Sex ratios and missing girls in nineteenth century Europe

By Francisco J. Beltrán Tapia (Norwegian University of Science and Technology)

This blog is part of our EHS Annual Conference 2020 Blog Series.

The flying girl. Available at Wikimedia Commons.

Gender discrimination – in the form of sex-selective abortion, female infanticide and the mortal neglect of young girls – constitutes a pervasive feature of many contemporary developing countries, especially in South and East Asia. Son preference stems from economic and cultural factors that have long influenced the perceived relative value of women in these regions and resulted in millions of ‘missing girls’.

But were there ‘missing girls’ in historical Europe? Although the conventional narrative argues that there is little evidence for this kind of behaviour (here), my research shows that this issue was much more important than previously thought, especially (but not exclusively) in Southern and Eastern Europe.

It should be noted first that historical sex ratios cannot be compared directly to modern ones. The biological survival advantage of girls was more visible in the high-mortality environments that characterised pre-industrial Europe. Subsequently, boys suffered higher mortality rates both in utero and during infancy and early childhood. Historical infant and child sex ratios were therefore relatively low, even in the presence of gender-discriminatory practices.

This is illustrated in Figure 1 below, which plots the relationship between child sex ratios (the number of boys per 100 girls) and infant mortality rates using information from European countries between 1750 and 2001. In particular, in societies where infant mortality rates were around 250 deaths (per 1,000 live births), a gender-neutral child sex ratio should have been slightly below parity (around 99.5 boys per 100 girls).

Figure 1: Infant mortality rates and child sex ratios in Europe, 1750-2001

Compared with this benchmark, infant and child sex ratios were abnormally high in some European regions (see Map 1 below), suggesting that some sort of gender discrimination was unduly increasing female mortality rates at those ages.

Interestingly, the observed differences in sex ratios are also visible throughout childhood. In fact, the evolution of sex ratios by age shows stark disparities across countries. Figure 2 shows how the number of boys per 100 girls changes as children grew older for a sample of countries, both in levels and in the observed trends.

In Bulgaria, Greece and France, for example, sex ratios increased with age, providing evidence that gender discrimination continued to increase female mortality rates as girls grew older. Importantly, the unbalanced sex ratios observed in some regions are not due to random noise, female under-registration or sex-specific migratory flows.

Likewise, although geography, climate and population density contributed to shaping infant and child sex ratios due to their impact on the disease environment, these factors cannot explain away the patterns of gender discrimination reported here.

Map 1: Child sex ratios in Europe, c.1880

Figure 2: Sex ratios by age in a sample of countries, c.1880

This evidence indicates that discriminatory practices with lethal consequences for girls constituted a veiled feature of our European past. But the actual nature of discrimination remains unclear and surely varies by region.

Excess female mortality was then not necessarily the result of ill treatment of young girls, but could have been just based on an unequal allocation of resources within the household, a circumstance that probably cumulated as infants grew older.

In contexts where infant and child mortality rates are high, a slight discrimination in the way that young girls were fed or treated when ill, as well as in the amount of work with which they were entrusted, was likely to have resulted in more girls dying from the combined effect of undernutrition and illness.

Although female infanticide or other extreme versions of mistreatment of young girls may not have been a systematic feature of historical Europe, this line of research would point to more passive, but pervasive, forms of gender discrimination that also resulted in a significant fraction of missing girls.

The Diaspora of a Diaspora: The Cassana and Rivarolo family network in the Atlantic, 1450-1530

By Andres Mesa (Università degli Studi di Teramo)

This research is due to be presented in the sixth New Researcher Online Session: ‘Spending & Networks’.

The Coast of Genoa, by Jasper Francis Cropsey, 1854. Available at Wikimedia Commons.

My project re-assesses the nature of Genoese family networks in the Atlantic, at the end of the fifteenth and early sixteenth centuries. The canonical understanding of these networks is based on three observations: family networks were highly co-dependent, centralized, and that Genoa was the centre of operations for all the Genoese. My research shows a multitude of scenarios and provides new explanations for these observations. Using a case study of the Rivarolo-Cassana family network, I show that this particular network functioned more in terms of cooperation, using a pluricentric language. As the title suggests, the economic endeavours of these merchants involved  a complex migration process. Consequently, their trading activities coincided with the interests of those in permanent settlements.

