## Matteo Cervellati and Uwe Sunde

Print publication date: 2017

Print ISBN-13: 9780262036627

Published to MIT Press Scholarship Online: May 2018

DOI: 10.7551/mitpress/9780262036627.001.0001

Show Summary Details
Page of

PRINTED FROM MIT PRESS SCHOLARSHIP ONLINE (www.mitpress.universitypressscholarship.com). (c) Copyright The MIT Press, 2018. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in MITSO for personal use. Subscriber: null; date: 24 October 2019

# Land Inequality, Education, and Marriage: Empirical Evidence from Nineteenth-Century Prussia

Chapter:
(p.183) 8 Land Inequality, Education, and Marriage: Empirical Evidence from Nineteenth-Century Prussia
Source:
Demographic Change and Long-Run Development
Publisher:
The MIT Press
DOI:10.7551/mitpress/9780262036627.003.0008

# Abstract and Keywords

This chapter offers an account of how institutional factors affect marriage, fertility, and education decisions. It reveals that higher landownership concentration was associated with lower enrollment rates as well as a negative relationship between marriage and enrollment rates, which suggests that political and economic inequality might be an important cofactor of the demographic and economic transitions. According to the study, the initial existence (or formation)of an economic elite will lead to the concentration of political power in the hands of few, which will in turn lead to the introduction of political institutions designed to sustain inequality. Such societies fail to adopt redistributive policies that allow for an optimal investment in physical and human capital. Therefore, in the long run, elite-based societies will not develop institutions conducive to sustainable economic growth.

# 8.1 Introduction

A vast literature documents the effects of economic and political inequality on development and growth. A significant part of this literature focuses on the political economy channel and argues the importance of the initial distribution of wealth for the distribution of political power. According to this literature, the initial existence (or formation) of an economic elite will lead to the concentration of political power in the hands of few, which will in turn lead to the introduction of political institutions designed to sustain inequality (see Acemoglu, Johnson, and Robinson 2001, 2002; Engerman and Sokoloff 1997; Sokoloff and Engerman 2000). Such societies fail to adopt redistributive policies that allow for an optimal investment in physical and human capital (see Alesina and Rodrik 1994; Persson and Tabellini 1994). Therefore, in the long run, elite-based societies will not develop institutions conducive to sustainable economic growth and thus will not match the income levels of economies with lower initial levels of inequality.

Elites generally oppose redistribution, using their political power to block the extension of voting rights and the expansion of school financing for mass education (see Acemoglu and Robinson 2006; Gallego 2010; Galor, Moav, and Vollrath 2009; Go and Lindert 2010; Naidu 2012; Ramcharan 2010). In a recent article, Cinnirella and Hornung (2016) propose an alternative mechanism based on the labor relations between the (landed) elite and the masses. Due to the decentralized (p.184) character of public schooling in Prussia (Lindert 2004), elites were not in the position to delay the expansion of mass schooling through the political channel. Yet, due to the prevailing institutions of the feudal system, the nobility exercised a direct authority over policing and jurisdiction and was patron over churches and schools in rural villages. Similarly, labor relations characterized by the coercion of labor lingered long into the nineteenth century. Thus, their position of power allowed them to directly interfere with the educational decisions of the peasantry.

Cinnirella and Hornung (2016) provide evidence for such exercise of authority by documenting the relationship between landownership concentration and peasants’ investments in education. They show that the regional concentration of noble large landowners is associated with lower enrollment rates in mass primary schooling. Furthermore, due to the introduction of agricultural reforms that eroded the authority of noble landowners, peasants gradually emancipated and increasingly enrolled in primary schools. Throughout the nineteenth century, the negative effect of the nobility on education vanishes. Further findings indicate that the nobility did not limit the provision of public schooling by restricting the supply of teachers and schools. Instead, the nobility restricted the demand for schooling through the coercion of labor services that prevailed even after serfdom was formally ended.

This chapter revisits the previous findings of a detrimental effect of landownership concentration on enrollment rates in nineteenth-century Prussia. It further proceeds to analyze a second aspect of possible interference of the landed elite with peasant’s individual life—the decision to get married. Until the agricultural reforms of the beginning of the nineteenth century, peasants were not allowed to get married without the consent of the noble landowner, which might have affected marriage patterns across Prussia in an unknown direction. It is likely that the landed elite favored the formation of families to increase the labor force bounded to land. On the other hand, the landed elite likely were interested in preserving a sufficient number of unmarried servants. Since celibacy and age at marriage are the most important determinants of fertility (Hajnal 1965; Voigtländer and Voth 2013), the potential relationship between land inequality and marriage is of great interest.

We expand the unique dataset compiled by Cinnirella and Hornung (2016) to include information about female marriage rates and sex ratios to estimate the relationship of interest throughout the nineteenth (p.185) century. This dataset further includes the distribution of landownership by size, primary school enrollment rates, and a range of development indicators (see also Becker et al. 2014). Using cross-sectional as well as panel analysis with county-fixed effects, we find no systematic evidence for the hypothesis that noble landowners directly interfered with the marriage decision.

However, we find a robust negative association of education with the share of married women in a panel analysis with county- and time-fixed effects. Thus, in regions where the authority of the landed nobility decreased and the demand for education increased, a smaller share of women got married. This finding aligns with the recent theoretical and empirical literature on the role of gender-specific human capital in the demographic transition (de la Croix and Donckt 2010; Galor and Weil 1996; Iyigun and Walsh 2007; Lagerlöf 2003). Educated women entering the industrial labor force earned higher wages and gained independence of male support. Thus, (female) investments in education changed the marriage pattern by postponing the age of marriage, increasing celibacy, and consequently reducing fertility over the course of the nineteenth century. By analyzing marriage patterns, this chapter provides preliminary evidence for an additional mechanism through which education could have affected the demographic transition in Europe.

The chapter is structured as follows: in section 8.2, we discuss some theoretical aspects of human capital formation and marriage patterns; section 8.3 reviews the empirical literature on inequality and education; section 8.4 provides a historical background regarding the Prussian agrarian reforms, school financing, and marriage patterns; section 8.5 describes the data; section 8.6 presents the empirical analysis of the relationship between landownership concentration and education; section 8.7 presents preliminary evidence on the association of landownership and education with marriage; and section 8.8 concludes.

# 8.2 Theory

## 8.2.1 Accumulation of Human Capital

Recent Theoretical Growth Models Assume A Complementarity Between Physical And Human Capital To Explain The Increasing Role Of Education In The Second Industrial Revolution. The Accumulation Of Physical Capital In The Process Of Development Increases The Importance Of Skills (p.186) And Human Capital In Production. This Escalation Creates A Conflict Of Interest Between Owners Of Capital And Land, Factors Of Production That Are Characterized By Different Complementarities With Human Capital (Galor And Moav 2006).

Galor, Moav, and Vollrath (2009) provide a theoretical framework that links landownership inequality with investments in public education. The model suggests that inequality in the distribution of landownership adversely affects the emergence of human capital-promoting institutions, thus affecting the transition process toward an industrial economy and the path to modern economic growth. Because of complementarities between physical capital and skills, capitalists strive for an educated labor force and support policies that promote public education. Conversely, the landowning elite are interested in reducing the mobility of the rural labor force and thus oppose policies favoring mass education. Due to a lower degree of complementarity between human capital and land, a rise in the level of education would push the rural labor force out of agriculture, thus reducing the return to land because of labor migration and higher agricultural wages. Large landowners have therefore fewer incentives to support public schooling for the masses.

If the political process is dominated by large landowners, they might prevent the implementation of public policies such as the expansion of public education. In fact, the model proposed by Galor, Moav, and Vollrath suggests that the negative impact of the landed elite on education increases with the concentration of landownership. This is a sensible prediction because there is substantial historical evidence supporting the notion that in preindustrial Europe, political power increased proportionally with the possession of land. The model also predicts that the landowners will stop opposing public policies in favor of education when their stakes in the industrialization process are large enough to reap the benefits of a higher level of skills.

The negative impact of landownership concentration on education also diminishes if the political power of the landed elite is eroded by institutional reforms. Galor and Moav (2006) suggest that capital-skill complementarity and the resulting conflict of interest between industrialists and agriculturalists in the provision of public education planted the seeds for the demise of the class structure. Therefore, a shift in the balance of power toward the owners of physical capital and the passing of land and educational reforms can be interpreted in the light of such a theory. In this vein, Cinnirella and Hornung (2016) show that agricultural reforms and the gradual emancipation of the peasants (p.187) during nineteenth-century Prussia changed the underlying relationship between landownership inequality and education.

Lindert (2004) describes the factors that foster and retard the expansion of primary and secondary education across the world. According to Lindert, elite self-interest, democracy, and decentralization are the crucial factors that define the success or failure of public policies on education. Similar to Galor and Moav (2006) and Galor, Moav, and Vollrath (2009), Lindert argues that powerful landed elites, such as the Tories in England, opposed the spread of education to the lower strata of the society (Lindert 2004, 100). However, regarding the role of capitalists, Lindert considers the social control hypothesis rather than the capital-skill complementarity. According to the former hypothesis, industrialists favor a centralized and mandatory school system to create a disciplined and obedient labor force, which is important for the factory system (Lindert 2004, 101). Both hypothesis, though based on different premises, predict a positive relationship between industrial development and the spread of mandatory primary education.

