DISCUSSION PAPER SERIES

IZA DP No. 3837

Luther and the Girls: Religious Denomination and

the Female Education Gap in 19th Century Prussia

Sascha O. Becker
Ludger Wößmann

November 2008

Forschungsinstitut
zur Zukunft der Arbeit
Institute for the Study
of Labor
 Luther and the Girls:
Religious Denomination and the Female
Education Gap in 19th Century Prussia

Sascha O. Becker
University of Stirling,
CESifo and IZA

Ludger Wößmann
University of Munich, Ifo Institute,
CESifo and IZA

Discussion Paper No. 3837

November 2008

IZA

P.O. Box 7240

53072 Bonn
Germany

Phone: +49-228-3894-0
Fax: +49-228-3894-180
E-mail: [email protected]

Any opinions expressed here are those of the author(s) and not those of IZA. Research published in
this series may include views on policy, but the institute itself takes no institutional policy positions.

The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center
and a place of communication between science, politics and business. IZA is an independent nonprofit
organization supported by Deutsche Post World Net. The center is associated with the University of
Bonn and offers a stimulating research environment through its international network, workshops and
conferences, data service, project support, research visits and doctoral program. IZA engages in (i)
original and internationally competitive research in all fields of labor economics, (ii) development of
policy concepts, and (iii) dissemination of research results and concepts to the interested public.

IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be
available directly from the author.
IZA Discussion Paper No. 3837
November 2008

ABSTRACT

Luther and the Girls: Religious Denomination and the

Female Education Gap in 19th Century Prussia*
Martin Luther urged each town to have a girls’ school so that girls would learn to read the
Gospel, evoking a surge of building girls’ schools in Protestant areas. Using county- and
town-level data from the first Prussian census of 1816, we show that a larger share of
Protestants decreased the gender gap in basic education. This result holds when using only
the exogenous variation in Protestantism due to a county’s or town’s distance to Wittenberg,
the birthplace of the Reformation. Similar results are found for the gender gap in literacy
among the adult population in 1871.

JEL Classification: I21, J16, N33, Z12

Keywords: gender gap, education, Protestantism

Corresponding author:

Ludger Wößmann
ifo Institute for Economic Research
at the University of Munich
Poschingerstr. 5
81679 Munich
Germany
E-mail: [email protected]

*
We received substantive comments during various seminar presentations. Discussions with and
comments from Davide Cantoni, Peter Egger, Nicola Fuchs-Schündeln, Monika Piazzesi, Martin
Schneider, and Holger Sieg were very fruitful. Erik Hornung and Clemens König provided capable
research assistance.
 “And would to God that every town also had a girls’ school,
in which the girls were taught the Gospel for an hour each day.”
Martin Luther (1520), To the Christian Nobility of the German
Nation Concerning the Improvement of the Christian Estate.

I. Introduction
In the history of mass education, Prussia is generally considered the role model of
educational reforms. Not only did it provide a decent basic education to most of its boys
early on, but it had virtually reached gender parity in primary education already by the
second half of the 19th century. Despite the importance of education for economic
prosperity in general and for the gender gap in economic outcomes in particular, the factors
that historically helped to close the female education gap are barely understood.
This paper suggests that Protestantism was a distinctive driving force in the
advancement of female education in Prussia. Martin Luther explicitly urged girls, as well as
boys, to be able to read the Gospel as a solely religious aim. This pledge for female reading
ability helped to promote schooling for girls. We use data on school enrolment from the
first Prussian Population Census in 1816 at the level of counties and towns to show that a
larger share of Protestants in a county or town was indeed associated with a larger share of
girls among the total school population. To our knowledge, this is the first time that these
early sub-regional data are ever used in econometric analyses.
We present instrumental variable estimates where each county’s and town’s share of
Protestants is instrumented by its distance to Wittenberg to suggest that the effect of
Protestantism can be causally interpreted. The emergence of the Reformation from Luther’s
city of Wittenberg and its diffusion in a roughly circular fashion provides exogenous
variation in Protestantism. Our results show that regions that were exogenously driven
towards Protestantism by their proximity to Wittenberg were the first to reach gender parity
in primary education in the early 1800s. The result that Protestantism was one factor that
helped to reduce the gender gap in education in Prussia is confirmed when using county-
level data on the gender gap in adult literacy in 1871. Thus, even by the time that universal
public primary education had closed the gender gap in enrolment in primary education at

1
the end of the 19th century, the gender gap in literacy was lower among Protestants than
among Catholics, and Protestant women had even more literacy than Catholic men.
While the motivation for our analysis is mostly historical in nature, the impact of
Protestantism on the gender gap in education is still visible in international data today.
Figure 1 plots the Protestant share of the population against the educational Gender Parity
Index (GPI), measured as the ratio of years of education in the female and male population,
for European countries in 1970.1 Across countries, a higher share of Protestants is clearly
associated with a higher GPI in years of education. All countries with a GPI under 0.88 are
Catholic countries. All Protestant countries lie above this value.
Of course, such cross-country comparisons are plagued by the difficulty of
disentangling the effect of religion from other possible causes of gender differences in
educational attainment that may vary across countries, such as institutions and geography.
By looking at sub-regional data within Prussia, this paper uses observations that are all
exposed to the same institutional and legal setting. Similarly, problems of geographical
variation are substantially smaller within Prussia than on a global scale, and will be dealt
with by proper control variables.
Given the substantive returns to education on the labor market (cf. Card 1999;
Psacharopoulos and Patrinos 2004), the difference in educational achievement between
women and men has crucial consequences for gender differences in economic outcomes. In
fact, education has played a major role in the evolutionary and revolutionary phases that
transformed the economic role of women (Goldin 2006). In addition, female education may
have substantial payoffs beyond the labor market. For example, Currie and Moretti (2003)
show that mothers’ education has important effects on the health of their children,
suggesting that the intergenerational transmission of human capital depends on female

1
The pattern is even more pronounced in a world-wide picture, but we stick to European countries in
this depiction to evade worries of comparisons across culturally diverse continents. Note, however, that the
relationship in Figure 1 is nonlinear and that there appears to be no relationship between the GPI and the
Protestant share of the population once the latter rises above 20 percent. The GPIs were calculated using data
on average years of education of both genders from Barro and Lee (2001). The population share of Protestants
is from Barrett, Kurian and Johnson (2001).

2
education. It is thus important to understand what forces helped to overcome the
historically poor participation of girls in basic education in the developed world.
Similarly, issues of gender parity in education are high on the agenda in developing
countries today. Gender parity constitutes one of the “Education for All” development
goals (UNESCO 2000). Only five out of 148 countries in the developing world have
achieved gender parity through tertiary education (UNESCO 2006). In line with the
religious factors emphasized in this paper, it seems that cultural factors play a major role in
preventing gender parity in education in developing countries today (UNESCO 2003, ch.
3).
The paper is structured as follows. Section II presents an overview of Luther’s and
other reformers’ teachings on female education and provides evidence on the practical
implementation of school reforms in the decades after the Reformation. Section III presents
data and empirical results on the effects of Protestantism on the gender gap in school
enrolment at the level of counties and towns from the earliest Prussian Census in 1816.
Section IV traces the development of the gender gap in school enrolment over the 19th
century and presents evidence on the gender gap in adult literacy in 1871. Section V
concludes and gives a brief outlook on the perpetuation of the denominational gender gap
in Germany at higher levels of education until today.