Genoese merchants chose cities in the Iberian peninsula for their homes and as a base for their  business activities. For example, Lisbon, Seville, and Valencia had a significant permanent Genoese population who were in the process of becoming naturalized Spanish and Portuguese citizens. In turn, some of the families that dominated the economic landscape of the 15th and 16th centuries disappeared in Genoa. Yet, their descendants still appear in Portugal and Spain with their original Ligurian surnames altered into Castilian or Portuguese.

The findings from my study indicate the need for a major reassessment of our understanding of Genoese family networks. The data I have collected shows that most of the day-to-day trade happened outside the family network, and the contractual relationships that emerged between partners  extended well beyond familial ties. Because the structure of private property ownership was connected to new interests and new markets, it was inevitable that these, in turn, were linked to the discovery of new lands. Consequently, The Genoese adopted a new business model based on owning the means of production for the goods they traded, particularly soap, wheat, and sugar.

Finally, I argue that the economic ties between families and family members, did not always translate into a share of business responsibility or welfare. The relationships and partnerships functioned in terms of very particular historical and geographical contexts. The contracts were between ‘individuals (Societas) to share losses and gains.’ Thus, liability was an individual matter despite the frequent use of jurists.

Andrés Mesa

Twitter: @mesaandres

The Paradox of Redistribution in time: Social spending in 54 countries, 1967-2018

By Xabier García Fuente (Universitat de Barcelona)

This research is due to be presented in the sixth New Researcher Online Session: ‘Spending & Networks’.

Money of various currencies. Available at Wikimedia Commons.

Why are some countries more redistributive than others? This question is central to current welfare state politics, especially in view of rising levels of inequality and the ensuing social tensions. Since coming to power in 2019, Brazil’s far-right government has restricted access to Bolsa Familia—a conditional cash-transfer program—despite its success at reducing poverty with a very low cost (less than 0.5% of national GDP). In richer countries, the social-democratic project is said to be obsolete, as left-wing parties forsake egalitarian policies to cater to economic winners (Piketty, 2020).

How can we make sense of this sort of distributive conflict? Are there common patterns in rich and middle-income countries? My research suggests that welfare state institutions show great inertia, so we need to observe the origins of social policies to explain current redistributive outcomes. Initial policy positions —how pro-poor or pro-rich social transfers were— determine what groups emerge as net winners or net losers when social expenditure increases, which crucially affects the viability and direction of policy change.

Korpi and Palme (1998) famously suggested the existence of a Paradox of Redistribution: ‘the more we target benefits at the poor … the less likely we are to reduce poverty and inequality’. In their framework, progressive programs may be more redistributive per euro spent, but they generate zero-sum conflicts between the poor and the middle-class and obstruct the formation of redistributive political coalitions. In contrast, universal programs align the preferences of the poor and the middle-class and lead to bigger, more egalitarian welfare states. In sum, redistribution increases as transfers become bigger and less pro-poor.

Using survey micro-data provided by the Luxembourg Income Study (LIS), my research updates Korpi and Palme’s (1998) study and addresses two gaps. First, I extend the sample to 54 rich and middle-income countries, including elitist welfare states in Latin America and other middle-income countries. As Figure 1 shows, extending the sample would clearly refute the Paradox: redistribution is higher in more pro-poor countries.

Second, in line with the dynamic political arguments suggested in the Paradox, I explore the evolution of social transfers and redistribution within countries over time. Overall, countries have increased redistribution by making their transfers less pro-poor, which matches the predictions of the Paradox (see Figure 2). The relationship is especially strong in Ireland, Canada, United Kingdom and Norway. Parting from highly progressive (pro-poor) policy positions, these countries have improved redistribution increasing expenditure and reducing their bias towards the poor.

Latin American countries are a notorious exception to this pattern. They are markedly pro-rich and, contrary to the cases above, they have improved redistribution very modestly by becoming more pro-poor (see Figure 3).

