a paucity of data tracking the outcomes for people with different ethnic backgrounds. An important emerging research area is gauging intersectional impacts – assessing outcomes for people categorised by status, gender and ethnicity – but the studies are few and far between. Wherever possible we highlight findings for the UK as a whole, but sometimes (especially on school education) we have to make do with data covering just England. Many of the most relevant large-scale data studies also come from the United States, but we will refer to international studies where relevant as well.
Intragenerational mobility, as opposed to intergenerational mobility, refers to movement between income or class positions during a person's own lifetime. Multigenerational mobility refers to transitions over not one, but multiple generations. Social mobility can be short range and long range, featuring a nudge along the income spectrum, or a full rags-to-riches leap. Average social mobility rates conceal large variations for those at the top and bottom of society. Considering the whole distribution as well as the nature of shifts at different points is critical for the study of intergenerational mobility patterns.
Social mobility can be measured in absolute or relative terms. For relative measures, if one person goes up, another one goes down. On the other hand, absolute mobility rates show the percentage of people whose income or class destinations improve or worsen compared with their income or class origins. This can be upward and downward in direction. An example of upward absolute intergenerational mobility would be earning more in real terms than your parents. Alternatively, it occurs when your social class is higher up the class structure. Relative mobility rates on the other hand describe the relative chances of people from different backgrounds moving up or down the income or social ladder. Sociologists sometimes call this social fluidity.
In education debates, social mobility has been used as a more generic term for improving the results of pupils from poorer backgrounds – tied to efforts to close achievement gaps between disadvantaged children and their more privileged peers. The guiding principle is that children should fulfil their potential irrespective of their background. However, as we do not know the status of students’ parents or outcomes of students as adults, these are only indicative measures of social mobility, based on the assumption that the individual benefits from education in the past will continue for future generations.
Social Mobility Research
To our knowledge, the term social mobility was first coined in the early twentieth century. The Russian–American sociologist Pitrim Sorokin was himself a story of upward mobility – a trait of many scholars in this field. In 1922, he fled from Russia, evading capture by Lenin's forces as the Bolsheviks consolidated power after the Russian Revolution. Sorokin emigrated to the United States where he founded Harvard University's sociology department.
Sorokin was a keen political activist and was interested in the stratification of societies into lower and upper classes defined by their wealth and power. Tracking the backgrounds of people entering various elites, Sorokin studied what he called vertical social mobility – ‘any transition of an individual, social object or value…from one social position to another’ (Sorokin, 1927). Vertical mobility could occur when individuals leaped from one class to another. On the other hand, it could describe the movement of whole groups or people closer or further apart.
Over the last century, the field has gone from data poor to data rich. The golden era of British sociological studies after the Second World War was founded on nationally representative cohort surveys offering rich details of parents, families and children. However, even these studies seem small scale compared with the big data era of the early twenty-first century.
Chetty et al. (2014b) used the tax records of some 40 million Americans to reveal a detailed map of upward mobility levels for children born between 1980 and 1982 in different cities, counties and states across the United States. This landmark work based on extremely rich data shows how far social mobility research has advanced.
Data are the life blood of social science. The UK has been blessed with national cohort studies enabling researchers to gain numerous insights into the changing life experiences and outcomes of successive generations. The 1970 British Cohort Study (BCS) follows the lives of 17,000 people born in England, Scotland and Wales in a single week of 1970. Other birth cohort studies include the 1946 Medical Research Council National Survey of Health and Development (NSHD), the 1958 National Child Development Study (NCDS) and the 2001 Millennium Cohort Study (MCS).
Without the findings from these studies, we would live in a less enlightened world, less equipped to face future societal challenges. It is criminal that social mobility research in the UK has itself suffered its own dark ages. We have no nationally representative datasets of similar quality tracking generations born in the 1980s and 1990s. The studies were victims of cuts to social science research under Margaret Thatcher's government. Thatcher's children are the lost generation at least in research terms. It is a tragedy that no British cohort study has been commissioned since the Millennium. Future researchers will be left guessing about the generations born in the 2010s (and perhaps 2020s), and will have to use different types of analysis and data to try to fill the gap.
Different Dimensions
Studies of intergenerational persistence have become increasingly multi-dimensional. Generational persistence is observed not only for earnings and occupational class but also for many other attributes as well: from wealth to health, education to happiness, crime, consumption and even divorce.
The extent of intergenerational persistence (or immobility) can be measured using various statistical methods. Table 1.1 summarises statistical measures used in the literature. The first four measures described in the upper panel are relative measures: a correlation coefficient of a particular measure of economic or social status between generations; the intergenerational elasticity (IGE) from statistical regression methods; a rank correlation which compares rank in the respective distributions from one generation to the next; and transitions between different parts of the economic or social status distribution of family members across generations.
Table 1.1
For the first three correlational measures, a value of 1 corresponds to complete immobility, with parents and offspring outcomes perfectly correlated. A correlation of 0 corresponds to complete mobility, with no relationship between family background and the adult outcomes of children. For transitions, the example given in the table splits parental and child measures into five equally sized groups – quintiles running from the bottom 20 per cent to the top 20 per cent. In this case, complete mobility corresponds to children growing up in any parental quintile having a 20 per cent chance of ending up in any of the five quintiles in their own generation. Complete immobility is where everyone stays in the same quintile as their parents. There is no movement and everyone remains on the diagonal of the five-by-five transition matrix.
Absolute mobility is about how well all children of a generation do compared with their parents. The metric of absolute mobility shown in the lower panel of Table 1.1 is therefore the fraction of children who do better for a given measure of economic or social status than their parents did in an earlier time. For example, it could be the percentage of children whose labour market earnings or family income are higher in real terms than their parents were at the same age. It could be the percentage of children with higher education levels or social class than their parents
It is possible, in theory, to have different patterns of absolute and relative mobility (Berman, 2019; Bukodi and Goldthorpe, 2016; Nybom, 2018). There could