Eighth Annual Historical Economic Demography Workshop

How the experience of slavery might have affected post emancipation trajectories remains an empirically unanswered question. I exploit local variation in enslaved workers' exposure to coercion intensity in two French Caribbean islands: Guadeloupe and Martinique. Using county-level data on enslaved mortality prior to abolition, I document that enslavement conditions were heavily influenced by planters' economic incentives and significantly deteriorated on sugarcane plantations compared to coffee, following heightened competition in the sugar market. Using newly digitized administrative records, I then track outcomes for all formerly enslaved families with children born within five years post-abolition. While exposure to sugarcane is linked to higher potential earnings and skilled positions for fathers, I find that fathers most exposed to sugarcane have a negative effect on their children's survival, relative to families with less coerced or absent fathers. Using place of enslavement and yearly data on enslaved deaths and births before abolition, I find that fathers' negative effect on children is driven by those most exposed to hardship during their childhood, and from the largest plantations. I find suggestive evidence that worse enslavement conditions might have led to more violent men, to which mothers may have responded through strategic matching decisions with less coerced partners. Taken together, my findings point to substantial inequality among descendants of formerly enslaved individuals, driven by paternal response to extreme coercion.

Previous work on the Francophone population of Ontario has highlighted regional differences in its size and importance relative to other groups and variation over time and space in the importance of continuing migration from Quebec (Ouellet 2012). This paper revisits the demographic history of the Franco Ontarian population using newly available complete count census data that permit fine-grained generalizations. A systematic linking of Ontario’s marriage records to the census points to the importance of phenomena inaccessible to previous scholarship: variation in the propensity of Francophones to marry non-Francophones, fertility patterns across such unions, and the long-run circumstances of children of different marital unions and in located inside or outside of major Francophone population centres.

Sattenspiel and colleagues (2025) have recently published “Death on the permafrost: revisiting the 1918-1920 influenza pandemic in Alaska using death certificates” in the American Journal of Epidemiology. This is by far the most detailed quantitative study of the 1918 pandemic mortality for the indigenous and non-indigenous populations of Alaska, and its headline conclusion is that the pandemic death rate was 8 times higher for the former than the latter. However, this study does not consider the completeness of available mortality data relative to the census. Using archival materials and indirect estimation, including intercensal cohort survival evaluation, I assess the coverage of the death registration data used in the study. The completeness of death registration relative to the census is markedly lower for Alaska Natives than Alaska non-natives, especially during the pandemic itself. As a result, the risk ratio of 8 is a significant underestimate. Additionally, the Alaska non-native population is unusual in its age and sex composition, and standardising their demographic structure to that of Alaska Natives pushes their excess mortality downwards. Accounting for this increases the risk ratio further. In general, this shows how issues of differential demographic structure and differential completeness of records can bias estimates of inequality in mortality, especially during historical crises.

Life tables are models of mortality that are used to estimate an expected number of additional years a person can expect to live if they experience the age-specific mortality rates of the given time period for the rest of their life. They are widely used in demography, public health, and social planning to assess population health and mortality trends. However, for the countries of the former Yugoslavia, existing international life expectancy estimates are often limited by unclear data sources and inconsistent methodologies, hindering cross-country comparisons and evidence-based policymaking. This is particularly prominent during the years preceding and throughout the transition to democracy. To improve access to data for this region, I have transcribed archival data and constructed 584 life tables for all former Yugoslav republics for the period 1970-2017. The database was developed following four guiding principles: country dialogue, transparency, comparability, and accessibility, and is publicly available to support further research and policy analysis. I present the process of compiling and validating population and deaths data from civil registration systems, assessing data quality and comparability over time. Key demographic methods, including the Gompertz, Brass Growth Balance, and Preston–Coale models, were applied, and estimates were cross-checked with international datasets. Then, using the new life tables, I analyse trends in life expectancy during the transition from socialism to democracy, a period marked by major socio-political crisis. I find that although life expectancy improved in all countries, the dynamics of change were diverse and context-specific. Initially, life expectancy increases were driven by reduction in infant and child mortality in all countries. During the transition period divergence of trends occurred. In Slovenia and Croatia further reductions in adult mortality led to an increase in life expectancy, while in Serbia and Montenegro increases in adult mortality reduced life expectancy. In the 2000s reductions in mortality at older ages led to increases in life expectancy in all countries, but the extent of the increase varied across countries. These findings show that the countries of ex-Yugoslavia experienced different patterns of mortality change during the transition compared to other Eastern European countries. This highlights the context specific variation of trends in population health in times of transition, inspiring further research on determinants of health, as well as development of targeted policies.