According to Lindert, the emergence of mass primary education depends strongly on the level of decentralization of the political process, which leads to the provision of public schooling. In particular, Lindert argues that local control of school finances could exert a positive effect on primary schooling in the context of early development. Decentralization of government allows the local population to vote for local taxes and schools. Therefore, the early leaders in education were those countries in which local governments were free to choose their optimal level of (tax-based) public expenditure on education. Lindert thus reinterprets the nineteenth-century success of German (Prussian) education, ascribing the high level of education to the high level of political decentralization of educational policy. In his view, the German educational leadership during the nineteenth century was the result of a localized, bottom-up process rather than of a process controlled by the political elite in Berlin (Lindert 2004, 115).

## 8.2.2 Human Capital, Fertility, and Marriage

In the second part of this chapter, we investigate how landownership concentration is related to female marriage rates. To our knowledge, economic theory has not directly linked land inequality to female marriage rates. Since the political channel cannot play a role in individual marriages decisions, such a connection seems debatable. However, Cinnirella and Hornung (2016) argue that land inequality determined the (p.188) individual demand for schooling. Here we discuss whether land inequality might also play a role in determining individual marriage decisions. This section presents a range of theories that explain marriage patterns.

One might argue that both land inequality and marriage rates are driven by some underlying mechanism, such as changes in the land-labor ratio. Late age at marriage and a relatively high celibacy rate define the so-called European Marriage Pattern (henceforth EMP) (Hajnal 1965). The EMP emerged as early as the fourteenth century in Western Europe (west of a line from St. Petersburg to Trieste) and contributed to limiting fertility by avoiding 25% to 40% of all possible births. Voigtländer and Voth (2013) argue that the Black Death played a crucial role in the rise of the EMP. According to their theory, the Black Death raised the land-labor ratio favoring animal husbandry, a sector in which women had a comparative advantage. Thus, after the Black Death, female employment opportunities in husbandry improved, which, in turn, provided an incentive to remain celibate because pregnancy and marriage resulted in termination of employment. The distinct cleavage between the EMP and the Eastern EMP alongside the so-called Hajnal line resembles the European border between regions that abolished serfdom comparatively early and regions that embraced the second serfdom—both divergences arose after the Black Death epidemics of the fourteenth century. Similar to the development of the EMP, the Black Death raised the land-labor ratio and changed the bargaining power between the serfs and the nobility, leading to the disappearance of serfdom in Western Europe and to an intensification of serfdom in Eastern Europe. As exemplified with the consequences of the Black Death during the fourteenth century, changes in the land-labor ratio might affect changes in serfdom institutions and marriage patterns and therefore qualify for the underlying mechanism that links landownership concentration and female marriage rates.

Alternative explanations for the EMP include De Moor and Van Zanden (2010), who argue that the pattern emerged as a result of the preaching of the Catholic Church promoting marriage based on consensus. In addition, they consider access to urban labor markets and the system of intergenerational transfers to explain the emergence of the EMP. Foreman-Peck (2011) develops a model that relates the EMP to investments in human capital and subsequent growth. In his model, later marriage raised the level of education of women because they had more time to acquire human capital before childbearing and child-rearing, (p.189) which in turn translated into higher income levels. Based on a review of the historical demography literature, Dennison and Ogilvie (2014) argue that there is no evidence of a relationship between the characteristics of the EMP and economic growth.

In this chapter, we present evidence for the lack of a direct link between landownership concentration and female marriage rates and for a significant negative association between the share of marriage rates and investments in education. The demand for skilled labor during the Industrial Revolution of the nineteenth century increased the female labor force participation (henceforth FLFP) and thus lowered female marriage rates. Galor and Weil (1996) model the relationship between the gender wage gap, fertility, and modern economic growth. In their model, an increase in capital per worker raises the relative wage of women as female labor is more complementary to capital than male labor. An increase in the female relative wage raises the opportunity cost of children by proportionally more than the couple’s full income, thus lowering fertility. A crucial element of their theory is that an increase in the capital intensity of the economy raises the relative wages of women. This is based on the notion that in the process of development, the economy rewards those characteristics more in which women have a comparative advantage.

In the spirit of Unified Growth Theory (Galor 2011), Diebolt and Perrin (2013a, 2013b) propose a model that emphasizes the role of women in the development process. In particular, they stress the role of gender equality in education and occupation as a crucial determinant of the demographic transition. According to their theory, once technological progress is sufficiently high and gender equality reaches a given threshold, women will enjoy the benefits of high returns to education. Higher investments in education by women will increase their bargaining power within the household, which, in turn, will allow them to enter the labor market. Investing in education will increase the opportunity cost of children for women, and they will have fewer children. The role of women in the model thus offers an alternative explanation of the demographic transition and can account for the reversal in the relationship between income and population growth.

We provide evidence that landownership concentration in nineteenth-century Prussia is not directly related to the share of married women. Yet, we provide robust evidence showing that increases in enrollment rates during the century are negatively associated with marriage rates.1 This preliminary result is consistent with the theory that (p.190) higher levels of development increased the returns to education and the employment opportunities for women who, in turn, married later or to a lesser extent, thus increasing the celibacy rate. These preliminary findings are broadly consistent with the theories on gender equality and growth of de la Croix and Donckt (2010); Diebolt and Perrin (2013b); Galor and Weil (1996); Iyigun and Walsh (2007); and Lagerlöf (2003).

# 8.3 Literature Review

There is a large empirical literature that analyzes the economic consequences of inequality for human capital. One strand of literature focuses on redistribution. Alesina and Rodrik (1994) and Persson and Tabellini (1994) hypothesize that in a more egalitarian society, taxation of physical and human capital is lower, enhancing economic growth. Ramcharan (2010) studies the relationship between land inequality and education expenditure in the United States for the period from 1890 to 1930. The paper shows that land inequality is negatively associated with redistribution and therefore with less expenditure on education. The issue of causality is addressed using geographic variables such as surface elevation, rainfall, and crop choice to identify exogenous variation in land inequality.2

The theoretical model of Galor, Moav, and Vollrath (2009) focuses on the low complementarity between agriculture and human capital and predicts an adverse effect of the concentration of landownership on education expenditure. The authors test this prediction for the United States for the period from 1900 to 1940 using variation in the distribution of landownership and education expenditure across states and over time. Indeed, they find that land inequality has a negative impact on education expenditures. They address causality through an instrumental variable approach. In particular, they identify variation in landownership inequality through the interaction between nationwide changes in the relative price of agricultural crops that are associated with economies of scale and variation in climatic characteristics across states. Kourtellos, Stylianou, and Tan (2013), by employing duration models in a sample of fifty-three countries, find that higher levels of land inequality delay the expansion of schooling. Their findings are consistent with the theory proposed by Galor, Moav, and Vollrath.

Go and Lindert (2010) stress the political-economy channel in the provision of public education in the United States. In particular, in explaining the gap in enrollment rates between the north and the south (p.191) in 1850, they point to local governments having more autonomy and the population having a more equally distributed political voice in the north. Among other things, they find that extending the voting power to lower-income groups raises the taxes paid for schooling and thus the primary school enrollment rates. Vollrath (2013) finds that landownership inequality predicts taxes for local school funding at the U.S. county level in 1890.

The so-called Engerman and Sokoloff hypothesis (Engerman and Sokoloff 1997) suggests that initial factor endowments create differences in economic inequality, which, in turn, generate low-quality institutions that induce low investments in human capital and cause differential development paths. Engerman and Sokoloff argue that the geography of Latin America, characterized by large economies of scale, led to the use of slave labor, which created high levels of inequality. In contrast, the natural endowments of North America led to smaller-scale family farms, generating less inequality and thus promoting the growth of a large middle class. Easterly (2007) tests the hypothesis of Engerman and Sokoloff and finds that inequality has a significant negative long-run effect on per capita income (in 2002), institutions, and secondary school enrollment rates for the period from 1998 to 2002 (Easterly 2007, 766, table 4). Easterly uses the soil suitability for the cultivation of sugar versus wheat to identify variation in inequality. Nunn (2008), when testing the Engerman and Sokoloff hypothesis, finds that slavery is associated with underdevelopment. Yet his analysis suggests that the detrimental effect of slavery on development does not work through its impact on initial economic inequality. In an attempt to reject geographic determinism, Clark and Gray (2014) argue that large-scale farming in south-east England in the nineteenth century is not associated with lower levels of education measured in terms of literacy rates.

Finally, another strand of literature considers variations in the levels of human capital as an outcome of institutions (Acemoglu, Johnson, and Robinson 2001, 2002; Acemoglu, Gallego, and Robinson 2014). According to this literature, countries that develop extractive institutions are generally characterized by the concentration of political power in the hands of a small elite and by a vast majority of population without effective property rights. On the contrary, countries with more inclusive political institutions tend to have economic institutions that are more conducive to economic growth. By estimating the correlation between schooling and literacy rates with inequality in political power, Mariscal and Sokoloff (2000) show to what extent the extension of the franchise (p.192) in Latin America increased schooling. The relationship between the extension of the franchise and schooling has also been analyzed theoretically in Acemoglu and Robinson (2000). Galor (2011) explores the role of historical variables and political institutions to explain differences in schooling in former colonies. He argues that the degree of democratization positively affects primary education, whereas decentralization of political power is more related to differences in higher levels of schooling, such as secondary and higher education. Dell (2010), instead, analyzes the persistent effect of coercive labor institutions on human capital.