What does it mean that redistribution increases as transfers become more or less pro-poor? United Kingdom and Mexico provide a good example (see Figure 4). In the United Kingdom, redistribution through social transfers increased from 7 Gini points in 1974 to 19 Gini points in 2016. In the same period, the share of total social transfers received by the poorest 20% of the population decreased from 35% to 18%. In Mexico, the share of total social transfers obtained by the poorest 20% went from 2% in 1984 to 10% in 2016, while the share obtained by the richest 20% decreased from 66% to 51%. Yet, despite these advances, redistribution through social transfers in Mexico remains very low (2.5 Gini points in 2016, from 0.1 Gini points in 1984).


In countries with pro-poor social transfers, extending coverage involves reaching up the income ladder to include richer constituencies, which narrows the gap between net winners and net losers. This reduces the salience of distributive conflicts and eases welfare state expansion, leading to higher redistribution. However, as transfers become more pro-rich the margin to leverage the progressivity-size trade-off narrows, which helps explain the inability of current welfare states to increase redistribution as inequality rises.

In countries with pro-rich social transfers, extending coverage involves reaching down the income ladder to include the poor. Launching programs for the poor requires rising taxes or cutting the benefits of privileged insiders, which creates a clearly delineated gap between net winners and net losers. This increases the salience of distributive conflicts, leading to smaller, less egalitarian welfare states.

In sum, social policy design is very persistent because it crucially shapes distributive conflicts. Advanced welfare states have increased redistribution by getting bigger and less progressive (less pro-poor). This fits with historical evidence that advanced welfare states grew from minimalist cores, but it also describes contemporary policy change. Following this same reasoning, elitist welfare states in developing regions will find it difficult to become more egalitarian. Figure 5 shows the persistency of distributive outcomes across welfare regimes.


Korpi, W. and Palme, J. (1998). The paradox of redistribution and strategies of equality: Welfare state institutions, inequality, and poverty in the western countries. American Sociological Review, 63(5):661–687.

Piketty, T. (2020). Capital and Ideology. Harvard University Press.

Xabier García Fuente

Twitter: @xabigarf

Settler capitalism: company colonisation and the rage for speculation (NR Online Session 5)

By Matthew Birchall (Cambridge University)

This research is due to be presented in the fifth New Researcher Online Session: ‘Government & Colonization’.


Scan from “Historical Atlas” by William R. Shepherd, New York, Henry Holt and Company, 1923. Available at Wikimedia Commons.

My research explores the little-known story of how company colonisation propelled the settler revolution. Characterised by mass emigration to Britain’s settler colonies during the long nineteenth century, the settler revolution transformed Chicago and Melbourne, London and New York, drawing all into a vast cultural and political network that straddled the globe. But while the settler revolution is now well integrated into recent histories of the British Empire, it remains curiously disconnected from the history of global capitalism.

Prising open what I call the inner lives of colonial corporations, I tell the story of how and why companies remade the settler world. It takes a fresh look at the colonial history of Australia and New Zealand in an attempt to map a new history of chartered colonial enterprise, one that is as sensitive to rhetoric as it is to ledgers documenting profit and loss. We tend to understand companies in terms of their institutional make-up, that is to say their legal and economic structure, but we sometimes forget that they are also cultural constructions with very human histories.

The story that I narrate takes us from the boardrooms of the City of London back out to the pastures of the colonial frontier: it is a snapshot of settler capitalism from the inside out. From the alleys and byways immortalised in Walter Bagehot’s Lombard Street (1873) to the sheep-runs of New South Wales and the South Canterbury plains, company colonisation has a global history – a history that links the Atlantic and the antipodes, Māori and metropolitan capital, country and the City of London. My study marks a first attempt at bringing this history to light.

In digging deep into the social and cultural history of company colonisation, I focus in particular on the legitimating narratives that underwrote visions of colonial reform. How did these company men make sense of their own ventures? What traditions of thought did they draw on to justify the appropriation of indigenous lands? How did the customs and norms of the City shape the boundaries of what was deemed possible, let alone appropriate in the extra-European world? I aim to show that company colonisation was as much an act of the imagination as it was the product of prudent capital investment.