II. The Reformers’ Urge for Female Schooling

Education of large parts of the population was unheard of before the advent of the
Reformation. Standard estimates suggest that less than one percent of the German
population was literate at the time of the Reformation (Engelsing 1973). Even after the
Reformation, educational efforts of the Catholic Church remained hugely different in scale
from the Protestant Church, especially for girls.
Luther’s views contrasted strongly with this Catholic neglect of schooling of boys and
girls. In what is generally viewed as his first major pamphlet that signified the breakthrough
of the Reformation among the general public, To the Christian Nobility of the German
Nation Concerning the Reform of the Christian Estate, Luther (1520) explicitly demanded
not only that every town should have a boys’ school, but also a girls’ school: “Above all,

3
the foremost and most general subject of study, both in the higher and the lower schools,
should be the Holy Scriptures, and for the young boys the Gospel. And would to God that
every town had a girl’s school also, in which the girls were taught the Gospel for an hour
each day… Ought not every Christian at his ninth or tenth year to know the entire holy
Gospel…?”
Luther and the other reformers not only pleaded for educational efforts of both
genders, but also influenced the implementation of school reforms on the practical front.
Towns and regions that joined the Reformation instituted new Church and School
Ordinances to align regulations with the Protestant ideals. Those were typically authored by
leading reformers. One of them, Johannes Bugenhagen, provided a basis for universal
education of both boys and girls in northern Germany, where he traveled during regular
extended leaves of absence from his duties as parish pastor in Wittenberg (cf. Green 1979).
His Church Ordinance for the city of Braunschweig in 1528, which provided for girls to
learn reading, was to set the standard for subsequent systems. In this ordinance,
Bugenhagen requested that the city should have both four boys’ schools and four girls’
schools. In his Church Ordinance for Wittenberg, he extended the request for girls’
schooling to writing and calculating.
Other leading reformers wrote schoolbooks for girls’ schools. Johann Agricola’s One
Hundred-and-Thirty Questions for the Young Children in the German Girls' School at
Eisleben (1527), a catechetical work for religious instruction, is considered the earliest
book published explicitly for use in a school for girls (cf. Green 1979).
These efforts to promote girls’ education seem to have been quite effective over the
first decades after the Reformation. Since Germany was fragmented into hundreds of states,
data coverage is necessarily limited to some of the bigger states. The Electorate
(Kurfürstentum) of Brandenburg was one of the biggest states at the time and also the core
state of what later became Prussia. Green (1979) examines documentary materials from the
visitations by church officials to the local parishes and finds that in the year that the
Reformation was introduced (1539), Brandenburg had 55 boys’ schools and 4 girls’ schools
(cf. Table 1). By 1572, the year that educational reform began under the leadership of the
reformer Andreas Musculus, the number had increased to 78 and 9, respectively.

4
Remarkable progress followed 1573-1600, when the number of schools for girls increased
five-fold to 45.
It seems that there was nothing similar in terms of developments in the number of
girls’ schools in Catholic regions. Quite to the contrary, in Bavaria, the biggest Catholic
state in Germany at the time, there were still strong objections against schools in the
countryside in general as late as 1614 (Gawthrop and Strauss 1984).
The supply of girls’ schools is arguably only one prerequisite for an increased
educational attainment of girls. In his Sermon on Keeping Children in School, Luther
(1530, p. 526) also extended his educational postulations to every individual Christian and
asked parents to send their children to school, thus also stimulating the demand for
education. A likely channel for this last effect is the non-monetary benefit associated with
literacy perceived by Protestants, namely the ability to read the Bible (cf. Becker and
Woessmann 2009).
To the extent that, in Protestant areas, more girls’ schools were built and Protestant
parents followed Luther’s request to send girls to schools, we would expect a narrower
gender gap in education among Protestants.

III. Protestantism and the Gender Gap in Basic Education in 1816 Prussia
This section tests empirically whether Protestantism did indeed lead to a reduction in
the gender gap in education, using the earliest point in time for which data across all
Prussian counties and municipalities are available.

Data for 1816 Prussia

Data coverage over the 17th and 18th centuries is, unfortunately, equally scarce as in the
16th century, for which we presented exemplary evidence in the previous section. In 1805,
Prussia founded its Prussian Statistical Bureau. This statistical office performed the first
full-scale Population Census in 1816, collecting data on population, occupation and
education which was later reported at the county and municipality level (see data appendix
for details). This is thus the earliest year which lends itself to a micro-econometric analysis

5
of education, gender and religion, and we are not aware that anybody has used these data
for micro-econometric analysis at all.
Prussia in 1816 was divided into 357 counties, 289 of which had the necessary data in
1816 to be included in the analysis. In addition, data are also reported for 172 “large and
medium-sized” towns (156 with necessary data in 1816; the smallest included town had
roughly 2,500 inhabitants). While these towns are also contained in the county averages of
the county-level data, the town data provide additional detailed information on specific
schools and other town characteristics.
In particular, the county-wide data contain only information on public primary schools.
Other school types, such as private schools and secondary schools, seem to barely exist or
not exist at all outside the large and medium-sized towns, and no data are reported in the
county totals. By contrast, the town data contain more detailed information. These include
both the distinction between public and private schools and different types of secondary
schools (see below).
Basic descriptive statistics for the county and the town samples are provided in the
appendix Tables A.1 and A.2. The average share of Protestants across the 289 counties is
59%, and 70% across the towns. However, in both samples, the share of Protestants varies
from 0% to 100%. This stark distinction between all-Protestant and all-Catholic counties in
Prussia (cf. also Figure 2) provides the interesting denominational variation within the
framework of a single country that enables the analysis of this paper.2
In the following analysis, we first look at the county data on enrolment in public
primary schools and then at the town data on enrolment in public and private primary and
secondary schools.

Evidence from Counties

If access to education were gender-neutral, the enrolment rate of girls would reflect the
gender composition among school-aged children. A descriptive look at the county-wide

2
As is evident from the figure, Prussia annexed several territories between 1816 and 1871, namely
Hohenzollern-Sigmaringen, Schleswig-Holstein, the Kingdom of Hannover, Hessen-Kassel, Nassau, and the
free city of Frankfurt.

6
data reveals that enrolment in primary schools is clearly lower for girls than for boys (cf.
Table A.1). The share of girls among pupils enrolled in public primary schools is 47.3% on
average across Prussian counties (compared to the share of girls in the relevant age group
of 49.1%), below gender parity.
To test whether these gender differences in educational participation are less
pronounced in Protestant areas, we perform simple regression analyses. Our dependent
variable is the share of female pupils among all pupils in public primary school. The first
column of Table 2 displays results from a bivariate regression of the share of girls in all
pupils against the share of Protestants in the county. A higher fraction of Protestants is
associated with a higher share of girls enrolled in primary school. The point estimate of
0.020 can be interpreted as follows: When a county goes from being all Catholic (0%
Protestants) to all Protestant, the share of female pupils among all pupils in public primary
school increases by two percentage points.
A demographic reason for differences in female shares in enrolment could be
differences in the gender composition of school-aged children. In column (2), we control
for the fractions of girls in the age group 8-14, as available in our data set. A higher fraction
of girls in school age is indeed associated with a higher share of girls in primary school.
However, the coefficient on Protestantism remains unaffected.
In column (3), we add geographical controls: latitude (in rad), longitude (in rad), their
interaction, and the fraction of the population living in towns. The coefficient on the
fraction of Protestants increases substantially, doubling to four percentage points. The
negative coefficient on the fraction of the population living in towns reflects an institutional
feature, as will become clear when looking at data for Prussian towns. Since, in towns,
enrolment of girls is more pronounced in private primary schools (constituting 17% of
primary school enrolment in towns), we expect a higher fraction of urban population in a
county to lower the share of girls enrolled in public primary school.
Another reason for the gender gap to vary across counties may be the stage of
development. Glewwe and Kremer (2006) argue that, at the international level, the gender
gap in education closes as national income increases. At the individual level, richer families
may be more inclined to send their daughters to school. Direct income measures are not