# 8.4 Historical Background

## 8.4.1 Noble Authority and Agricultural Reforms

A distinguishing feature of the old Prussia and the western territory gained at the Congress of Vienna is the organization of agriculture.3 The river Elbe is the geographical reference point generally used to distinguish the two agricultural regimes:4 The dominant regime in the western provinces was Grundherrschaft and the dominant regime in the eastern province was Gutswirtschaft. In the western system of Grundherrschaft, the nobility owned the land but did not participate in its cultivation and relied on cash rents paid by the tenants. In the Eastelbian system of Gutswirtschaft, noble lords operated large parts of the estate and relied on labor services provided by serf peasants. Due to the close, personal relationship between noble landowners and peasants in the Eastelbian system, the exertion of noble authority and legal control was more prevalent than in the western system.

Control and authority of noble landlords over the peasantry included policing, jurisdiction (Patrimonialgerichtsbarkeit), tax collection, and the appointment of priests and schoolteachers (Patronatsrecht) (Bowman 1980; Eddie 2013). As patron of the estate, the noble landowner was responsible for the education and religiosity of the individuals. He appointed the schoolmaster and provided the instructions under which the school was to operate. In some cases, estate owners even prescribed the content of the school curriculum (Berdahl 1988, 55–63).5 From the age of fourteen, children could be drafted for compulsory domestic service (Gesindezwangsdienst) at the manor. Furthermore, the noble landowner had the ability and power to interfere with a range of individual decisions, such as marriage, land transfer, conflict resolution, (p.193) and migration. Ogilvie (2005) argues that noble intervention in individual decisions was highly effective because it needed to be exercised rarely. In return for their services, landowners had the obligation to protect their subjects and to provide relief in cases of need, such as accidents and bad harvests. Benefits for the serfs further included the rights to the commons, including the grazing rights to the commons, and might include the right to fish and cut wood.

Noble authority and personal subjection started to resolve toward the end of the eighteenth century, and peasant emancipation gained considerable momentum after the agricultural reforms of the early nineteenth century. The Prussian king aimed to inspire nobles by abolishing serfdom on the royal domains in 1799. After losing the war against the French Revolutionary Army, the Prussian Reforms (Stein-Hardenberg Reforms) enacted by a small group of enlightened bureaucrats introduced a series of changes of agricultural institutions. The October Edict of 1807 abolished the legal institution of servitude starting from November 11 (St. Martin’s Day), 1810. In 1811, another edict clarified how former serfs could acquire property and become owners of their plots. Due to the opposition of the nobility, in 1816, the scope of the edict was restricted and excluded peasants on small plots (those that did not use drought animals to cultivate land). Consequently, the nobility maintained a de facto authority over two-thirds of the peasantry that continued to provide mandatory labor services (Harnisch 1984). The legal process at the local level remained under the control of the landed nobility until the constitution of 1849 changed the system of noble mediation. From 1850, all peasants gained complete legal emancipation and were able to discard and redeem labor services (Pierenkemper and Tilly 2004; Bowman 1980).

In the western Prussian regions of Rhineland and Westphalia, most noble prerogatives were abolished during the French occupation period between 1794 and 1814. Serfdom was abolished de jure in the Rhineland in 1805 and in Westphalia in 1808. However, Westphalian nobles tried to block many of the changes introduced by the Napoleonic Code. While the Napoleonic Code remained effective after 1815 in the areas left of the river Rhine, other areas formally (re-)established Prussian law (Allgemeines Landrecht).6

## 8.4.2 Education and School Financing

According to Schleunes (1979, 317), the war against France triggered a “decade of feverish activity” leading to a first surge in primary (p.194)

Figure 8.1 Development of the primary school enrollment rate in Prussia, 1816–1900.

Source: Prussian census data.

schooling (see figure 8.1). Prussia is well-known to be the world leader in education during the nineteenth century, especially in primary education. The roots of mass schooling can be traced back to Martin Luther’s call for establishing Protestant schools in rural areas (see Becker and Woessmann 2009). Also, the Prussian king pushed for compulsory education first in 1717 and again in 1763. However, the state did not provide financing, and the king’s edicts were attempts to persuade the landed nobility to provide education at their own expenses (Lindert 2003, 2004). Thus, it took until the defeat of the Prussian Army by the French Empire in 1806 for mass education to move into the public focus in Prussia.

According to Lindert (2004), schooling had never been of special national interest and was left to the local leaders. He interprets Prussian educational history as a bottom-up rather than a top-down story. In fact, resources for school funding derived from a range of local sources. A large share of the expenses was provided by assets held by the local school such as real estate, entitlements or capital rents. Similarly, teacher salaries were derived from a range of sources including tuition fees (17%), school assets (11%), local school taxes (60%), and state funds (12%).7 So-called schooling societies, bodies of the municipality (p.195) received funds from the school tax. Taxpayers were the heads of each household (excluding noble landowners) proportional to their wealth and income, independent of their religious denominations and number of children.

Tuition fees were usually charged if society funds were insufficient. Noble landowners were to contribute to school financing only if local families could not afford to pay for schooling (Kuhlemann 1991, 181). In addition, poor school districts received financial support from the state (Berlin 1889, 58–59). Tuition fees were abolished in 1888, and the exemption of noble estate owners from school financing and school taxes was abolished in 1906.

The Prussian Statistical Office provides comparative information on school financing for the period from 1861 to 1886. Table 8.1 presents an overview of total per capita school funds at the provincial level. Similar to Lindert (2003, 2004), we find that eastern provinces had relatively lower school funds compared to western provinces. For example, primary school funds amounted to about 1.4 marks per capita in East and West Prussia in 1861,8 whereas it amounted to 1.8 marks in the western province of Rhineland. However, it is important to note that this difference may reflect differentials in the cost of living. In fact, if we deflate school funds per capita with rye prices in 1861, funds in the provinces of East and West Prussia are 10% and 14% higher than in Rhineland, respectively.

Table 8.1 Total Expenditure for Public Primary Schools, 1861–1886

1861

1864

1867

1871

1878

1886

East Prussia

1.414

1.469

1.723

1.781

3.053

3.637

West Prussia

1.458

1.603

1.798

1.915

3.124

3.673

Poznan

1.190

1.253

1.399

1.695

2.687

3.508

Silesia

1.276

1.337

1.517

1.716

2.792

3.488

Pomerania

1.670

1.727

1.945

2.305

3.668

4.533

Brandenburg

1.963

1.953

2.109

2.468

3.515

4.220

Saxony

2.079

2.124

2.233

2.536

3.727

4.678

Westphalia

1.559

1.597

1.812

2.110

3.840

4.742

Rhineland

1.883

1.950

2.124

2.603

4.664

4.991

Note: Total expenditure for public primary schools in German marks per capita at the provincial level.

Source: Own calculations according to Königliches Statistisches Bureau in Berlin (1889).

(p.196)

Table 8.2 State Expenditures for Public Primary Schools, 1861–1886

1861

1864

1867

1871

1878

1886

East Prussia

0.079

0.077

0.095

0.142

0.636

0.742

West Prussia

0.103

0.096

0.111

0.169

0.499

0.667

Poznan

0.085

0.080

0.103

0.141

0.638

0.798

Silesia

0.035

0.038

0.050

0.074

0.325

0.424

Pomerania

0.063

0.052

0.060

0.119

0.807

0.908

Brandenburg

0.111

0.083

0.106

0.129

0.461

0.501

Saxony

0.090

0.068

0.079

0.109

0.356

0.378

Westphalia

0.058

0.044

0.069

0.096

0.392

0.376

Rhineland

0.039

0.041

0.058

0.087

0.397

0.410

Note: State expenditures for public primary schools in German marks per capita at the provincial level.

Source: Own calculations according to Königliches Statistisches Bureau in Berlin (1889).

Consistent with the finding that deflated funds did not systematically differ across Prussia, Cinnirella and Hornung (2016) observe no appreciable difference between the East and the West when looking at the number of schools and teachers per child at school age. A further subdivision of the sources of school funding in table 8.2 shows that, throughout the period considered, less-developed eastern provinces such as East Prussia, West Prussia, and Poznan benefited to a larger extent from state contributions than did the western provinces of Westphalia and Rhineland. On the contrary, school funds in eastern provinces relied to a much less extent on tuition fees. In 1861, in the provinces of Rhineland and Westphalia, tuition fees accounted for 24% and 27% of the total school funds, respectively; instead, in East and West Prussia tuition fees accounted for only 11% and 13% of the total school funds, respectively.

Therefore, the lower (nominal) level of financial support in the rural eastern areas arguably impacted neither the availability of schools nor that of teachers.9 Supply of education was not restrained in the eastern regions. As we argue in this chapter, it was the noble authority that affected the private demand for education, which explains the delay in educational attainment of the agricultural eastern regions of Prussia to a large extent.