My research engages with large questions of contemporary relevance: the role of corporations in the making of the modern world; the relationship between empire and global capitalism; and the salience of social and cultural factors in the development of corporate enterprise. I hope to enrich these debates by injecting the discussion with greater historical context.



Coordinating Decline: Governmental Regulation of Disappearing Horse Markets in Britain, 1873-1957 (NR Online Session 5)

By Luise Elsaesser (European University Institute)

This research is due to be presented in the fifth New Researcher Online Session: ‘Government & Colonization’.


Milkman and horse-drawn cart – Alfred Denny, Victoria Dairy, Kew Gardens, Est 1900. Available at Wikimedia Commons.

The enormous horse drawn society of 1900 was new. An unprecedented amount of goods and people could only be moved by trains and ships between terminal points and therefore, horses were required by anybody and for everything to reach its final destination. But, the moment the need for horsepower peaked, new technologies had already started to make the working horse redundant for everyday economic life. The disappearance of the horse was a rapid process in the urban areas, whereas the horse remained an economic necessity much longer in other areas of use such as agriculture. The horses decline left behind deep traces causing fundamental changes in soundscapes, landscapes, and smells of human environment and economic life.


Against prevailing narratives of a laissez-faire approach, the British government monitored and shaped this major shift in the use of energy source actively. The exploration of the political economy of a disappearing commercial good examines the regulatory practices and ways the British government interacted with producers and consumers of markets. This demonstrates that governmental regulations are inseparable from modern British economy and that government intervention follows the careful assessment of costs and benefits as well as self-interest over the long time period.

Public pressure groups such as the RSPCA as well as social and business elites were often strongly connected to government circles embracing the opportunity to influence policy outcomes. For instance, the Royal Commission on Horse Breeding was formed in December 1887 is telling because it shows where policy making power that passed through Westminster originated. The commissionaires were without exception holders of heredity titles, members of the gentry, politicians, or businessmen, and all were avid horsemen and breeders. To name but two, Henry Chaplin, the President of the Board of Agriculture, had a family background of Tory country gentlemen and was a dedicated rider, and Mr. John Gilmour, whose merchant father grew rich in the Empire, owned a Clydesdale stud of national reputation. Their self-interest and devotion to horse breeding seems obvious, especially in the context of the agricultural depression when livestock proved more profitable than the cultivation of grain.

Although economic agents of the horse markets were often moving within government circles, they had to face regulations. For example, a legal framework was developed which fashioned the scope of manoeuvre for import and export markets for horses. The most prominent case during the transition from horse to motor-power was the emergence of an export market of horses for slaughter. British charitable organisations such as the RSPCA, the Women’s Guild for Empire, and the National Federation of Women’s Institute pressured the government to prevent the export of horses for slaughter on grounds of “national honour” since the 1930s. However, though the government never publicly admitted it, the meat market was endorsed to manage the declining utility of horsepower. With technologies becoming cheaper, horsemeat markets were greeted by large businesses such as railway companies as way to dispose of their working horses without making a financial loss. Hence, the markets for working horses were not merely associated with the economic use and demand for their muscle power but were linked to government regulation.

Ultimately, an analysis of governmental coordination can be linked to wider socio-cultural and economic systems of consumption because policy outcome indeed influenced the use of the horse but likewise coordination was monitored by the agents of the working horse markets.

Luise Elsaesser

Twitter: @Luise_Elsaesser

How to Keep Society Equal: The Case of Pre-industrial East Asia (NR Online Session 4)

By Yuzuru Kumon (Bocconi University)

This research is due to be presented in the fourth New Researcher Online Session: ‘Equality & Wages’.


Theatrum orbis terrarum: Map Tartaria, by Abraham Ortelius. Available at State Library Victoria.



Is high inequality destiny? The established view is that societies naturally converge towards high inequality in the absence of catastrophes (world wars or revolutions) or the progressive taxation of the rich. Yet, I show that rural Japan, 1700-1870, is an unexpected historical case in which a stable equality was sustained without such aids. Most peasants owned land, the most valuable asset in an agricultural economy, and Japan remained a society of land-owning peasants. This contrasts with the landless laborer societies of contemporary Western Europe which were highly unequal. Why were the outcomes so different?