7
available in the 1816 Prussian Census data. However, natural proxies are variables
measuring agricultural productivity, in particular after controlling for the share of people
living in towns. We transform the following livestock numbers at the county level to per
capita numbers: number of horses, number of foals, number of bulls, number of oxen,
number of cows, number of young cattle, number of sheep and number of goats. In column
(4) of Table 2, we use only the number of horses and bulls per capita, but results are very
similar when using all agricultural variables (with the other agricultural variables being
statistically insignificant). Again, the coefficient on the share of Protestants remains
unchanged.
An interesting question is whether school enrolment is mostly demand-driven or
supply-driven, i.e. whether the reduced female enrolment gap is partly explained by better
school supply in Protestant areas. To probe this point, we include the number of primary
schools per 1,000 inhabitants as a measure of school supply in the county. Column (5)
shows that including supply of primary schools slightly reduces the effect of Protestantism
on the fraction of girls enrolled, from 0.041 to 0.035. The effect of the number of primary
schools per 1,000 inhabitants itself on the fraction of girls enrolled is positive and
significant. Taken together, this indicates that part of the enrolment effect for girls works
through higher school density in Protestant areas, but that the bulk of the effect comes from
the demand side. Protestant parents seem to have a higher preference for schooling of their
daughters than Catholic parents.

Endogeneity of Protestantism? Using Distance to Wittenberg as an Instrument

Several historical facts suggest that the origin of denominational differences in Prussia
can be viewed as an exogenous shock (cf. Becker and Woessmann 2009 for details). Most
of the denominational variation across Prussia in the 19th century can be traced back to
denominational choices of local rulers during the Reformation in 16th and early 17th
century, mostly motivated by religious conviction and power politics vis-à-vis the Pope and
the German Emperor. In addition, the increased literacy of Protestants was an unintended
side effect of Luther’s Gospel-reading aims, unrelated to a work ethic or any other
economic thought.

8
 Still, there may be worries that the adoption of Protestantism may have been
endogenous to pre-existing ethical, educational or economic differences. To rule out such
potential remaining worries about endogeneity of the share of Protestants in a county, we
employ an instrumental variable strategy that uses a particular aspect of the historical
diffusion of Protestantism across the German Empire in order to restrict the variation in
Protestantism used in the estimation to a part that is credibly exogenous. Reformation
historians refer to the diffusion of Protestantism as resembling the propagation of a wave
caused by a stone thrown into water. Luther’s preaching had its most imminent effect in the
area surrounding his city of Wittenberg, and there is a tendency for the impact to diminish
with distance to Wittenberg. In effect, in the German Empire, Protestantism dispersed
around Wittenberg in a mostly concentric pattern. As evidenced in Figure 2, it seems that
the Reformation spread out from Wittenberg in all directions, but then came to a halt after
some distance.
The main reasons for a roughly circular dispersion around Wittenberg may have been
the costs of traveling and of information diffusion through space, and these transportation
and transaction costs played a crucial role at the time. Among other things, rulers who
wanted to convert their church system to Protestantism had to send priests to study the new
denomination in Wittenberg. Thus, thousands of students came to Wittenberg to hear
Luther’s sermons and speeches, and they spread the word as preachers back in their home
regions (cf. Bunkowske 1985). Given the arduousness of travel in the early 1500s, the
propensity to come to Wittenberg to listen to Luther and his successors likely declined with
distance to Wittenberg. The fact that the German regions spoke ever more different dialects
the further distant the regions may also have contributed to a concentric pattern of the
dispersion of Protestantism, both in oral and written means of dissemination, and in the
dissemination both to rulers and to the population at large.
It is generally accepted that Wittenberg was an “unimportant place” (Holborn 1942, p.
133) until 1517. Therefore, distance to Wittenberg should be unrelated to a county’s
economic and educational state before it adopted Protestantism. However, it is hard to
ascertain this rigorously across our Prussian counties because there are hardly any data on
the economic or educational situation for the time of Reformation, and the 1816 Population

9
Census is explicitly the first occasion where state-wide data on school enrolment was
surveyed.
But several pieces of evidence support the idea that distance to Wittenberg is unrelated
to the economic and educational situation before 1517. In Becker and Woessmann (2009),
we show that, in the sample of 452 Prussian counties as of 1871, distance to Wittenberg is
completely insignificant in predicting the probability of being a free imperial city
(measured in pre-Reformation status), considered to be centers of economic activity before
the Reformation. Second, the same is true for the probability of being a free Hanseatic city,
which constituted major trading hubs in pre-Lutheran times. Third, distance to Wittenberg
is similarly uncorrelated with urban population density in 1500, a proxy often used for
economic progressiveness before industrialization. Fourth, in terms of proxies for the
educational situation before 1517, we estimate whether distance to Wittenberg predicts
whether a county had a university before 1517. We also regress the year of foundation of
universities in existence before 1517 on distance to Wittenberg. In both exercises, distance
to Wittenberg is completely unrelated to the spread of universities before Lutheran times.
Fifth, we perform the same exercise for those present-day German schools that trace back
their history to pre-Reformation times. Both a dummy for the existence of and the year of
foundation of those schools are unrelated to distance to Wittenberg. Sixth, the density of
monasteries, guardians of learnedness in medieval times that preserved the skill of literacy
and often contained substantial libraries, measured in 1517 is equally unrelated to distance
to Wittenberg.
As a consequence, the geographically concentric pattern of the dispersion of the
Reformation provides a means to obtain a specific variation in Protestantism that is credibly
exogenous to economic and educational considerations: the variation due to distance to
Wittenberg. We thus use distance to Wittenberg as an instrument for the share of
Protestants in a county in 19th century Prussia.
Table 3 reports instrumental variable (IV) estimates of the effect of Protestantism on
girls’ share in primary school enrolment, where Protestantism is instrumented by distance
to Wittenberg. As is evident from the F-statistic of the instrument in the first stage, distance
to Wittenberg is a strong instrument for the share of Protestants in a county. Each 100 km

10
distance to Wittenberg is associated with a Protestant share that is 13-19 percentage points
lower (see columns (1)-(4)). The second stage uses only that part of the Protestant share
that is due to distance to Wittenberg to predict the girls’ share in primary school enrolment.
The positive effect of Protestantism on girls’ share in primary school enrolment is
highly robust in the IV specification. In fact, the point estimate is higher, with a difference
in girls’ share in primary school enrolment of 4.6-4.9 percentage points between an all-
Protestant and an all-Catholic county, depending on the specification (see columns (5)-(8)).
The finding that school supply has a negligible effect on girls’ enrolment (column (8))
shows up also in the IV estimates.
We conclude from the data on county totals that there is an economically and
statistically significant effect of Protestantism on the female enrolment gap. The result is
robust across several specifications and remains stable when instrumenting the fraction of
Protestants by distance the Wittenberg.