## 8.4.3 Marriage Patterns

In Prussia, the decision to marry was effectively influenced by institutions prevailing from the feudal system and the church as well as (p.197) individual incomes. The marriage pattern in Catholic regions is distinctively different from Protestant regions; with a higher propensity to be married in Protestant regions. Feudal institutions provided that marriage was allowed only with the consent of the noble landlord. Marriage decisions were also determined by labor relations and therefore not possible for domestic workers and servants (Gesinde) usually accommodated within the household of their employers.10 Consequently, the serf institution of compulsory domestic service for the nobility might have led to a lower share of marriages.11 The share of females that provided domestic service to the nobility was 1.8% in Prussia east of the Elbe and 0.8% in Prussia west of the Elbe in 1816. This pattern seems to remain relatively stable over time; in 1849, it was 2.0% in the east and 1.0% in the west. Therefore, variation in the concentration of noble landowners across Prussia might affect differential marriage patterns.

According to Eddie (2013), noble landowners were obligated to support their peasants in lean times. Since female peasants from outside of the village became the concern of the landowners upon marriage, nobles might have paid attention to the economic circumstances of the marriage. Yet, Ogilvie (2005, 98) finds that noble lords rarely interfered with serf marriages. Of the 111 requests for permission to marry that were denied and brought to the noble court on the Bohemian estate of Friedland, only 18% were ultimately refused. The reason to refuse the request was usually that the female serf was subject to another lord, creating incentives for the male serf to run away and indicating that the nobility might have decreased marriage rates. On the other hand, individual incomes might affect marriage decisions. Upon marriage, peasants were expected to establish a new household on the estate of the lord. Running an independent household or a farm was considered infeasible without a wife. Adult children of serfs would often wait to get married until they inherited the land of their parents to have a sufficient income. This constraint might have postponed marriage to later stages in life. According to Ogilvie (2005), the nobility might thus have ordered male peasants to marry and thus increased marriage rates. With the agricultural reforms of 1807, the power and authority of noble landowners over the peasant population eroded. Fees that had to be paid to the noble landowner upon marriage were abolished. Peasants were able to choose to get married without the consent of the landowner. This allowance might have affected the propensity to get married in an unknown direction.

During the nineteenth century, the Industrial Revolution created a range of new employment opportunities that affected individual (p.198) incomes and provided independence from the necessity to inherit land. Consequently, the share of married women across Prussia might have increased. However, the effect would be reversed if an increasing number of women entered the industrial labor force and earned an independent income. According to Wrigley (2006), women in regions with a large textile industry postponed marriage much more than in regions with a large mining and metals industry. Wrigley assumes that part of the effect operates through sex ratios, leading to a high ratio of women to men in textile regions and a high ratio of men to women in mining regions, resulting in a higher share of the population having trouble finding an adequate spouse. Data on marriage rates and female labor force participation across Prussian counties in 1882 support the hypothesis that employment opportunities for women are associatedwith a lower propensity to marry. In figure 8.2, we plot the share of married women in 1885 against female labor force participation in 1882. The two variables are significantly negatively correlated. In a following section, we show that landownership concentration is, in fact, not related to marriage rates, whereas enrollment rates show a significant, robust negative relationship with marriage rates.

Figure 8.2 Female marriage and female labor force participation rate.

Source: Prussian census data.

# (p.199) 8.5 Data

Our analysis of the relationship between landownership, education, and marriage rates is based on county-level data from nineteenth-century Prussia. Our dataset includes data from five periods (1816, 1849, 1864, 1886, and 1896) collected from various sources published on behalf of the Royal Prussian Statistical Bureau in Berlin. This section describes those variables that are particularly important for this project: information on land ownership structure by size, enrollment rates in primary schools, and the marital status and age structure of the female population. Control variables such as population size, urbanization, religious denomination, and industrial structure are also included in the regression analysis.12

The main variable of interest, public primary school enrollment rates, refers to school attendance of the six to fourteen year olds. Consistent with the definition of mandatory schooling at the time, we consider both elementary schools (Elementarschulen) and middle schools (Mittelschulen) as primary schools.13

For the analysis of marriage patterns, we combine a range of data from different population censuses. We define the share of married women as the total number of married women over the total number of women older than fifteen years. It is important to note that we measure marriage rates at the same time or close to all other explanatory variables in our five cross sections. Unfortunately, county-level information on married women is unavailable during the period from 1850 to 1870. For the cross section labelled 1864, we thus construct the share of married women using data for 1871, while we include enrollment rates for 1864 and landownership concentration for 1858.

The Prussian censuses report the number of landholdings per county and classify them into size bins. The first full census in 1816 classifies land holdings into three groups: properties or leasehold estates of up to 15 Prussian morgen (henceforth PM), from 15 to 300 PM, and more than 300 PM.14 This categorization reflects the contemporary agricultural structure of farming. Farmers with less than 15 PM usually required some additional form of nonfarming income. Landholdings between 15 and 300 PM were generally large enough for the subsistence of a family, whereas farms with more than 300 PM were usually cultivated by paid laborers and coerced labor, while the owner was not expected to perform any manual work (Harnisch 1984). The 1849 census classifies landownership into 5 groups: up to 5 PM, 5 to 30 PM, 30 to 300 PM, (p.200) 300 to 600 PM, and more than 600 PM. The agricultural census in 1855 follows the same structure. From 1882 onward, the census considers only arable land, and the unit of measurement changes from PM to hectare (ha). In particular, the occupation census in 1882 classifies farms into six groups: farms with arable land up to 1 ha, 1 to 2 ha, 2 to 10 ha, 10 to 50 ha, 50 to 100 ha, and more than 100 ha. Finally, the occupation census in 1895 classifies farms into 7 groups: up to 0.5 ha, 0.5 to 2 ha, 2 to 5 ha, 5 to 20 ha, 20 to 100 ha, and more than 100 ha. Our measure for the concentration of large landownership is based on the share of farms larger than 300 PM or 100 ha, respectively.

Our rich dataset allows us to account for several potentially confounding factors. In particular, our analysis includes control variables for religious affiliation, urbanization, the share of people employed in industry, the share of people employed in agriculture, the child dependency ratio (defined as the ratio of people aged 0 to 14/18 over the working population between 15/19 and 65/70 years), population density, and school density. We also include a binary variable for the inheritance rule: the northeastern parts of Prussia are dominated by nonpartible inheritance (Anerbenrecht), while the southwestern parts are characterized by partible inheritance (Realteilung). These different rules likely affected the average size and the distribution of land holdings and might therefore influence our relationships of interest. We also include a control for the share of people whose first language was not German. Thus, we can account for the ethnic-linguistic heterogeneity of the population. Finally, we include a measure for the sex ratio, defined as the ratio of men over women, which is an important determinant of the female marriage rate.

Descriptive statistics of these variables for the five periods are presented in table 8.3. The increasing number of observations at the bottom of the table reflects the formation and acquisition of new territories over the course of the nineteenth century. One can immediately observe a high initial level of enrollment of 60% in 1816 that increased to more than 90% at the end of the nineteenth century. On the contrary, we observe a decrease in the average female marriage rate from 54% to 50%, which is a substantial drop given the slow-moving character of adult marriage rates. The descriptive statistics further show that the share of large landholdings increased during the first half of the nineteenth century, reaching a share of 2.5% in 1858. Because of the change in the unit of measurement and in the definition of large (p.201)

Table 8.3 Descriptive Statistics

VARIABLES

(1) 1816

(2) 1849

(3) 1864

(4) 1886

(5) 1896

Enrollment rate (6–14)

0.603

0.802

0.753

0.936

0.943

(0.195)

(0.117)

(0.104)

(0.077)

(0.076)

Large landholdings (share)

0.017

0.024

0.025

0.008

0.006

(0.021)

(0.027)

(0.026)

(0.008)

(0.007)

Married women (share)

0.544

0.541

0.511

0.502

0.504

(0.054)

(0.046)

(0.039)

(0.041)

(0.045)

Protestant (share)

0.616

0.605

0.600

0.642

0.645

(0.402)

(0.394)

(0.391)

(0.374)

(0.367)

Urban (share)

0.244

0.246

0.260

0.295

0.317

(0.182)

(0.186)

(0.194)

(0.242)

(0.275)

Industrial (share)

0.009

0.072

0.080

0.119

0.127

(0.023)

(0.039)

(0.048)

(0.058)

(0.062)

Agricultural (share)

0.088

0.550

0.186

0.196

0.188

(0.038)

(0.187)

(0.066)

(0.079)

(0.090)

Child-dependency ratio

0.631

0.646

0.602

0.894

0.773

(0.074)

(0.077)

(0.067)

(0.111)

(0.105)

Population density

0.760

1.774

2.236

4.135

5.237

(1.855)

(8.430)

(11.238)

(16.231)

(17.597)

School density

0.131

0.177

0.195

0.178

0.201

(0.263)

(0.730)

(0.756)

(0.336)

(0.360)

Inheritance (dummy)

0.246

0.245

0.245

0.253

0.255

(0.432)

(0.431)

(0.431)

(0.435)

(0.436)

Non-German speakers (share)

0.107

0.134

0.134

0.106

0.119

(0.237)

(0.259)

(0.259)

(0.242)

(0.245)

Sex ratio

0.962

0.992

0.966

0.962

0.968

(0.057)

(0.046)

(0.048)

(0.055)

(0.063)

Observations

272

335

335

463

549

Note: Standard deviation in parenthesis.