My research shows that the relative equality of pre-industrial Japan can partly be explained by the widespread use of adoptions in Japan, which was used as a means of securing a male heir. The reasoning becomes clear if we first consider the case of the Earls Cowper in 18th century England where adoption was not practiced. The first Earl Cowper was a modest landowner and married Mary Clavering in 1706. When Mary’s brother subsequently died, she became the heiress and the couple inherited the Clavering estate. Similar (miss)fortunes for their heirs led the Cowpers to become one of the greatest landed families of England. The Cowpers were not particularly lucky, as one quarter of families were heirless during this era of high child mortality. The outcome of this death lottery was inequality.

Had the Cowpers lived in contemporary Japan, they would have remained modest landowners. An heirless household in Japan would adopt a son. Hence, the Claverings would have an adopted son and the family estate would have remained in the family. To keep the blood in the family, the adopted son may have married a daughter if available. If unavailable, the next generation could be formed by total strangers but they would continue the family line. Amassing a fortune in Japan was unrelated to demographic luck.

Widespread adoptions were not a peculiarity of Japan and this mechanism can also explain why East Asian societies were landowning peasant societies. China also had high rates of adoption in addition to equal distributions of land according to surveys from the 1930s. Perhaps more surprisingly, adoptions were common in ancient Europe where the Greeks and Romans practiced adoptions to secure heirs. For example, Augustus, the first emperor of the Roman Empire, was adopted. Adoptions were a natural means of keeping wealth under the control of the family.

Europe changed due to the church discouraging adoptions from the early middle ages, leading to adoptions becoming rarities by the 11th century. The church was partially motivated by theology but also by the possibility that heir-less wealth would get willed to the church. They almost certainly did not foresee that their policies would lead to greater wealth inequality during the subsequent eras.


Figure 1. Land Distribution under Differing Adoption Regimes and Impartible Inheritance



My study shows by simulation that a large portion of the difference in wealth inequality outcomes between east and west can be explained by adoption (see figure 1). Societies without adoption have wealth distribution that are heavily skewed with many landless households unlike those with adoptions. Therefore, family institutions played a key role in determining inequality which had huge implications for the way society was organized in these two regions.

Interestingly, East Asian societies still have greater equality in wealth distributions today. Moreover, adoptions still amount to 10% of marriages in Japan which is a remarkably large share. Adoption may have continued creating a relatively equal society in Japan up to today.

Did the Ottomans Import the Low Wages of the British in the 19th Century? An Examination of Ottoman Textile Factories (NR Online Session 4)

By Tamer Güven (Istanbul University)

This research is due to be presented in the fourth New Researcher Online Session: ‘Equality & Wages’.


The Istanbul Grand Bazaar in the 1890s. Available at Wikimedia Commons.

Compared to the UK and Western Europe, there are a limited number of studies on wages and standards of living in the Ottoman empire. For the Ottoman empire the only source that can provide regular wage data for industry are the Ottoman state factories established in the 1840s to meet the needs of the state’s growing and centralized military and bureaucracy. Limitations in the sources of data are explained by the relative absence of industrial wage series in the monographies on Ottoman industrial institutions, and that manufacturing mainly comprised small manufacturers who did not keep records. This paucity in data may change as the Ottoman Archives become fully catalogued. The main aim of this study is to construct a wage series using the wage ledgers of those working in state factories. Consequently, I examined four prominent textile-related factories: Hereke Imperial Factory, Veliefendi Calico Factory, Bursa Silk Factory, and İzmit Cloth Factory. Only the Hereke Factory offers a 52-year wage series between 1848-1899. The data for the Veliefendi Factory started in 1848 but are disrupted in 1876 as the factory was transferred to military rule; the same applies to the İzmit Factory, which was established in 1844, but transferred to military rule in 1849.