Evidence from Towns

The county-wide data are limited to information about public primary schools. The
data on Prussian towns, in contrast, have information on both public and private schools
and on primary as well as secondary schools. The descriptive pattern that female enrolment
numbers are lower than male enrolment numbers is also visible in towns: Across the 156
towns with non-missing education data, 41.5% of pupils are female (compared to a
population share of 50.5% in the relevant age group; cf. Tables 4 and A.2).3
Table 4 splits down the educational gender gap by type of school. The fraction of
females falls with the level of schooling. While in primary schools, girls make up 46.5% of
all pupils, in (lower) secondary schools, they only constitute 38.5% of pupils, and only boys
attend upper-secondary schools (Gelehrtenschulen, i.e. Gymnasien).4

3
There are two outliers in terms of the share of girls in the relevant age group, Rogasen (79.6%) and
Hirschberg (70.4%), which also have larger shares of females in other age groups. Dropping these two
outliers does not change any of our qualitative results.
4
Only after 1900, girls were allowed to obtain the Abitur, the university-entry certificate (see Vogt
1997).

11
 In private schools, female enrolment exceeds male enrolment. The reason for higher
enrolment of girls in private as opposed to public schools is that the Prussian state did not
yet consider female education a public duty. Secondary schools for girls were thus mostly
run privately. However, private schools constitute only a small fraction of all schools, so
that on the whole, female school enrolment is clearly below gender parity.
Table 5 presents regression analyses using the town sample. The first three columns
refer to primary school enrolment, the last three columns to secondary school enrolment.
The number of observations is lower in the latter case because not all towns host secondary
schools.
Results are broadly similar to the county sample. Column (1) shows OLS estimates for
a specification that includes the same geographic control variables as for the county data.
Instead of the agricultural controls, which are not available in the town data, we use other
indicators measuring the stage of economic development of the towns: the fraction of
houses that have massive walls (as opposed to timbered or wooden walls), the number of
businesses per capita, the number of looms per capita and the number of retailers per capita.
The coefficient on the fraction of Protestants is significantly positive at 4 percent and
thus in the same range as in the regression on the sample of county totals. Similar to the
regression on county totals, we find a small reduction of the direct effect of Protestantism
on girls’ enrolment share when controlling for the number of primary schools per 1,000
inhabitants (not reported).
Using the Wittenberg instrument again, the coefficient estimate on the fraction of
Protestants nearly doubles and stays statistically significant. Similar to the county totals
regressions, the part of the variation in the fraction of Protestants that we can arguably
attribute to the exogenous spread of the reformation yields stronger effects on the share of
girls among pupils.
When looking at secondary school enrolment in columns (4)-(6), the point estimates
for the effect of Protestantism on the female enrolment share are even larger, albeit
statistically insignificant. This may be attributed to the considerably smaller number of
observations in the town data compared to the county data set.

12
 It is instructive to control for school supply when estimating the effect of Protestantism
on girls’ enrolment in secondary schools. In contrast to primary schools, there are separate
middle schools for boys and girls. We can thus control for the number of boys’ and girls’
middle schools and upper-secondary Gymnasien per 1,000 inhabitants. The point estimate
of the direct effect of Protestantism goes down to 0.02, whereas there is a strong positive
effect of the number of girls’ middle schools on girls’ enrolment in these schools (detailed
results available from the authors upon request). When regressing the number of girls’
middle schools per 1,000 inhabitants on the fraction of Protestants, there is a strong positive
association. Given the fact that the vast majority of girls’ middle schools are privately run,
whereas boys’ middle schools are publicly run, there seems to be more private initiative to
operate girls’ middle schools in Protestant areas.
It has been argued that girls’ middle schools originally catered to relatively well-off
families eager to demarcate themselves from the lower classes and endow their daughters
with the knowledge and competences required for their future social positions as bourgeois
married women. The issue of professional training did not arise, as women of a certain rank
were not expected to work (Küpper 1987, p. 181-184). It thus appears that girls’
participation in secondary education at the time was largely demand driven. While
throughout most of the 19th century, the Prussian state authorities remained passive with
regard to female education beyond the primary level, the emerging bourgeoisie did attach
value to promoting their daughters’ education. Hence, we should expect a positive
association between female education in secondary education and the prevalence of
bourgeois elements in the population. This is precisely what we observe. When regressing
the share of girls enrolled in secondary schools (both public and private) on religion as well
as a number of demographic and economic controls, religion turns out to be statistically
insignificant, whereas the number of retailers per capita is positively and highly
significantly correlated with women’s participation (cf. Table 5).
We read the 1816 results as evidence of significant denominational differences in the
gender education gap in primary and lower-secondary education (although the latter does
not reach standard levels of statistical significance): School enrolment of girls is higher in
Protestant areas.

13
IV. The Evolution of the Gender Gap over the 19th Century
This section documents how the educational gender gap evolved between 1822 and
1864 and then estimates whether Protestantism had an effect on the gender gap in literacy
among the adult population in 1871.

The Evolution of Educational Gender Ratios from 1822 to 1864

Between 1822 and 1864, the Prussian Statistical Office published school enrolment
statistics only at the level of provinces or districts (Regierungsbezirke; see Preussische
Statistik 1889, p. 1). Figure 3 shows the evolution of the Gender Parity Index (GPI) of
primary and secondary school enrolment in 1822-1864. In primary education, the GPI starts
from about 0.92 and reaches 0.98 at the end of the period. In 1864, 1.43 million boys and
1.40 million girls are enrolled in primary school. After 1864, the regular publication of
school enrolment numbers was discontinued and only taken up again in 1886. Using data
from the 1886 Education Census, we can show that the closing of the gender enrolment gap
is virtually complete. In Prussia as a whole, in 1886, 2,422,044 boys and 2,416,203 girls are
enrolled in primary school. Girls thus make up 49.94% of pupils in primary school.
In secondary schools, however, the gender gap persists over the whole period 1822-
1864. The straight line depicts enrolment in public secondary school, whereas the dotted
line includes private secondary schools (not included in the school statistics before 1858).
Even including the private schools, the GPI in secondary education stands at only 0.57 in
1864. The absolute numbers of secondary enrolment are also telling. In 1864, only 131,430
boys and 74,761 girls are enrolled, compared to the 1.43 million boys and 1.40 million girls
enrolled in primary school, showing the limited nature of secondary school enrolment in
Prussia in the middle of the 19th century.
In sum, whereas the educational gender gap in primary school enrolment observed in
1816 closes over the century due to compulsory schooling regulations, it persists in
secondary schooling.

