Social mobility in pre-industrial times has been argued to be much lower than often estimated and that it hardly changes over time. This column uses family records from central China over 1300–1900 to estimate social mobility. Within this sample, intergenerational mobility was not necessarily low and changed over time: mobility was substantially higher in the 19th century compared to the 17th, possibly reflecting the 18th-century eradication of hereditary class barriers across Chinese society. Social mobility is also inversely correlated with inequality.
Work by Chetty et al. (2014) using US administrative records has shown that mobility in the US since the 21st century is decidedly mixed, a finding that casts doubt on whether the US has in fact been a uniform ‘land of opportunity’, where a person’s success doesn’t depend on their family background.
For much earlier times, our understanding of social mobility is necessarily limited by the fact that linked socioeconomic status data on parents and children are not easily available. Thus, researchers have relied on pseudo-intergenerational links and the socioeconomic content of names (Clark 2014, Olivetti and Paserman 2015, and Guell et al. 2015) to connect parents’ outcomes to child outcomes. For instance, analysing economies across the globe, Gregory Clark concluded that, due to latent factors, social mobility in pre-industrial times is not only much lower than has often been estimated, but hardly changes over time (Clark 2013, 2014).
Genealogical data allow us to estimate social mobility using actual intergenerational linkages from family records for a population in Central China over the period 1300 to 1900. My analysis (Shiue 2024) suggests that social mobility in pre-industrial economies was not necessarily very low, and social mobility can evolve over time. 1
The results indicate, first, that intergenerational mobility in this sample population changed substantially over time. See Figure 1 for results from the canonical regression approach to relative mobility, which relates the socioeconomic rank of the son to that of the father in an ordinary least squares (OLS) regression.
Figure 1 Intergenerational mobility over time
Note: Shown are βc coefficients from OLS regressions of percentile rank son on percentile rank father (equation (11)). Ten 25-year birth cohorts (1575 to 1825), plus birth cohorts before 1575 and after 1825. Median birth year of son in cohort given on horizontal axis. Earliest son birth year in sample is 1330.
Figure 1 shows that the intergenerational mobility estimate varies from about 0.7 in cohorts born around 1650 to about 0.4 for cohorts born in the early 19th century. In other words, a 10 percentage point higher status of the father translates to a 7 percentage point higher status of the son around the year 1650, but two centuries later, the same parental advantage corresponds to a smaller boost of just over half of what it was, at 4 percentage points. In my paper, I detail how the trends are unlikely to be an artefact of variation in data quality or selection.
The changes correspond to a substantially higher level of mobility in the 19th century compared to the 17th century in this part of China. The timing of the trends corroborates what historians of China have called “the most striking development with respect to mobility” in 18th-century China: the eradication of hereditary class barriers across Chinese society (Mann 1991).
It is also apparent from Figure 1 that during much of the sample period, the level of intergenerational mobility is estimated with a regression coefficient indicating substantial mobility, that is, well below one. When the coefficient is constrained to be the same for all birth cohorts, the estimate is 0.53. This turns out to be within the range of results for certain parts of 19th-century Sweden (Lindahl et al. 2015).
Another important question concerns the relationship between social mobility and inequality. Because a society’s level of mobility quantifies the advantage that children of rich parents have over the children of poor parents, one might think of social persistence as the dynamic, generation-to-generation counterpart of a society’s cross-sectional inequality.
Well-known models of parental investment, such as those proposed by Becker and Tomes (1986), predict that high inequality results in low social mobility. For intuition, suppose that career advancement depends on passing a key exam (such as China’s civil service exam, the keju), which requires investment by the parents to train the young. Consider a scenario of maximal inequality with a Gini coefficient of 1 – where, for example, 1% of society owns all the wealth while the remaining 99% merely subsists – then only wealthy families can afford to prepare their children for career advancement. As a result, only the wealthy can succeed, and consequently, inequality is perpetuated.
The question of whether societies with high cross-sectional inequality tend to have low social mobility has been discussed as the so-called Great Gatsby curve question (Krueger 2012), and cross-country data for advanced countries during the 20th century confirms such a relationship (Corak 2013). As in other cross-country analyses, a difficulty in interpreting this result is that countries differ in other ways that might potentially be correlated with the ‘low mobility to high inequality’ relationship of this Great Gatsby curve. To address this problem, I present time-series evidence on the relationship between inequality and mobility from the 14th to the 19th century (Figure 2).
Figure 2 Social mobility and inequality: Great Gatsby in the time series
Note: Shown are βc and the Theil index of status in the father generation for 12 birth cohorts. Earliest son’s birth is 1330.
Figure 2 shows that social mobility for birth cohorts characterised by high inequality tends to be low, and vice versa. The correlation between the intergenerational coefficient and the measure of inequality is 0.77. This provides evidence that the mobility-inequality relationship of the Great Gatsby curve holds even when potentially confounding factors in a cross-country analysis are eliminated.
In sum, the study of historic samples using actual links between family members shows that over the long run: (1) intergenerational mobility was not necessarily low, (2) it can evolve over time, and (3) mobility is inversely correlated with inequality in a society.
Source : VOXeu