I created two separate daily and monthly wage series to determine how many days workers worked per month and how this changed during the nineteenth century. Thus, not only the workers’ potential wages but also the workers’ observed monthly wages can be analysed. Some groups of workers were eliminated from the dataset for a variety of reasons. For example, civilian officials and masters working in factories were excluded because of their relatively high wages. Conversely, because of their relatively low wages, I also exclude carpet weavers —  mostly young girls and children.  I preferred to use median values for monthly wage series to include as many workers as possible in the analysis. As with much historical data, the wage series created in this study are incomplete. To overcome this I complement data for the Hereke Factory wage series with data from the  Veliefendi and Bursa Factories.

My results indicate that daily real wages increased by only by 0.03 per cent,  per annum,  between  1852 and 1899.  However, the real monthly wages of Hereke Factory workers rose by 0.11 per cent, per annum,  between 1848 and 1899, but by 0.24 per cent per annum using 1852 as a starting point.  Monthly wages increased faster than daily wages, but at the cost of more workdays for workers. Average workdays increased by 0.44 per cent, per annum over the span of the period. Although the Veliefendi Factory provides a narrower wage series from 1848 to 1876, it supports this pattern. Limited,  but prominent examinations of Ottoman wage history claim that construction, urban, and agricultural workers’ wages increased, albeit at different rates in the same period. How can we explain the increase in wages of other sectors when the wages of textile workers were stagnant?

Many observations on the Ottoman cities has shown that industrial production, particularly in the textile sector, shifted from urban to rural,  or from craft workshops to houses, to compete with cheap British yarn and fabric in the 19th century. According to my calculations, imports of Ottoman cotton yarn increased by a factor of 25 to 50 in the 19th century. This trend was most pronounced after the 1838 Anglo-Turkish Convention, when cheap English products were imported into the Ottoman Empire, and Ottoman producers sought cheaper labour.  Labour-saving machines both facilitated the export of  British yarns and fabrics to, and lowered wages in, the Ottoman empire.  Although the wage series for the Hereke factory,  and, to a more limited extent,  the Veliefendi factory provide evidence in support of this hypothesis, numerous studies on Ottoman industry in the 19th-century support the same argument, though without a wage series.

Women in the German Economy: A Long Way to Gender Equality (NR Session 4)

By Theresa Neef (Freie Universität Berlin)

This research is due to be presented in the fourth New Researcher Online Session: ‘Equality & Wages’.


Scanned image of a mid-1930s postcard depicting Unter den Linden in Berlin. Available at Wikimedia Commons.

Female employees in the European Union (EU-27) earn, on average, about 85 per cent of the wages received by male employees. While some countries such as France and Sweden exhibit closer pay equality, women in Germany face a larger gap and receive just 79 per cent of the average male wage, according to the 2018 results from Eurostat 2020. How did this state of affairs emerge?

To understand contemporary pay inequality, it is vital to take a long-run perspective and look at the development of the gender pay ratio in Germany since 1913.  An in-depth analysis of historical inquiry reports and publications by the statistical offices reveals that in 1913 women in Germany earned around 44 per cent of male wages. Although  World War I led to a temporary increase in women’s pay in blue-collar occupations, this trend was soon reversed and the gender-segregated labour market was re-established following demobilization.

The interwar period brought about the most dynamic leap in gender relations during the 20th century. While in 1920 German women earned on average 45% of a man’s average pay, by 1937 this share had increased to 61%, a consequence of women’s occupational transition and the more progressive institutional framework adopted during the Weimar Republic.

With the growing number of white-collar jobs, young females had job opportunities that were better paid and more socially accepted than the work in low-paid domestic services or agriculture. That was an opportunity they took: from 1910 to 1960, women increased their share in those fast-growing occupations from 18% to 45%, while their share decreased in agricultural work. This trend most likely contributed to women’s wage gains relative to men.

During the Weimar Republic, a new constitution and a more progressive institutional framework fostered further equalization of earnings, especially in the white-collar occupations. In 1919, the Weimar constitution introduced compulsory schooling for all youths under 18 years irrespective of gender. For the first time, this law provided girls with the same chances to receive vocational education and an apprenticeship as their male peers. All youths that worked in commercial and industrial firms were obliged to attend vocational commercial school at least once a week for two to three years.  Before the introduction of this law, employers hardly invested in girls’ apprenticeships because women were seen as transient employees leaving the labour force upon marriage. This non-gendered schooling obligation led to a dynamic convergence of vocational training between boys and girls.