14
Evidence of Denominational Effects on Gender Ratios in Literacy in 1871

The Prussian Population Census in 1871 collected data on literacy for all household
members. The 1871 Population Census is explicitly the very first census ever to survey
literacy in Prussia. Literacy is measured as the ability to read and write among the
population aged 10 years or older. As an outcome measure, it captures whether the learning
of basic skills in primary education was successful.
Although young girls had virtually caught up with boys in terms of primary school
enrolment by 1871, there are still significant differences in literacy in the adult population.
As the Prussia-wide cross-tabulations of literacy by gender and denomination of Table 6
show, Protestant women had even higher literacy than Catholic men. The numbers imply
that the GPI of literacy rates, i.e. the literacy rate of females divided by the literacy rate of
males, is larger for Protestants (94.9) than for Catholics (92.6).
An interesting exception to the general pattern can be observed in predominantly
Protestant regions. When looking at literacy rates in counties with more than 80%
Protestants, Catholic women in Protestant areas seem to profit from the better school
supply.5 Although they do not reach the level of the Protestant women there, they nearly
reach the level of Catholic men.
Regression analysis allows us to analyze the gender education gap more closely and to
control for demographic and geographic factors that might influence it (Table 7). The result
that a larger share of Protestants reduced the gender gap in literacy is very robust in the
different specifications. The specification of column (1) contains a set of demographic
control variables derived from the 1871 Population Census. They include age structure,
gender, native population, household and county size, and recent population growth, which
is included to capture possible effects of the Franco-Prussian war of 1870/71. In column
(2), we add geographic control variables: latitude and longitude of the county capitals and
their interaction, as well as the fraction of the population in the county that lives in towns.

5
Becker and Woessmann (2009) show that, similar to the results for 1816 presented above, Protestant
areas have a higher school density also in the second half of the 19th century. In 1886, the fraction of pupils
with distance to school of more than 3 kilometers is significantly lower in Protestant areas; see data appendix
for details on the 1886 Education Census.

15
To control for the stage of economic development, we use two measures: the fraction of
workers in manufacturing and services (column (3))6 and per capita income (column (4)),
proxied by the per capita income of primary school teachers.7 Using teacher income as a
proxy may actually “over-adjust” for income to the extent that better paid teachers may be
directly responsible for higher literacy rates.
The results across all specifications clearly show that the gender literacy gap
significantly decreases with an increasing share of Protestants in the population. To judge
the size of the effect, note that the dependent variable, the fraction of females among the
literate, varies between 43.8% and 49.9% across the 452 Prussian counties as of 1871 (cf.
Table A.3). The first decile is at 46.7% and the ninth decile is at 49.6%. A point estimate of
0.007 thus corresponds to the following thought experiment: If a county went from an all
Catholic to an all Protestant population, the fraction of females among the literate would
increase by two-thirds of a percentage point, which is quite sizeable, considering the
relatively small variance in the outcome.
In column (5), we again use distance to Wittenberg as an instrument for the fraction of
Protestants in a county. The point estimate on the fraction of Protestants drops slightly to
0.006 and remains clearly significant. We conclude that, in 1871, Protestantism narrows the
gender literacy gap. The smaller adult gender literacy gap in Protestant areas may indeed be
the direct consequence of better access to primary schools for Protestant women in earlier
decades, as evidenced by our 1816 results above.

V. Conclusions and Outlook

This paper traces the development of female education in what is generally viewed as
the role model of educational reforms, Prussia, over the 19th century. We suggest that
Protestantism helped to reduce the gender gap in basic education, because Martin Luther
urged each town to have a girls’ school so that girls would learn to read the Gospel.

6
Occupation measures stem from the 1882 Occupation Census; see data appendix for details.
7
Income of primary school teachers was reported in the 1886 Education Census; see data appendix for
details.

16
Sporadic evidence from the time of the Reformation confirms that the Reformation did
indeed lead to a surge in the building of girls’ schools in Protestant areas.
Our empirical analysis uses the exogenous variation in Protestantism stemming from a
county’s or town’s distance to Wittenberg to show that Protestantism led to a decrease in
the gender gap in basic education in 1816 as well as in the gender gap in adult literacy in
1871. The evidence reveals that there was a response to Luther’s postulation of female
education by Protestant authorities and – in the case of middle schools – private entities to
build girls’ schools and by Protestant parents to send their girls to school. Together, this
helped to reduce the gender gap in educational enrolment in early 19th century Prussia and
in literacy in late 19th century.
Our result of a significant causal effect of Protestantism on the female education gap
adds to the literature on the historical relevance of religious factors in education outcomes
(e.g., Botticini and Eckstein 2005, 2007; Becker and Woessmann 2009) by highlighting the
gender dimension. Given the religious norm stressed in Botticini and Eckstein’s important
work that literacy transmitted from father to son in Judaism, this dimension may well be an
important difference in the otherwise similar inheritance of religion-led educational
advancement in Jewish and Protestant history.8
The growing evidence that women’s education is instrumental for important economic
and non-economic outcomes beyond its own sake – e.g., for the economic role of women
(Goldin 2006) and for the health of children (Currie and Moretti 2003) – provides
additional relevance for the finding that women’s education is affected by historical causes
of religious denomination.
While the motivation of this study is mostly historical in nature, the result may have
wider relevance for the developing world today. In terms of UNESCO’s “Education for
All” development goals (UNESCO 2000) that emphasize gender parity, our evidence
suggests that cultural factors may constitute a relevant factor hindering equal access to
education for both genders, as in the case of Catholics in 19th century Prussia. This enforces

8
While this paper concentrates on Luther’s role in advancing female (and male) human capital
accumulation in general, it is an unfortunate legacy of Luther’s embarrassing late writings on Jews that they
are partly responsible for the expulsion of Jews, leading to a serious loss of human capital.

17
the importance of strict enforcement of compulsory schooling to ensure gender parity in
those countries where women are traditionally disadvantaged.
This is in line with the Prussian experience, where we show that compulsory schooling
regulations managed to close the educational gender gap in primary education over the
course of the 19th century. However, it is interesting to note that the pattern of effects of
religious denomination on the gender gap in primary school enrolment and adult literacy
observed in 19th century Prussia continues in secondary and tertiary education in the 20th
and 21st century. For instance, in the first year when women were admitted to university in
Prussia in 1908, there were more than eight times as many female students of Protestant
denomination than of Catholic denomination (359 vs. 43; see Preussische Statistik 1910, p.
71), compared to a proportion of roughly two to one in the total female population.
Protestant women stayed ahead of Catholic women in West Germany even after World War
II. In 1951/52, 59% of female university students were Protestant, clearly exceeding their
population share of 52%. For male students, the denominational difference was less
pronounced, with 56% of male students being Protestant (see Herder-Korrespondenz 1954).
Even in contemporary Germany, according to data from the German Socioeconomic Panel,
Protestant women continue to stay ahead of Catholic women in educational attainment. It is
an interesting topic for future research to establish how the Protestant lead in female
education perpetuated across the education levels.

