Were Victorian liberals right to extol nineteenth-century English society as one of openness and high mobility? Or were opportunities few and far between? PhD student Ziming Zhu’s research uses individual-level data spanning 60 years (1851 to 1911) to answer these questions. He argues that his new dataset offers a better representation of social mobility during the period and corrects previous estimates that overstated the level of social mobility between sons and their fathers.

In 1863, Samuel Smiles expounded the prospect of social advancement in nineteenth-century Britain in his work Self Help, a book central to the ideology of Victorian liberalism. Social commentators of the past believed strongly that social mobility was a virtuous by-product of a system that rewarded those who worked hard. But do the data support the idea of a mobile society?

Using a newly constructed and improved set of linked data from the full-count England and Wales decennial censuses, this paper estimates the intergenerational elasticity (IGE) of occupational status in England between 1851 and 1911, following the Becker-Tomes model of intergenerational transmission of human capital (Becker and Tomes, 1986). The results show that, contrary to the findings of some earlier works, social mobility was rather limited during the Victorian (and Edwardian) era. Measurement errors cause significant attenuation bias to estimates of social mobility. Correcting for this bias could raise the IGE obtained from 0.4 to 0.6-0.7, or as much as 64 per cent. The implication: sons are much more likely to stay in the same social level as their fathers than previous studies had suggested.

Most previous studies have relied on marriage registers (Miles, 1993, 1999; Mitch, 1993, 2005) or on surname-based measures (Clark and Cummins, 2015). Long (2013), in a first-of-its-kind study, estimated rates of social mobility using linked census data for England. However, the high rate of mobility found by Long may not represent the true state of nineteenth-century English society. First, his sample size was restricted by the use of a 2 per cent sample of the 1851 census. This raises issues of representativeness and increases the likelihood of false-positives in linking. Second, Long’s methodology does not address the issue of measurement errors highlighted by Ward (2021) in his work on historical mobility in the US. High rates of false-positives and measurement errors could have caused significant attenuation bias, leading to results that overestimate the degree of social mobility.

This paper addresses the previous issues by using the Integrated Census Microdata (I-CeM) complete-count census data, greatly expanding the number of observations. The paper provides revised intergenerational elasticities of occupational status after accounting for both classical measurement errors and false positives in census linking. The results are robust to alternative methods of census linkage and to different occupational indices. False positives and reweighting do not have a significant impact on my findings.