Where Is the Middle Class? Evidence from 60 Million English Death and Probate Records, 1892–1992

Abstract

This article analyzes a newly constructed individual level dataset of every English death and probate from 1892–1992. This analysis shows that the twentieth century’s “Great Equalization” of wealth stalled in mid-century. The probate rate, which captures the proportion of English holding any significant wealth at death rose from 10 percent in the 1890s to 40 percent by 1950 and has stagnated to 1992. Despite the large declines in the wealth share of the top 1 percent, from 73 to 20 percent, the median English individual died with almost nothing throughout. All changes in inequality after 1950 involve a reshuffling of wealth within the top 30 percent. I translate the individual level data to synthetic households; the majority have at least one member probated. Yet the bottom 60 percent of households hold only 12 percent of all wealth, at their peak wealth-holding level, in the early 1990s. I also compare the new wealth data with existing estimates of top wealth shares, home-ownership trends, wealth survey distributions, aggregate wealth, and the wealth Gini coefficient.

The Second World War, Inequality and the Social Contract in Britain

“…Using these data, we can infer how many individuals are missing from the probate data. Following Cummins (2021), we can impute the wealth of those not in the probated sample by assigning to them half of the probate threshold. This procedure allows us to recover a measure of wealth for the full population, rather than only for those people wealthy enough to be probated.…”

“…Probates needed to be filed based on a wealth cut‐off. Cummins (2021) documents the proportion probated each year and the probate threshold each year. Somewhere between 30% and 70% of individuals who pass away in a given year are probated, depending on the year.…”

“…To investigate these issues, we put together the most comprehensive existing dataset of wealth-at-death for all people in England and Wales (we do not have data for Scotland), by year, between 1930 and 1954. Though these data have been examined in aggregate time series before (for example, by Harrison and Atkinson 1978;Alvaredo et al 2018;Cummins 2021), the microdata have never been analysed empirically. This allows us to construct a geolocated measure of wealth inequality (these 'probate' records record the street address of the deceased from which we identified the latitude and longitude).…”

“…Though these data have been examined in aggregate time series before (for example, by Harrison and Atkinson 1978; Alvaredo et al . 2018; Cummins 2021), the microdata have never been analysed empirically. This allows us to construct a geolocated measure of wealth inequality (these ‘probate’ records record the street address of the deceased from which we identified the latitude and longitude).…”

The rise of machine learning in the academic social sciences

“…The Fragile Families Challenge (Salganik et al 2020) aimed to generate a better understanding of social determinism, but not every emergent application need be survey based. The use of optical character recognition (OCR) for digitizing archi-val population records (Cummins 2021) and the prediction of history (Risi et al 2019) are prime examples of other recent and exciting applications of what ML makes possible. There are substantial public policy applications and opportunities for intervention based upon prediction, too; if we can more accurately predict rain tomorrow, we can better plan to bring an umbrella.…”