18
References
Barrett, David B., George T. Kurian, Todd M. Johnson (2001). World Christian
Encyclopedia. 2nd edition. Oxford: Oxford University Press.
Barro, Robert J., Jong-Wha Lee (2001). International Data on Educational Attainment:
Updates and Implications. Oxford Economic Papers 53 (3): 541-563.
Becker, Sascha O., Ludger Woessmann (2009). Was Weber Wrong? A Human Capital
Theory of Protestant Economic History. Quarterly Journal of Economics 124:
forthcoming.
Botticini, Maristella, Zvi Eckstein (2005). Jewish Occupational Selection: Education,
Restrictions, or Minorities? Journal of Economic History 65 (4): 922-948.
Botticini, Maristella, Zvi Eckstein (2007). From Farmers to Merchants, Conversions and
Diaspora: Human Capital and Jewish History. Journal of the European Economic
Association 5 (5): 885-926.
Bunkowske, Eugene W. (1985). Was Luther a Missionary? Concordia Theological
Quarterly 49 (2-3): 161-179.
Card, David (1999). The Causal Effect of Education on Earnings. In: Orley Ashenfelter,
David Card (eds.), Handbook of Labor Economics, vol. 3A, pp. 1801-1863. Amsterdam:
North-Holland.
Currie, Janet, Enrico Moretti (2003). Mother’s Education and the Intergenerational
Transmission of Human Capital: Evidence from College Openings. Quarterly Journal of
Economics 118 (4): 1495-1532.
Engelsing, Rolf (1973). Analphabetentum und Lektüre: Zur Sozialgeschichte des Lesens in
Deutschland zwischen feudaler und industrieller Gesellschaft. Metzler.
Galloway, Patrick R., Eugene A. Hammel, Ronald D. Lee (1994). Fertility Decline in
Prussia, 1875-1910: A Pooled Cross-Section Time Series Analysis. Population Studies
48 (1): 135-158.
Gawthrop, Richard, Gerald Strauss (1984). Protestantism and Literacy in Early Modern
Germany. Past and Present 104: 31-55.
Glewwe, Paul, Michael Kremer (2006). Schools, Teachers, and Education Outcomes in
Developing Countries. In: Eric A. Hanushek, Finis Welch (eds.), Handbook of the
Economics of Education, vol. 2, ch. 16. Amsterdam: North-Holland.
Goldin, Claudia (2006). The Quiet Revolution That Transformed Women’s Employment,
Education, and Family. American Economic Review 96 (2): 1-21.
Green, Lowell (1979). The Education of Women in the Reformation. History of Education
Quarterly 19 (1): 93-116.
Herder-Korrespondenz (1954). Meldungen aus der katholischen Welt: Eine Konfessions-
statistik der westdeutschen Studentenschaft. Herder-Korrespondenz 1954 (9): 99-101.

19
Holborn, Louise W. (1942). Printing and the Growth of a Protestant Movement in Germany
from 1517 to 1524. Church History 11 (2): 123-137.
Küpper, Erika (1987). Die höheren Mädchenschulen. In: Karl-Ernst Jeismann, Peter
Lundgreen (eds.), Handbuch der deutschen Bildungsgeschichte, vol. 3, ch. 4. Munich:
C.H. Beck.
Luther, Martin (1520). An den christlichen Adel deutscher Nation von des christlichen
Standes Besserung. (To the Christian Nobility of the German Nation Concerning the
Improvement of the Christian Estate.) In: Dr. Martin Luthers Werke: Kritische
Gesamtausgabe. Vol. 6, pp. 381-469, 1888, Weimar.
Luther, Martin (1530). Eine Predigt, daß man Kinder zur Schule halten solle. (A Sermon on
Keeping Children in School.) In: Dr. Martin Luthers Werke: Kritische Gesamtausgabe.
Vol. 30, Part 2, pp. 522-588, 1909, Weimar.
Psacharopoulos, George, Harry A. Patrinos (2004). Returns to Investment in Education: A
Further Update. Education Economics 12 (2): 111-134.
Schleunes, K.A. (1989). Schooling and Society: The Politics of Education in Prussia and
Bavaria 1750-1900. London: St. Martin’s Press.
UNESCO (2000). The Dakar Framework for Action – Education for All: Meeting Our
Collective Commitments. Paris: United Nations Educational, Scientific and Cultural
Organization.
UNESCO (2003). Gender and Education for All - The Leap to Equality - EFA Global
Monitoring Report 2003/4. Paris: United Nations Educational, Scientific and Cultural
Organization.
UNESCO (2006). Strong Foundations: Early Childhood Care and Education – EFA
Global Monitoring Report 2007. Paris: United Nations Educational, Scientific and
Cultural Organization.
Vogt, Annette (1997). Elsa Neumann - erste Promovendin an der Berliner Uni. Berlinische
Monatsschrift 8: 27-32.

20
Appendix: Data Sources
Demographers have found county-level data for 19th century Prussia to be a unique
source of highest-quality data for analyses at a disaggregate level (cf. Galloway, Hammel,
and Lee 1994). We have compiled the county-level data as well as municipality-level data
from respective archives.

1816 Population Census

1816 is the earliest year for which the Prussian Statistical Office, founded in 1805,
collected detailed data at the county and municipality level. It is thus the earliest year which
lends itself to a micro-econometric analysis of education, gender and religion. To our
knowledge, these data have not been used for econometric analyses before. The 1816 data
refer to 172 large and medium-sized9 towns and 357 counties10 in Prussia.
The town data provide information on the number of schools, teachers and students of
the following school types: public primary schools (Öffentliche Elementarschulen), private
primary schools (Privat-Elementarschulen), public (lower) secondary schools (Öffentliche
Buerger- oder Mittelschulen für Söhne und Töchter) for boys and girls, private secondary
schools (Private Bürger- oder Mittelschulen für Söhne und Töchter), as well as upper-
secondary schools (Gelehrte Schulen). For all school types, we have gender-specific
enrolment numbers. In contrast, the county data report only on public primary schools
(Öffentliche Elementarschulen).
The source of the 1816 Population Census data is Mützell (1825).

School Enrolment Statistics, 1822-1864

Between 1822 and 1864, the Prussian Statistical Office published school enrolment
statistics at the province or district (Regierungsbezirk) level only. Those are reviewed in
Engel (1869).

9
The Prussian statistical office does not explicitly state what the threshold is for being classified as a
medium-sized town, but the smallest town (Jülich) had 2,470 inhabitants in 1816.
10
Data on the 357 Prussian counties includes the 172 towns mentioned above.

21
1871 Population Census

The second point in time for which we have detailed county-level education data is
1871. The 1871 Population Census is explicitly the very first census ever to survey literacy
in Prussia.11 Literacy is measured as the ability to read and write among the population
aged 10 years or older. As a measure of educational outcome, literacy may be a more
informative measure of accumulated human capital than standard enrollment data, which
may partly capture years in school that did not lead to effective educational outcomes. The
1871 Population Census also provides data on a host of demographic variables. The source
of the 1871 Population Census data is Preussische Statistik (1874).

1882 Occupation Census

The 1882 Occupation Census collected information on employment and self-

employment across two-digit sectors. We calculate the share of the labor force working in
the manufacturing sector and in the service sector, using the classification provided by the
Prussian Statistical Office to classify the two sectors. The manufacturing sector (sector B in
the 1882 classification) includes mining, construction, and manufacture of metals,
machinery, equipment, chemicals, textiles, paper, leather, food products, and wood. The
service sector (sector C in the 1882 classification) includes trade business, insurance,
transport, lodging, and restaurants. Note that the service sector C does not include servants
and housemaids, nor does it include those working in the public administration and the
military. The source of the 1882 Occupation Census data is Preussische Statistik (1884/85,
Vol. 76b, pp. 232-695 and Vol. 76c, p. 239).

1886 Education Census

The 1886 Education Census collected information mainly on pupil enrolment in

primary schools. It also contains county-level data on the share of pupils who had a
distance to school of more than three kilometers, as well as on the average annual income
of full-time male primary school teachers in a county. Given that teacher incomes were

11
Other parts of the German Empire did not survey literacy in the 1871 census.

22
almost entirely financed from local contributions, they provide a reasonable proxy for the
average income of the county (cf. Schleunes 1989). The source of the 1886 Education
Census data is Preussische Statistik (1889, pp. 2-391).