Source: See Cinnirella and Hornung (2016) for data sources and details.

landownership, we cannot directly compare the figures for the first three periods with those of the last two periods. In the regression analysis, we use standardized values for landownership concentration (mean zero and unit standard deviation) to eliminate biases due to the unit of measurement.

(p.202) The descriptive statistics document trends in urbanization as well as in industrialization and population density over the course of the century. The apparent strong changes in the share of people employed in agriculture over time can be explained by inconsistencies in the definition of agricultural laborers. Religious affiliation, linguistic heterogeneity, inheritance rule, and the sex ratio are rather time invariant.

# 8.6 Land Inequality and Education

## 8.6.1 Cross-Sectional Evidence

This section provides cross-sectional evidence for the relationship between landownership concentration and education. The relationship is estimated in five separate cross sections spanning the entire nineteenth century, for the years 1816, 1849, 1864, 1886, and 1896. Cinnirella and Hornung (2016) find a significant negative relationship between landownership concentration and education in these five cross sections. In their paper, the administrative borders are held constant, resembling the administrative structure in place around 1816 to allow for the highest possible comparability over time. However, throughout the century, Prussia acquired vast new territories in central Germany.15 These territories, west of the river Elbe, fill the gap between the eastern and the western part of Prussia and might add significant variation to the data.

Table 8.4 shows cross-sectional results using the actual county structure in place at the time of each of the censuses, including the newly acquired territories. We also control for a range of demand and supply factors that might have an effect on enrollment rates. The estimates indicate a significant negative association of the share of large landownership with the enrollment rate for each period. Since we standardize the share of large landownership with mean zero and unit standard deviation, we can directly compare the magnitude of the coefficients over time. The pattern of results suggests that the relationship decreases throughout the nineteenth century. A one standard deviation increase in landownership concentration (e.g., from the average value of 1.7% to 3.8% in 1816) translates into a 7 p.p. decrease in enrollment rates in 1816 but only to a 0.7 p.p. decrease in 1896. These findings align with the previous findings by Cinnirella and Hornung (2016) and confirm that the negative relationship between landownership concentration and education holds true when including the vast new territories in central Germany. (p.203)

Table 8.4 Land Concentration and Enrollment Rates—OLS Estimates

VARIABLES

(1) 1816

(2) 1849

(3) 1864

(4) 1886

(5) 1896

Share of large

0.070***

0.038***

0.032***

0.009***

0.007***

landholdings (std)

(0.010)

(0.007)

(0.005)

(0.002)

(0.002)

Protestant (share)

0.189***

0.055***

0.045***

0.016**

0.015**

(0.031)

(0.016)

(0.013)

(0.008)

(0.006)

Urban (share)

0.078

0.086**

0.043

0.133***

0.101***

(0.064)

(0.041)

(0.056)

(0.022)

(0.019)

Industrial (share)

0.440*

0.117

0.312

0.615***

0.497***

(0.232)

(0.142)

(0.201)

(0.120)

(0.078)

Agricultural (share)

0.232

0.022

0.320*

0.629***

0.489***

(0.274)

(0.036)

(0.167)

(0.107)

(0.060)

Child dependency ratio

0.347**

0.057

0.186**

0.182***

0.251***

(0.137)

(0.083)

(0.089)

(0.035)

(0.031)

Population density

0.065***

0.008***

0.004***

0.002***

0.001*

(0.014)

(0.002)

(0.001)

(0.001)

(0.001)

School density

0.422***

0.087***

0.046***

0.114***

0.081***

(0.090)

(0.014)

(0.010)

(0.032)

(0.027)

Inheritance (dummy)

0.003

0.021*

0.029**

0.000

0.002

(0.026)

(0.012)

(0.012)

(0.007)

(0.006)

Non-German speakers (share)

0.205***

0.158***

0.135***

0.038***

0.015

(0.059)

(0.028)

(0.024)

(0.012)

(0.010)

Observations

272

335

335

463

549

R-squared

0.46

0.38

0.40

0.59

0.61

Note: The table shows county-level OLS estimates for five separate cross sections. Land concentration is standardized with mean zero and unit standard deviation. Robust standard errors in parentheses. Significance:

(***) p < 0.01,

(**) p < 0.05,

(*) p < 0.1.

Source: See Cinnirella and Hornung (2016) for data sources and details.

This chapter does not aim to establish causality in the relationship under analysis. By using contemporary census information on the geological composition of the soil (i.e., the amount of sand, clay, and loam in the terrain [soil texture]), Cinnirella and Hornung (2016) identify exogenous variation in farm size and therefore in the concentration of large landownership. The identification strategy is based on the notion that differences in soil texture are related to differences in soil quality, which historically generated a heterogeneous demand for land. Thus, they provide two-stage, least-squares estimates of the relationship of interest and address the issue of causality. Instrumental variable (p.204) estimates confirm a significant negative effect of large landownership for the years 1816 and 1849, whereas the coefficients for the other periods are not statistically different from zero. Since the historical data on soil texture are not available for the Prussian territories acquired in the second half of the century, we refrain from providing instrumental variable estimates here.

## 8.6.2 Panel Analysis

The cross-sectional results are likely to be afflicted by unobserved heterogeneity at the county level. We thus proceed to analyze the relationship of interest using the panel structure of the data. Since data for the newly acquired territories are not available for the early periods, we proceed to aggregate the data to the administrative structure in place at the beginning of the period, similar to Cinnirella and Hornung (2016).16

We start by showing bivariate estimates in the pooled sample in column 1 of table 8.5; this shows the average relationship between landownership concentration and education throughout the nineteenth century. Column 2 excludes time-invariant unobserved heterogeneity by introducing a full set of county-and time-fixed effects. According to the results, a standard deviation increase in landownership concentration decreases enrollment rates by 2 percentage points. Controlling for various county characteristics in column 3 leaves the coefficient virtually unaffected.17

Although we account for time-invariant heterogeneity across counties, time-varying characteristics that affect both landownership and education might bias our fixed-effects estimates. In particular, if crops differ regarding their human capital intensity, time heterogeneity in the productivity of the soil could be an important omitted variable. In column 4, we address these concerns by including controls for land rent and crop yields.18 Since land rent is time-invariant and was determined only in 1865, we interact the variable with the different periods. Similarly, information on yields per hectare is available for the years 1886 and 1896. For the earlier periods (1816, 1849, and 1864), we compute the average across the available years 1886 and 1896 and interact it with the period dummies. The results indicate that such controls do not affect the relationship between land concentration and education.

Finally, in columns 5 and 6, we estimate our model for the counties west and east of the river Elbe, respectively. As already mentioned, the two parts differ regarding the extent of serfdom and the level of development. Indeed, our panel estimates indicate that the relationship (p.205)

Table 8.5 Land Concentration and Enrollment Rates—Panel Estimates

Dep. var.: Enrollment rate

(1) Pooled

(2) FE

(3) W/Controls

(4) W/Controls

(5) West

(6) East

Land concentration

–0.032***

–0.020**

–0.018**

–0.021***

0.030

–0.013*

(0.005)

(0.009)

(0.008)

(0.008)

(0.021)

(0.007)

Urban (share)

0.024

–0.012

–0.021

0.143

(0.064)

(0.071)

(0.057)

(0.197)

Protestant (share)

0.515***

0.483**

–0.193

1.028***

(0.196)

(0.200)

(0.168)

(0.254)

School density

0.023*

0.012

0.008**

2.629***

(0.012)

(0.009)

(0.004)

(0.654)

Industrial (share)

–0.345**

–0.183

0.494***

–0.876***

(0.159)

(0.153)

(0.149)

(0.229)

Agricultural (share)

0.013

0.003

–0.027

–0.047

(0.035)

(0.036)

(0.043)

(0.071)

Child-dependency ratio

0.263***

0.266***

–0.253***

0.437***

(0.075)

(0.078)

(0.075)

(0.111)

Non-German speakers (share)

0.248

0.366**

1.099***

0.358***

(0.171)

(0.156)

(0.271)

(0.127)

Time-fixed effects

No

Yes

Yes

Yes

Yes

Yes

County-fixed effects

No

Yes

Yes

Yes

Yes

Yes

GRE

No

No

No

Yes

Yes

Yes

Yields

No

No

No

Yes

Yes

Yes

Observations

1387

1387

1387

1357

615

742

Number of counties

280

280

277

127

150

Note: The table shows county-level panel estimates. Measures of land concentration are standardized with mean zero and unit standard deviation. Standard errors in parentheses are clustered at the county level. Thirteen observations drop out from the analysis because of missing information for the district of Cologne in the 1816 data. Significance:

(***) p < 0.01,

(**) p < 0.05,

(*) p < 0.1.

Source: See Cinnirella and Hornung (2016) for data sources and details.

(p.206) between landownership concentration and enrollment rates is negative and significant only for the eastern counties, where the authority of the landowners over the peasantry was more accentuated.

# 8.7 Land Inequality, Education, and Marriage

The previous section provides evidence for a negative relationship between landownership concentration and human capital formation. We have shown that the negative impact of large landownership on education can be detected when analyzing variation between counties as well as within counties.