In the post-1945 period, the gender pay gap decreased in Germany from 65 percent in 1960 to 74 per cent  twenty years later. In contrast,  Sweden took the lead among European countries and by 1980, the gender pay gap was just 14 percentage points. However, since the 1980s, the gender pay gap has stagnated in many European countries.


Figure 1: Gender pay ratio, Germany, Sweden, and the USA. Swedish and German series based on mean earnings; US-American time series based on median earnings if not indicated differently. The German time series covers the German Reich, the Federal Republic of Germany and reunified Germany (hollow items).


All in all, the long-run perspective shows that since the beginning of the 20th century Germany has persistently exhibited a lower gender pay equality than other European economies, such as Sweden, despite the important improvement observed in the interwar period. In the postwar period, the gap between Germany and Sweden widened further due to slower progress in the young Federal Republic. These results suggest that differences in gender pay inequality across countries can be traced back to historical roots that go beyond the developments in the past forty years.

Before the fall: quantity versus quality in pre–demographic transition Quebec (NR Online Session 3)

By Matthew Curtis (University of California, Davis)

This research is due to be presented in the third New Researcher Online Session: ‘Human Capital & Development’.


Map of East Canada or Quebec and New Brunswick, by John Tallis c.1850. Available at Wikimedia Commons.

While it plays a key role in theories of the transition to modern economic growth, there are few estimates of the quantity-quality trade-off from before the demographic transition. Using a uniquely suitable new dataset of vital records, I use two instrumental variable (IV) strategies to estimate the trade-off in Quebec between 1620 and 1850. I find that one additional child who survived past age one decreased the literacy rate (proxied by signatures) of their older siblings by 5 percentage points.

The first strategy exploits the fact that twin births, conditional on mother’s age and parity, are a random increase in family size. While twins are often used to identify the trade-off in contemporary studies, sufficiently large and reliable historical datasets containing twins are rare. I compare two families, one whose mother gave birth to twins and one whose mother gave birth to a singleton, both at the same parity and age. I then look at the probability that each older non-twin sibling signed their marriage record.

For the second strategy, I posit that aggregate, province-wide infant mortality rate during the year a younger child was born is exogenous to individual family characteristics. I compare two families, one whose mother gave birth during a year with relatively high infant mortality rate, both at the same parity and age. I then look at older siblings from both families who were born in the same year, controlling for potential time trends in literacy. As the two different IV techniques result in very similar estimates, I argue there is strong evidence of a modest trade-off.

By using two instruments, I am able to rule out one major source of potential bias. In many settings, IV estimates of the trade-off may be biased if parents reallocate resources towards (reinforcement) or away from (compensation) children with higher birth endowments. I show that both twins and children born in high mortality years have, on average, lower literacy rates than their older siblings. As one shock increases and one shock decreases family size, but both result in older siblings having relatively higher human capital, reinforcement or compensation would bias the estimates in different directions. As the estimates are very similar, I conclude there is no evidence that my estimates suffer from this bias.

Is the estimated trade-off economically significant? I compare Quebec to a society with similar culture and institutions: pre-Revolutionary rural France. Between  1628 and 1788, a woman surviving to age 40 in Quebec would expect to have 1.7 additional children surviving past age one compared to her rural French peers. The average literacy rate (again proxied by signatures) in France was about 9.5 percentage points higher than in Quebec. Assuming my estimate of the trade-off is a linear and constant effect (instead of just a local average), reducing family sizes to French levels would have increased literacy by 8.6 percentage points in the next generation, thereby eliminating most of the gap.

However, pre-Revolutionary France was hardly a human capital-rich society. Proxying for the presence of the primary educators of the period (clergy and members of religious orders) with unmarried adults, I find plausible evidence that the trade-off was steeper in boroughs and decades with greater access to education. Altogether, I interpret my results as evidence that a trade-off existed which explains some of the differences across societies.


Data Sources

Henry, Louis, 1978. “Fécondité des mariages dans le quart Sud-Est de la France de 1670 a 1829,” Population (French Edition), 33 (4/5), 855–883.