1908 University Census

The 1908 University Census collected information on student enrolment in the winter
semester 1908/09 in all Prussian universities. 1908 was the year women were admitted to
university. The source of the 1908 University Census data is Preussische Statistik (1910).

Data References

Engel, Ernst (1869). Beiträge zur Geschichte und Statistik des Unterrichts, insbesondere
des Volksschul-Unterrichts, im preussischen Staate. Zeitschrift der königlich
preussischen statistischen Bureaus vol. 9, pp. 99-152.
Mützell (1825). Neues Topographisch-statistisch-geographisches Wörterbuch des
Preussischen Staats. Halle: Karl August Kümmel, 1825.
Preussische Statistik (1874), Die Gemeinden und Gutsbezirke des Preussischen Staates und
ihre Bevölkerung: Nach den Urmaterialien der allgemeinen Volkszählung vom 1.
December 1871, Berlin: Verlag des Königlichen Statistischen Bureaus.
Preussische Statistik (1884/85). Die Ergebnisse der Berufsstatistik vom 5. Juni 1882 im
preussischen Staat, Preussische Statistik vol. 76, Berlin: Verlag des Königlichen
Statistischen Bureaus.
Preussische Statistik (1889). Das gesammte Volkschulwesen im Preußischen Staate im
Jahre 1886, Preussische Statistik vol. 101, Berlin: Verlag des Königlichen Statistischen
Bureaus.
Preussische Statistik (1910). Statistik der Preußischen Landesuniversitäten, Studienjahr
Ostern 1908/09, Preussische Statistik vol. 223, Berlin: Verlag des Königlichen
Statistischen Landesamts.

23
 Figure 1: PROTESTANTISM AND EDUCATIONAL GENDER RATIOS ACROSS EUROPE

IRL GBR
1

DNK
FIN
FRA
HUN DEU SWE
BEL
CHE
.9

POL NLD NOR

Gender Parity Index

ITA

ESP
.8

GRC
.7

AUT
.6

PRT

0 .2 .4 .6 .8 1
Protestant share of population

Gender Parity Index = ratio of years of education in the female population to years of education in the male
population. Data refer to 1970.
Figure 2: PROTESTANTISM IN 19TH CENTURY PRUSSIA

(a) 1816

Wittenberg

(b) 1871

Wittenberg
 Figure 3: GENDER PARITY INDEX OF SCHOOL ENROLMENT, 1822-1864

(a) Primary School

1.00

0.95
Gender Parity Index

0.90

0.85

0.80
1822 1825 1828 1831 1834 1837 1840 1843 1846 1849 1852 1855 1858 1861 1864

(b) Secondary School

1.0

0.8
Gender Parity Index

0.6

0.4

0.2

0.0
1822 1825 1828 1831 1834 1837 1840 1843 1846 1849 1852 1855 1858 1861 1864

Straight line: public schools; dotted line: public and private schools.
Table 1: Diffusion of Schools in Brandenburg in the 16th Century
Number of Number of
Year(s) boys’ schools girls’ schools

Middle Ages until 1539 55 4

1540-1572 76 9
1573-1600 100 45

Data for 102 towns in the Electorate (Kurfürstentum) of Brandenburg.

Source: Green (1979).
 Table 2: Protestantism and the Female Enrolment Gap, County Totals, 1816

% Girls among pupils

Dependent variable: enrolled in public primary schools
(1) (2) (3) (4) (5)

% Protestants .020 .021 .040 .041 .035

(.005)∗∗∗ (.005)∗∗∗ (.006)∗∗∗ (.006)∗∗∗ (.006)∗∗∗
% Girls in age group 8-14 .872 .461 .461 .451
(.143)∗∗∗ (.150)∗∗∗ (.148)∗∗∗ (.147)∗∗∗
Latitude (in rad) * 100 -.006 -.006 -.006
(.003)∗ (.003)∗∗ (.003)∗
Longitude (in rad) * 100 .0002 .008 .010
(.009) (.009) (.009)
Latitude * Longitude * 100 -.0007 -.010 -.012
(.010) (.010) (.011)
% Population living in towns -.023 -.013 -.006
(.010)∗∗ (.011) (.011)
Number of horses per capita .150 .138
(.051)∗∗∗ (.051)∗∗∗
Number of bulls per capita .175 .120
(.815) (.812)
Prim. schools per 1,000 inhabitants .005
(.002)∗
Obs. 289 289 289 289 289
R2 .046 .156 .279 .306 .315

Coefficients and standard errors from ordinary least squares (OLS) estimation.
Standard errors in parentheses: ∗ significance at ten, ∗∗ five, ∗∗∗ one percent.
Data for Prussian counties from the 1816 Census; see main text for details.
 Table 3: Instrumenting Protestantism by Distance to Wittenberg

1st stage 2nd stage

Dependent variable: % Protestants % Girls among pupils in public primary schools
(1) (2) (3) (4) (5) (6) (7) (8)

% Protestants .049 .049 .049 .046

(.011)∗∗∗ (.009)∗∗∗ (.009)∗∗∗ (.012)∗∗∗
Distance to Wittenberg in 100 km -.126 -.190 -.189 -.156
(.014)∗∗∗ (.015)∗∗∗ (.015)∗∗∗ (.015)∗∗∗
% Girls in age group 8-14 -3.284 2.324 2.317 1.804 .906 .416 .417 .415
(1.493)∗∗ (1.273)∗ (1.274)∗ (1.188) (.151)∗∗∗ (.155)∗∗∗ (.153)∗∗∗ (.152)∗∗∗
Latitude (in rad) * 100 -.091 -.093 -.052 -.008 -.008 -.008
(.034)∗∗∗ (.034)∗∗∗ (.033) (.004)∗∗ (.003)∗∗ (.004)∗∗
Longitude (in rad) * 100 -.567 -.571 -.399 -.003 .005 .006
(.108)∗∗∗ (.110)∗∗∗ (.105)∗∗∗ (.010) (.010) (.010)
Latitude * Longitude * 100 .640 .645 .455 .003 -.006 -.008
(.120)∗∗∗ (.122)∗∗∗ (.117)∗∗∗ (.011) (.011) (.011)
% Population living in towns .120 .137 .298 -.024 -.015 -.011
(.090) (.095) (.092)∗∗∗ (.011)∗∗ (.011) (.013)
Number of horses per capita -.320 -.610 .155 .147
(.439) (.411) (.051)∗∗∗ (.052)∗∗∗
Number of bulls per capita 8.951 5.810 .100 .066
(7.016) (6.544) (.821) (.816)
Prim. schools per 1,000 inhabitants .124 .003
(.019)∗∗∗ (.003)
Obs. 289 289 289 289 289 289 289 289
R2 .228 .551 .553 .615 .069 .273 .300 .308
F statistic (instrument) 83.72 157.00 155.25 108.39

Coefficients and standard errors from instrumental variables (IV) estimation.

Standard errors in parentheses: ∗ significance at ten, ∗∗ five, ∗∗∗ one percent.
Data for Prussian counties from the 1816 Census; see main text for details.
 Table 4: The Female Enrolment Gap, Towns, 1816
% girls among pupils % private enrolm. in enrolm. in school type
School type all public private school type as % of total enrolm.
Elementary school 46.5 (8.0) 43.8 (12.7) 55.1 (21.6) 16.9 73.9

Middle school 38.5 (30.9) 28.0 (27.3) 61.1 (36.2) 29.4 18.0

Gymnasium 0.0 (0.0) – – – 8.1

All schools 41.5 (7.2) – – – 100.0

Columns 1-3 display means (standard deviations in parentheses).