## 8.7.1 Land Inequality and the Female Marriage Rate

In section 4.3, we discuss the potential role that large landowners might have played in peasants’ marriage decisions. In this section, we test whether landownership concentration is in fact associated with marriage patterns. We start by estimating cross-sectional models for each period:

(1)
$Display mathematics$

where y is the share of married women in county i, and Land is the share of large landowners.19 The subscript t refers to the period. The vector of covariates X includes urbanization, the share of Protestants, the sex ratio, employment in industry and agriculture, and the share of non-German speakers—variables that likely affect marriage rates independent of land concentration. The results are reported in table 8.6. As one can see, with the exception of the year 1864, the cross-sectional estimates do not show any significant relationship between landownership concentration and the share of married women. We find that the share of Protestants is positively related to female marriage rates, whereas urbanization presents a systematic negative correlation. The share of people employed in industry seems to be positively related to female marriage rates, especially in the second half of the nineteenth century.

The cross-sectional estimates can be severely affected by unobserved heterogeneity. Therefore we proceed to present panel estimates that account for time-invariant, county-specific factors that can be related to marriage patterns. We estimate the following model:

(2)
$Display mathematics$

where α‎i and τ‎t are county- and time-fixed effects, respectively. The panel estimates are presented in table 8.7. (p.207)

Table 8.6 Female Marriage and Land Concentration—OLS Estimates

VARIABLES

(1) 1816

(2) 1849

(3) 1864

(4) 1886

(5) 1896

Share of large

0.004

0.006**

0.001

0.000

0.002

landholdings (std)

(0.003)

(0.002)

(0.002)

(0.001)

(0.002)

Protestant (share)

0.042***

0.040***

0.058***

0.065***

0.074***

(0.008)

(0.006)

(0.005)

(0.005)

(0.005)

Urban (share)

0.085***

0.097***

0.069***

0.072***

0.073***

(0.016)

(0.019)

(0.012)

(0.013)

(0.009)

Industrial (share)

0.121*

0.036

0.147***

0.220***

0.257***

(0.069)

(0.050)

(0.052)

(0.059)

(0.049)

Agricultural (share)

0.120*

0.022

0.018

0.027

0.028

(0.066)

(0.016)

(0.040)

(0.056)

(0.039)

Population density

0.004***

0.001

0.001**

0.000

0.000

(0.001)

(0.001)

(0.000)

(0.000)

(0.000)

Sex ratio

0.161

0.167***

0.235***

0.259***

0.228***

(0.101)

(0.051)

(0.038)

(0.030)

(0.026)

Inheritance (dummy)

0.004

0.001

0.001

0.002

0.001

(0.007)

(0.005)

(0.004)

(0.004)

(0.004)

Non-German

0.123***

0.056***

0.081***

0.053***

0.077***

speakers (share)

(0.013)

(0.008)

(0.006)

(0.008)

(0.009)

Observations

272

335

335

463

549

R-squared

0.54

0.41

0.61

0.52

0.56

Note: The table shows county-level OLS estimates for five separate cross sections. Robust standard errors in parentheses. Significance:

(***) p < 0.01,

(**) p < 0.05,

(*) p < 0.1.

Source: See Cinnirella and Hornung (2016) for data sources and details.

(p.208) In column 1, when estimating the relationship using the five cross sections pooled, results show a positive coefficient. Column 2 introduces time- and county-fixed effects. When exploiting only the within-county variation in the data, the relationship between land concentration and the female marriage rate is insignificant and close to zero. The same result holds when including the set of control variables (column 3). Contrary to the cross-sectional estimates, within-county variation indicates that the share of Protestants is negatively related to the share of married women. That relation suggests the existence of a county-specific factor that affected both the share of Protestants and marriage rates. The coefficient for the sex ratio and employment in industry are positively related to female marriage rates.20

Table 8.7 Female Marriage and Land Concentration—Panel Estimates

Dep. var.: Share of married women

(1) Pooled

(2) FE

(3) W/Controls

(4) W/Controls

(5) West

(6) East

Land concentration

0.012***

–0.001

–0.001

0.001

–0.003

–0.001

(0.003)

(0.002)

(0.001)

(0.001)

(0.004)

(0.001)

Urban (share)

–0.066***

–0.057***

–0.074***

–0.046

(0.015)

(0.016)

(0.018)

(0.029)

Protestant (share)

–0.158***

–0.122***

–0.122**

–0.163***

(0.048)

(0.045)

(0.057)

(0.052)

Sex ratio

0.245***

0.213***

0.058

0.452***

(0.067)

(0.069)

(0.042)

(0.047)

Industrial (share)

0.278***

0.203***

0.226***

0.109***

(0.049)

(0.044)

(0.054)

(0.034)

Agricultural (share)

0.018

0.003

–0.028

0.026*

(0.015)

(0.015)

(0.021)

(0.015)

Non-German speakers (share)

0.034

–0.006

0.369

–0.035

(0.030)

(0.031)

(0.233)

(0.026)

Time-fixed effects

No

Yes

Yes

Yes

Yes

Yes

County-fixed effects

No

Yes

Yes

Yes

Yes

Yes

GRE

No

No

No

Yes

Yes

Yes

Yields

No

No

No

Yes

Yes

Yes

Observations

1387

1387

1387

1357

615

742

Number of counties

280

280

277

127

150

Note: The table shows county-level panel estimates. Standard errors in parentheses are clustered at the county level. Thirteen observations drop out from the analysis because of missing information for the district of Cologne in the 1816 data. Significance:

(***) p < 0.01,

(**) p < 0.05,

(*) p < 0.1.

Source: See Cinnirella and Hornung (2016) for data sources and details.

(p.209) In column 4, we add controls for land rent and agricultural yields in wheat and rye. Similar to the previous panel analysis, we interact land rent and agricultural yields with the five periods.21 These control variables should account for changes in land productivity that might have affected female marriage decisions. Accounting for land rent and crop yields does not change our findings—there is no significant relationship between the concentration of large landownership and the share of married women. Estimating the fixed-effects model separately for the east and the west (columns 5 and 6) does not yield heterogeneity in the relationship either.

## 8.7.2 Education and the Female Marriage Rate

Thus far, we have presented evidence for the relationship between the institutions prevailing from the feudal system and marriage rates. However, as discussed previously, relative changes in female incomes might be the reason for changes in marriage rates. Since data on relative changes in income are unavailable for the period under observation, we proxy income by an important determinant of income—the investment in education.

In this section, we test whether investments in education—that are to a large extent determined by variations in landownership concentration—are associated with different marriage patterns. According to the theoretical model of Galor and Weil (1996), an increase in the capital-labor ratio raises the relative wage of women because female labor has a higher complementarity with capital than male labor. An increase in the relative female wage increases the opportunity cost of children, potentially lowering fertility. In the context of nineteenth-century Prussia, we test whether higher investments in education, a proxy for higher returns to human capital, and thus, possibly, higher employment opportunities for women affected the marriage pattern. In fact, figure 8.2 indicates that there is a significant negative relationship between female labor force participation in industry and the share of married women.

Further descriptive evidence indicates that education and female marriage rates are negatively related. Figure 8.3 shows the evolution of enrollment rates in primary schools and marriage rates over time across Prussian districts. The overall picture reveals a strong divergence over the century. Higher enrollment rates, arguably proxying for higher returns to education, which are beneficial to women (who have a comparative advantage in nonmanual tasks), seem to be negatively correlated with marriage rates.

(p.210)

Figure 8.3 Female marriage and enrollment rates.

Source: Prussian census data.

(p.211) We test this hypothesis by estimating the relationship of interest using our panel dataset in a model similar to equation (2), substituting enrollment rates for land concentration. The results are presented in table 8.8. Column 1 presents results from a simple bivariate pooled OLS regression: the estimated coefficient for the enrollment rate is negative and significant. In column 2, we include county- and time-fixed effects. We find a significant negative relationship between enrollment rates and the share of married women. The finding is confirmed in column 3 after including the set of control variables: the relationship of interest remains negative and highly significant. The coefficients for the control variables confirm the previous pattern: both urbanization and Protestantism are associated with lower marriage rates. As expected, a higher sex ratio is positively related to the female marriage ratio—more men lead to a higher share of married women. Higher employment in the secondary sector is positively associated with female marriage rates. We interpret this coefficient as capturing an income effect (Galor and Weil 1996; Becker, Murphy, and Tamura 1990).

Consistent with the estimates presented in table 8.7, we include additional controls for land rent and agricultural yields in column 4. The coefficient for enrollment rate remains significant and of similar magnitude. In column 5, we test whether, accounting for enrollment rates, land concentration has a residual impact on female marriage rates. Consistent with the previous results, we find that landownership concentration has no direct impact on the share of married women.

So far, we have interpreted enrollment rates as a proxy for higher returns to education, which would be comparatively more beneficial to women. However, it is also interesting to estimate the relationship between lagged investments in education and current marriage rates. In this way, we can capture, at least to some extent, the marriage decision of those women who were in school in the previous period.22 Thus, in column 6, we include lagged values of the enrollment rate. Indeed, we find a significant negative coefficient of similar magnitude, which indicates a strong persistence in the relationship of interest. Such an estimate also rules out the issue of reverse causality.