IMPQ. 2019. Infrastructure intégrée des microdonnées historiques de la population du Québec (XVIIe – XXe siècle) (IMPQ). [Dataset].Centre interuniversitaires d’études              québécoises (CIEQ).

Programme de recherche en démographie historique (PRDH). 2019. Registre de la population du Québec ancien (RPQA). [Dataset]. Département de Démographie, Université de Montréal.

Projet BALSAC. 2019. Le fichier BALSAC. [Dataset]. L’Université du Québec à Chicoutimi.

Honest, sober and willing: Oxford college servants 1850-1939 (NR Online Session 3)

By Kathryne Crossley (University of Oxford)

This research is due to be presented in the third New Researcher Online Session: ‘Human Capital & Development’.


The library of Christ Church, Oxford from Rudolph Ackermann’s History of Oxford (1813). Available at Wikimedia Commons.



Oxford colleges were among the earliest employers in England to offer organised pension schemes for their workers. These schemes were remarkable for several reasons: they were early, the first was established in 1852; they included domestic servants, rather than white-collar workers; and colleges were unlike typical early adopters of pension schemes, which tended to be large bureaucratic organisations, such as railways or the civil service.

The schemes developed from various motives: from preventing poverty in workers’ old age to promoting middle-class values, like thrift and sobriety, through compulsory savings.

Until the Second World War, college servants were often described as a ‘labour aristocracy’, and while there were many successful senior servants, equally there were many casual, part-time and seasonal workers. The experience of these workers provides an unusually detailed look at the precarity of working-class life in the nineteenth and early twentieth centuries, and the strategies that workers developed to manage uncertainty, especially in old age.

My research uses a wide variety of archival sources, many previously unknown, from 19 Oxford colleges to consider why these colleges decided to overhaul servants’ pension provisions during this period, how retirement savings schemes were designed and implemented, and to try and understand what workers thought of these fundamental changes to the labour contract.

During this period, Oxford was a highly seasonal, low-waged economy. It was hard for many people to find enough work during the year to earn an adequate living, much less save for an old age they usually did not expect to see. Most men and women worked as long as they were capable, often past what we think of as a typical retirement age today.

It’s no surprise then that the protections against illness, disability, old age and death offered by these paternalistic employers encouraged a highly competitive labour market for college work, and the promise of an ex gratia, or traditional non-contributory pension, was one of the most attractive features of college employment.

For centuries, colleges awarded these traditional pensions to workers. Rights to these pensions, which usually replaced about a quarter to a third of a worker’s total earnings, were insecure and awards were made entirely at the discretion of the college.

In 1852, the first retirement savings scheme for Oxford college servants was created at Balliol College. By the 1920s, traditional non-contributory pensions had been replaced by contributory schemes at most Oxford colleges, shifting the risk of old age from employers to employees. Even though making contributions often meant a decrease in take-home pay, servants always preferred a guaranteed pension entitlement over traditional non-contributory pensions.

The earliest savings schemes mandated the purchase of life insurance policies. These were intended not only to protect a servant’s dependent family members, but also to limit the college’s financial liability in the event of a servant’s death. Servants were similarly risk-averse and often purchased multiple policies when they could afford to; many joined friendly societies and purchased insurance privately, in addition to employer-directed schemes.

The popularity of these schemes among Oxford colleges mirrors the growth of the insurance industry and the development of actuarial science during this period. By the 1870s, nearly all schemes included annuities or endowment assurance policies, which provided a guaranteed income for servants, usually at age 60-65, and facilitated the introduction of mandatory retirement ages for these workers.

Traditional paternalism remained influential throughout the period. Colleges insisted on controlling insurance policies, naming themselves as beneficiaries and directing the proceeds. Women, who were more likely to be in low-waged seasonal work, were nearly always excluded from these schemes and had to depend on ex gratia pension awards much longer than their male colleagues.

These early pension schemes offered no protection against inflation and colleges were usually slow to increase pension awards in response to rising prices. By the end of the Great War, dissatisfaction with inadequate pensions was one of several factors that pushed college servants to form a trade union in 1919.