The Code Napoléon attempted to impose a New Regime of the Family. In a conscious effort to adhere to the Revolutionary ideal of equality, universal partition of inheritance was introduced. Although intensely patriarchal in most other regards (Tudor, 2022), this included the introduction of legal equality of women in inheritance. Using a new dataset of 1.8 million observations of wealth at death and employing a Regression Discontinuity Design, I investigate whether pre-Revolutionary customary legal norms, which were not outlawed by the Code Napoléon, had persistent effects on the gender wealth gap. I find that pre-revolutionary customary marital community property norms (communauté des acquests) had persistent negative effects on women’s wealth at death. Counterintuitively, I find that previous customary exclusion of women from inheritance had positive effects on women’s wealth after women had been included in inheritance by the Code Napoléon. I hypothesize that this is because societies which disadvantage women in inheritance developed offsetting mechanisms, such as higher dowries, which also persisted into the Code Napoléon era. These findings are in line with Bessière and Gollac (2022) and show how inequality between genders is (re-)produced within families through private practices and how legal equality is not a sufficient condition for economic equality.

We document changes in global inequality of opportunity in education for women and men born between 1941 and 1983, using individual-level census and survey data on 46.7 million individuals from 95 countries, representing all major regions of the world. We measure global inequality of opportunity in education as inequality in education due to circumstances beyond the control of individuals. In addition to gender and social origin, we treat a person's country of residence as a circumstance that produces inequality of opportunity, because the country of residence is, to a large extent, beyond an individual's control. We test whether global inequality of opportunity in education has increased or decreased. Our results show a decline in global inequality of opportunity. The decline is stronger among women than men, although inequality of opportunity remains higher among women than men.

Migrant networks reduce migration costs and create self-reinforcing migration flows, a phenomenon known as chain migration. Empirical evidence for migrant network effects come sfrom a limited number of cases, such as contemporary Mexico-US migration or historical European mass migration. This paper explores whether being networked to migrants would affect people’s propensity to migrate in the case of 19th- and early 20th-century Chinesemigration, with Southeast Asia being the major destination. It uses Chinese genealogical data to explore the variation in migrant networks arising from the structure of migrants’ extended families, which is unavailable in other sources and therefore under-explored in the literature. I study the most recent 10 to 12 generations of male descendants from 5 genealogies, mostly born between the 17th and early 20th centuries. Using the length of the shortest path in graph theory, I quantify an individual’s kin-based migrant networks by how connected they are to previous migrants based on the genealogical distances. This measure captures both the number of previous migrant relatives an individual has and their position relative to these migrants in the family tree. It allows me to detect the within-surname-group variation in migrant networks with logistic regression. I show that closer connections to migrants in one’s kinship networks increase the odds of migrating. This meaningful variation is flattened when migrant networks are measured solely by their size, which treats all kin as equally relevant. I further compute individuals’ connectedness to specific types of migrants and show that some migrant relatives had no detectable influence, even though they contribute to the network size in simple counts. Connections to migrants whose destinations are unknown to the wider clan had little impact on an individual’s decision to move. Connections to migrants who were too distantly related or too old to a focal person also had little impact on their decision to move. Taken together, the results suggest that migration network should not be viewed as an undifferentiated whole, but a structurally differentiated set of ties with varied relevance for migration

We investigate how exposure to universal healthcare in infancy influences early life health outcomes and the extent to which impacts persist into later life. The introduction of universal healthcare through the National Health Service (NHS) in the United Kingdom in July 1948 presented a fundamental re-organisation of the healthcare environment. Immediate large decreases in infant mortality ensued. We show an immediate decrease in infant mortality ensued, particularly in i) the neo-natal period, ii) for amenable causes of death, and iii) among low-income individuals. We use administrative data on hospitalisations and death records to compare individuals born in the immediate cohorts around the NHS introduction in a regression discontinuity design. Our findings indicate that age-specific survival rates in later life are systematically higher among those whose post-natal care expanded through the NHS. These long run impacts of infancy exposure to universal healthcare coverage through the NHS are economically significant. Exploiting geographical variation in the change in per-capita medical services induced by the introduction of universal healthcare reveals evidence of crowding out effects between pre-existing and incoming patients to the new system. Our analysis suggests that supply side constraints play a key role in the design of universal healthcare systems.

The secular trend of growth in height suddenly slowed for Americans born after the middle of the 20th century, and the health and human capital of these cohorts as adults appears to have declined, or at least stagnated, more broadly. This paper presents evidence that the physical growth of these unhealthy cohorts was particularly stunted during adolescence. Using data from NHANES and its predecessors, I show that males born in the 1960s were the same height in childhood as those born a decade earlier, but then fell behind and were half an inch shorter in adolescence. By adulthood, the heights of the two cohorts were nearly identical. This suggests that males born in the 1960s had a later or smaller adolescent growth spurt than those born a decade earlier. Similar patterns are not evident in the height of females; however, females born in the 1960s experienced menarche later than those born a decade earlier. The delayed puberty of cohorts born in the 1960s appears to be a short-term blip in a long-run trend towards earlier puberty. Later-born cohorts were taller in adolescence but approximately the same height in adulthood as those born in the 1950s and 1960s. The findings strongly suggest that something had already gone wrong by at least adolescence for American cohorts born after mid-century.