Finally, in columns 7 and 8, we estimate separate fixed-effects models for the counties west and east of the river Elbe. In both cases, we find a significant negative relationship between investments in education and the share of married women. The point estimates suggest that the relationship is stronger in the more industrialized western counties, where enrollment rates likely capture better employment opportunities and higher wages for women. (p.212)

Table 8.8 Female Marriage and Enrollment Rates—Panel Estimates

Dep. variable: Share of married women

(1) Pooled

(2) FE

(3)

(4)

(5)

(6)

(7) West

(8) East

Enrollment rate (6–14)

–0.080***

–0.085***

–0.056***

–0.049***

–0.048***

–0.065***

–0.021*

(0.011)

(0.011)

(0.011)

(0.010)

(0.010)

(0.015)

(0.011)

Lagged enr. rate (6–14)

–0.040***

(0.008)

Land concentration

–0.000

–0.001

–0.002

(0.001)

(0.004)

(0.001)

Urban (share)

–0.062***

–0.057***

–0.058***

–0.063***

–0.074***

–0.045

(0.015)

(0.017)

(0.017)

(0.015)

(0.019)

(0.030)

Protestant (share)

–0.143***

–0.103**

–0.102**

–0.051

–0.131**

–0.140**

(0.047)

(0.046)

(0.046)

(0.053)

(0.056)

(0.054)

Sex ratio

0.217***

0.195***

0.195***

0.173***

0.057

0.431***

(0.062)

(0.065)

(0.065)

(0.029)

(0.039)

(0.045)

Industrial (share)

0.256***

0.196***

0.198***

0.118***

0.253***

0.093***

(0.045)

(0.041)

(0.042)

(0.043)

(0.051)

(0.034)

Agricultural (share)

0.014

0.002

0.002

0.006

–0.028

0.024

(0.015)

(0.015)

(0.015)

(0.019)

(0.021)

(0.015)

Non-German speakers (share)

0.059**

0.015

0.014

–0.024

0.431*

–0.027

(0.028)

(0.030)

(0.030)

(0.026)

(0.224)

(0.025)

Time-fixed effects

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

County-fixed effects

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

GRE

No

No

No

Yes

Yes

Yes

Yes

Yes

Yields

No

No

No

Yes

Yes

Yes

Yes

Yes

Observations

1389

1389

1389

1358

1357

1085

615

742

Number of counties

280

280

277

277

277

127

150

Note: The table shows county-level panel estimates. Standard errors in parentheses are clustered at the county level. Thirteen observations drop out from the analysis because of missing information for the district of Cologne in the 1816 data. Signficance:

(***) p < 0.01,

(**) p < 0.05,

(*) p < 0.1.

Source: See Cinnirella and Hornung (2016) for data sources and details.

(p.213) In brief, by exploiting within-county variation and controlling for a large set of confounding factors, we find that land concentration is not directly related to marriage rates. Yet, increasing levels of primary school enrollment rates are strongly negatively associated with changes in the female marriage rate. This result is consistent with the hypothesis that higher enrollment rates capture higher returns to skills that are comparatively more beneficial to women and resulted in higher female employment opportunities and higher female wages (Galor and Weil 1996). We argue that such higher employment opportunities and wages for women determine the observed higher celibacy rate. Clearly, these are reduced-form estimates that do not allow any claim of causality and force us to be speculative. Yet, these preliminary findings, if confirmed by more thorough analysis, would suggest that higher returns to education and the accumulation of human capital, by offering more female employment opportunities and higher wages, played a crucial role in limiting fertility not only through the quantity-quality tradeoff (Becker, Cinnirella, and Woessmann 2010, 2012) but also by increasing the female celibacy rate.

# 8.8 Conclusion

This chapter integrates two distinct parts of the literature on long-run development. A part of this literature analyzes the effect of inequality on development by focusing on the detrimental effect of inequality on investment in human capital. Another part of the literature analyzes the role of human capital investment in changing the female fertility choices that triggered the demographic transition.

After discussing the main theoretical models on the long-run determinants of human capital and reviewing the main empirical studies on the topic, we present some evidence for the relationship between land inequality and primary schooling in nineteenth-century Prussia. We show that landownership concentration is negatively associated with enrollment rates over the course of the nineteenth century and that the relationship is robust to the introduction of a large set of confounding factors.

We further aim at integrating the literature by investigating the relationship between land inequality and female marriage patterns, which is a major determinant of female fertility choices. In particular, we investigate to what extent the authority of noble landowners that prevailed (p.214) even after the abolition of feudal institutions influenced female marriage rates. On the one hand, the landed elite might have favored the formation of families to increase the labor force bounded to land. On the other hand, the landed elite might have been interested in preserving a sufficient number of domestic servants who were required to remain celibate. In fact, cross-sectional and panel estimates do not reveal any systematic relationship between landownership concentration and the share of married women.

Yet, as land inequality evidentially affected investments in human capital, we proceed to analyze the relationship between enrollment rates and the female marriage rate to understand whether inequality had an indirect effect on female fertility choices. We find evidence for a robust negative relationship between primary school enrollment rates and female marriage rates in fixed-effects panel analysis. We argue that this relationship can be explained by the fact that, during the Industrial Revolution, the capital-labor ratio changed and increased the relative wage of women. In our empirical model, higher enrollment rates proxy for higher returns to skills and thus higher employment opportunities for women, which, in turn, increased the economic independence of women and resulted in an increase in the female celibacy rate. These preliminary results, if confirmed, could imply that higher returns to education and the accumulation of human capital played a crucial role in limiting fertility not only through the quantity-quality tradeoff but also by increasing the female celibacy rate.

(p.216) References

Bibliography references:

Acemoglu, D., D. Cantoni, S. Johnson, AND J. A. Robinson (2011): “The Consequences of Radical Reform: The French Revolution”, American Economic Review, 101, 3286–3307.

Acemoglu, D., F. A. Gallego, AND J. A. Robinson (2014): “Institutions, Human Capital, and Development”, Annual Review of Economics, 6, 875–912.

Acemoglu, D., S. Johnson, AND J. A. Robinson (2001): “The Colonial Origins of Comparative Development”, American Economic Review, 91, 1369–1401.

——— (2002): “Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution”, Quarterly Journal of Economics, 117, 1231–1294.

Acemoglu, D., AND J. A. Robinson (2000): “Why Did the West Extend the Franchise? Democracy, Inequality and Growth in Historical Perspective”, Quarterly Journal of Economics, 115, 1167–1199.

——— (2006): Economic Origins of Dictatorship and Democracy, Cambridge: Cambridge University Press.

Alesina, A., AND D. Rodrik (1994): “Distributive Politics and Economic Growth”, Quarterly Journal of Economics, 108, 465–490.

Banerjee, A., AND L. Iyer (2005): “History, Institutions, and Economic Performance: The Legacy of Colonial Land Tenure Systems in India”, American Economic Review, 95, 1190–1213.

Becker, G. S., K. M. Murphy, AND R. Tamura (1990): “Human Capital, Fertility, and Economic Growth”, Journal of Political Economy, 98, 12–37.

Becker, S. O., F. Cinnirella, E. Hornung, AND L. Woessmann (2014): “iPEHD: The ifo Prussian Economic History Database”, Historical Methods, 47, 57–66.

Becker, S. O., F. Cinnirella, AND L. Woessmann (2010): “Education versus Fertility: Evidence from before the Demographic Transition”, Journal of Economic Growth, 15, 177–204.

——— (2012): “The Effect of Investment in Children’s Education on Fertility in 1816 Prussia”, Cliometrica, 6, 29–44.

——— (2013): “Does Women’s Education Affect Fertility? Evidence from Pre-Demographic Transition Prussia”, European Review of Economic History, 17, 24–44.

Becker, S. O., AND L. Woessmann (2009): “WasWeberWrong? A Human Capital Theory of Protestant Economic History”, Quarterly Journal of Economics, 124, 531–596.

Berdahl, R. M. (1988): The Politics of the Prussian Nobility, New Jersey: Princeton University Press.

Berlin, Königlichen statistischen Bureau (1889): Preussische Statistik, volume 101, Ernst Kuehn.

Bowman, S. D. (1980): “Antebellum Planters and Vormärz Junkers in Comparative Perspective”, American Historical Review, 85, 779–808.

Cinnirella, F., AND E. Hornung (2016): “Landownership Concentration and the Expansion of Education”, Journal of Development Economics, 121, 135–152.

(p.217) Clark, G., AND R. Gray (2014): “Geography Is Not Destiny: Geography, Institutions and Literacy in England, 1837–63”, Oxford Economic Papers, 66, 1042–1069.

de la Croix, D., AND M. V. Donckt (2010): “Would Empowering Women Initiate the Demographic Transition in Least Developed Countries?” Journal of Human Capital, 4, 85–129.

De Moor, T., AND J. L. Van Zanden (2010): “GirlPower: The European Marriage Pattern and Labor Markets in the North Sea Region in the Late Medieval and Early Modern Period”, Economic History Review, 63, 1–33.

Dell, M. (2010): “The Persistent Effects of Peru’s Mining Mita”, Econometrica, 78, 1863–1903.

Dennison, T., AND S. Ogilvie (2014): “Does the European Marriage Pattern Explain Economic Growth?” Journal of Economic History, 74, 651–693.

Diebolt, C., AND F. Perrin (2013a): “From Stagnation to Sustained Growth: The Role of Female Empowerment”, AFC Working Paper No. 4.

——— (2013b): “From Stagnation to Sustained Growth: The Role of Female Empowerment”, American Economic Review, 103, 545–49.

Easterly, W. (2007): “Inequality Does Cause Underdevelopment: Insights From a New Instrument”, Journal of Development Economics, 84, 755–776.

Eddie, S. A. (2013): Freedom’s Price: Serfdom, Subjection, and Reform in Prussia, Oxford.

Engerman, S. L., AND K. L. Sokoloff (1997): “Factor Endowments, Institutions, and Differential Paths of Growth among New World Economies: A View from Economic Historians of the United States”, in How Latin America Fell Behind: Essays on the Economic Histories of Brazil and Mexico, 1800–1914, ed. S. H. Haber, Stanford University Press, 260–304.

Foreman-Peck, J. (2011): “The Western European Marriage Pattern and Economic Development”, Explorations in Economic History, 48, 292–309.

Gallego, F. A. (2010): “Historical Origins of Schooling: The Role of Democracy and Political Decentralization”, Review of Economics and Statistics, 92, 228–243.

Galor, O. (2011): Unified Growth Theory, NJ: Princeton University Press.

Galor, O., AND O. Moav (2006): “Das Human-Kapital: A Theory of the Demise of the Class Structure”, Review of Economic Studies, 73, 85–117.

Galor, O., O. Moav, AND D. Vollrath (2009): “Inequality in Land Ownership, the Emergence of Human Capital Promoting Institutions and the Great Divergence”, Review of Economic Studies, 76, 143–179.

Galor, O., AND D. N. Weil (1996): “The Gender Gap, Fertility, and Growth”, American Economic Review, 86, 374–387.

Go, S., AND P. Lindert (2010): “The Uneven Rise of American Public Schools to 1850”, Journal of Economic History, 70, 1–26.

Hajnal, J. (1965): “European Marriage Pattern in Perspective”, in Population in History, London: Arnold, D.V. Glass and D.E.C. Eversley.

Harnisch, H. (1984): Kapitalistische Agrarreform und Industrielle Revolution, Weimar: Hermann Böhlaus Nachfolger.

(p.218) Iyigun, M., AND R. P. Walsh (2007): “Endogenous Gender Power, Household Labor Supply and the Demographic Transition”, Journal of Development Economics, 82, 138–155.

Kopsidis, M., AND N. Wolf (2012): “Agricultural Productivity across Prussia during the Industrial Revolution: A Thuenen Perspective”, Journal of Economic History, 72, 634–670.

Kourtellos, A., I. Stylianou, AND C. M. Tan (2013): “Failure to Launch? The Role of Land Inequality in Transition Delays”, European Economic Review, 62, 98–113.

Kuhlemann, F.-M. (1991): “Schule, Hochschulen, Lehrer”, in Handbuch der deutschen Bildungsgeschichte, ed. C. Berg, München: Verlag C. H. Beck, 170–370.

Lagerlöf, N.-P. (2003): “Gender Equality and Long-Run Growth”, Journal of Economic Growth, 8, 403–426.

Lindert, P. H. (2003): “Voice and Growth: Was Churchill Right?” Journal of Economic History, 63, 315–350.

———(2004): Growing Public: Social Spending and Economic Growth since the Eighteenth Century, Cambridge: Cambridge University Press.

Mariscal, E., AND K. L. Sokoloff (2000): “Schooling, Suffrage, and Inequality in the Americas, 1800–1945.” in Political Institutions and Economic Growth in Latin America. Essay in Policy, History, and Political Economy, ed. S. H. Haber, Stanford: Hoover Institution Press, 159–218.

Naidu, S. (2012): “Suffrage, Schooling, and Sorting in the Post-Bellum U.S. South”, NBER Working Paper 18129.

Neugebauer, W. (1992): “Das Bildungswesen in Preußen seit der Mitte des 17. Jahrhun-derts”, in Handbuch der Preussischen Geschichte, ed. O. Busch, Berlin, New York: Walter de Gruyter, volume 2, 605–799.

Nunn, N. (2008): “Slavery, Inequality, and Economic Development in the Americas: An Examination of the Engerman-Sokoloff Hypothesis”, in Institutions and Economic Performance, ed. E. Helpman, Cambridge: Harvard University Press.

Ogilvie, S. (2005): “Communities and the ‘Second Serfdom’ in Early Modern Bohemia”, Past & Present, 187, 69–119.

Persson, T., AND G. Tabellini (1994): “Is Inequality Harmful for Growth? Theory and Evidence”, American Economic Review, 84, 600–621.

Pierenkemper, T., AND R. Tilly (2004): The German Economy during the Nineteenth Century, New York and Oxford: Berghan Books.

Ramcharan, R. (2010): “Inequality and Redistribution: Evidence from U.S. Counties and States, 1890–1930”, Review of Economics and Statistics, 92, 729–744.

Rosenberg, H. (1944): “The Rise of the Junkers in Brandenburg-Prussia, 1410–1653: Part 2”, American Historical Review, 49, 228–242.

Schleunes, K. A. (1979): “Enlightenment, Reform, Reaction: The Schooling Revolution in Prussia”, Central European History, 12, 315–342.

Sokoloff, K. L., AND S. Engerman (2000): “Institutions, Factor Endowments, and Paths of Development in the New World”, Journal of Economic Perspectives, 14, 217–232.

(p.219) Voigtländer, N., AND H.-J. Voth (2013): “How the West ‘Invented’ Fertility Restriction”, American Economic Review, 103, 2227–2264.

Vollrath, D. (2013): “School Funding and Inequality in the United States, 1890”, Explorations in Economic History, 50, 267–284.

Wrigley, E. A. (2006): Industrial Growth and Population Change, Cambridge: Cambridge University Press. (p.220)

## Notes:

(*) Corresponding author: Ifo Institute, Poschingerstr. 5, 81679 Munich, Germany; cinnirella@ifo.de.

() University of Bayreuth, Universitaetsstr. 30, 95447 Bayreuth, Germany; erik.hornung@uni-bayreuth.de.

(1.) Unfortunately, we cannot separate the effect of female from male enrollment rates because the two variables are highly correlated. In addition, enrollment rates by gender are not available for all periods.

(2.) For the long-run effects of property rights on health and educational outcomes, the reader is referred to Banerjee and Iyer (2005).

(3.) At the Congress of Vienna, Prussia received two-fifths of Saxony, Swedish Pomerania, the Grand Duchy of Posen, Danzig, and the provinces of Rhineland and Westphalia.

(4.) The Elbe does not present a clear discontinuity between the regimes, and the transition between the systems was gradual (Rosenberg 1944).

(5.) The appointment of the school teacher for the estate-school remained within the authority of the estate owner even around 1900 (Neugebauer 1992, 684). Note that (p.215) teachers might be selected by noble landowners; the payment of teacher salaries, however, fell on the entire community.

(6.) For more details about the regional adoption of reforms, you are referred to Acemoglu et al. (2011) and the corresponding online appendix.

(7.) Shares correspond to the distribution of funds across sources as of 1886 (Berlin 1889, 84).

(8.) Note that both provinces of East and West Prussia belong to the East-Elbe part of Prussia.

(9.) We cannot exclude that it affected their quality.

(10.) Similarly, apprentices living in the household of their masters were not allowed to marry.

(11.) The compulsion to provide domestic service to the nobility was abolished with serfdom institutions in 1810. Since the nobility still required servants, a large proportion of the population provided domestic services by choice and against wage payment.

(12.) Much of the data are stored in the ifo Prussian Economic History Database (iPEHD). For more details on variables and sources of the iPEHD, you are referred to Becker et al. (2014).

(13.) In a few cases, enrollment rates exceed 100%. This could be due to children commuting from neighboring counties or because of children older than fourteen years being enrolled in school.

(14.) One PM is equal to circa 0.25 hectare.

(15.) Acquisitions include Hohenzollern (1850), Schleswig and Holstein (1865), the Kingdom of Hanover (1866), the Electorate of Hesse (1866), the Duchy of Nassau (1866), the free City of Frankfurt (1866), and Saxe-Lauenburg (1876).

(16.) Changes in the administrative boundaries throughout the nineteenth century complicate the analysis of Prussian county-level data. Becker et al. (2014) provide instructions on how to merge datasets from different sources and periods to obtain a panel structure.

(17.) These results were first shown in Cinnirella and Hornung (2016).

(18.) The variable for land rent is based on the Grundsteuerreinertrag, defined as the income from agrarian use of land less the costs of farming (Kopsidis and Wolf 2012). Crop yields refer to yields of wheat and rye.

(19.) The share of married women is defined as the number of married women over the total number of women older than fifteen years.

(20.) In figure 8.2, we show the relationship between female marriage rates and female labor force participation in the industrial sector in 1882. As information on labor force participation by gender is unavailable for earlier periods, such a variable cannot be included in the panel analysis.

(21.) See the appendix in Cinnirella and Hornung (2016) for more details on data on land rent and crop yields.

(22.) Becker, Cinnirella, and Woessmann (2013) adopt a similar approach to estimate the effect of mothers’ education on their fertility.