Data for 156 Prussian medium-sized and large towns from the 1816 Census; see main text for details.
 Table 5: Protestantism and the Female Enrolment Gap, Towns, 1816

Primary schools Secondary schools

OLS IV OLS IV
1st stage 2nd stage 1st stage 2nd stage
% Girls % Girls % Girls % Girls
Dependent variable: among pupils % Protestants among pupils among pupils % Protestants among pupils
(1) (2) (3) (4) (5) (6)

% Protestants .041 .079 .097 .159

(.020)∗∗ (.034)∗∗ (.061) (.105)
Distance to Wittenberg in 100 km -.187 -.171
(.020)∗∗∗ (.021)∗∗∗
% Girls in age group 8-14 .202 -.598 .213 -.228 -.595 -.203
(.156) (.498) (.159) (.420) (.489) (.423)
Latitude (in rad) * 100 -.019 -.120 -.024 -.002 -.125 -.008
(.012) (.045)∗∗∗ (.012)∗ (.032) (.046)∗∗∗ (.034)
Longitude (in rad) * 100 -.027 -.642 -.036 .007 -.636 -.004
(.035) (.144)∗∗∗ (.036) (.099) (.151)∗∗∗ (.100)
Latitude * Longitude * 100 .029 .712 .039 -.003 .709 .009
(.039) (.160)∗∗∗ (.040) (.110) (.167)∗∗∗ (.111)
% Buildings w/ massive walls -.012 -.057 -.002 .044 -.051 .059
(.025) (.081) (.026) (.075) (.089) (.078)
Businesses per capita .059 .603 .067 -.681 -.116 -.618
(.118) (.384) (.119) (.361)∗ (.428) (.372)∗
Looms per capita -.011 .378 -.023 .130 .377 .109
(.070) (.220)∗ (.071) (.186) (.215)∗ (.189)
Retailers per capita -.592 -1.878 -.566 4.138 -.865 4.094
(.651) (2.072) (.659) (1.805)∗∗ (2.109) (1.813)∗∗
Obs. 156 156 156 141 141 141
R2 .108 .534 .087 .100 .506 .093
F statistic (instrument) 90.06 67.40

Coefficients and standard errors from ordinary least squares (OLS) and instrumental variables (IV) estimation.
Standard errors in parentheses: ∗ significance at ten, ∗∗ five, ∗∗∗ one percent.
Data for Prussian medium-sized and large towns from the 1816 Census; see main text for details.
 Table 6: Literacy Rates by Denomination and Gender, 1871
Males Females
all Prussian counties
Protestants 93.36 88.72
Catholics 84.75 78.48

counties with >80% Protestants

Protestants 93.79 89.23
Catholics 88.35 87.95

% Literate among those aged ≥ 10.

Data for Prussian counties from the 1871 Population Census; see main text for details.
 Table 7: Protestantism and the Female Literacy Gap, 1871

OLS IV

(1) (2) (3) (4) (5)

% Protestants .008 .009 .007 .007 .006

(.001)∗∗∗ (.001)∗∗∗ (.001)∗∗∗ (.001)∗∗∗ (.003)∗∗
% Missing education info .021 .019 .020 .025 .025
(.038) (.037) (.038) (.037) (.039)
% Age below 10 -.162 -.139 -.157 -.143 -.162
(.019)∗∗∗ (.020)∗∗∗ (.019)∗∗∗ (.019)∗∗∗ (.019)∗∗∗
% Jews -.174 -.193 -.174 -.188 -.185
(.034)∗∗∗ (.035)∗∗∗ (.033)∗∗∗ (.033)∗∗∗ (.038)∗∗∗
% Females -.105 -.012 -.075 -.059 -.109
(.036)∗∗∗ (.040) (.037)∗∗ (.037) (.036)∗∗∗
% Born in municipality .028 .005 .024 .025 .026
(.004)∗∗∗ (.006) (.004)∗∗∗ (.004)∗∗∗ (.006)∗∗∗
% Of Prussian origin -.064 -.032 -.060 -.063 -.067
(.021)∗∗∗ (.022) (.021)∗∗∗ (.021)∗∗∗ (.022)∗∗∗
Average household size -.019 .044 .095 .055 -.074
(.151) (.161) (.155) (.150) (.174)
ln(Population size) -.137 -.052 -.192 -.200 -.134
(.103) (.103) (.105)∗ (.103)∗ (.104)
Popul. growth 1867-1871 (in %) .016 .005 -.0009 .006 .012
(.011) (.012) (.013) (.011) (.013)
Geographic controls yes

% of labor force in manu and serv .011

(.004)∗∗∗
ln(Per capita income) 1.071
(.292)∗∗∗
Obs. 452 452 452 452 452
R2 .556 .587 .564 .570 .554

Dependent variable: fraction of females among the literate.

Standard errors in parentheses: ∗ significance at ten, ∗∗ five, ∗∗∗ one percent.
Column 5 uses distance to Wittenberg as an instrument for the fraction of Protestants in county.
Data for Prussian counties from the 1871 Population Census and the 1882 Occupation Census; see main text and appendix for details.
Further controls: % Blind, % Deaf-mute, % Insane.
 Table A.1: Descriptive Statistics: Prussian Counties, 1816
Mean StdDev Min Max
(1) (2) (3) (4)

% Girls among public prim. school pupils 47.13 3.77 23.04 57.57
% Protestants 59.30 40.72 .00 99.96
% Girls in age group 8-14 49.13 1.42 44.67 53.93
Latitude (in rad) * 100 90.96 2.45 85.93 97.24
Longitude (in rad) * 100 23.81 8.61 10.52 39.40
% Population living in towns 11.56 19.58 .00 100.00
Number of horses per capita .10 .06 .00 .35
Number of foals per capita .02 .02 .00 .11
Number of bulls per capita .005 .003 .00 .03
Number of oxen per capita .07 .06 .00 .25
Number of cows per capita .22 .07 .00 .60
Number of young cattle per capita .12 .06 .00 .49
Number of sheep per capita .87 .66 .00 3.10
Number of goats per capita .02 .02 .00 .12
Number of pigs per capita .15 .10 .00 .44
Prim. schools per 1,000 inhabitants 1.97 .94 .00 4.86
Distance to Wittenberg in km 340.12 164.43 .00 731.46

Number of observations: 289.

Data for Prussian counties from the 1816 Census.
 Table A.2: Descriptive Statistics: Prussian Towns, 1816
Mean StdDev Min Max
(1) (2) (3) (4)

% Girls among prim. school pupils 46.50 8.05 12.20 69.70

% Girls among second. school pupils 21.65 21.14 .00 100.00
% Protestants 69.53 35.39 .11 100.00
% Girls in age group 8-14 50.45 4.05 42.28 79.64
Latitude (in rad) * 100 90.77 2.21 85.93 97.24
Longitude (in rad) * 100 22.37 7.53 10.52 38.75
% Buildings w/ massive walls 23.85 27.88 .00 100.00
Businesses per capita .19 .06 .07 .45
Looms per capita .03 .09 .00 1.04
Retailers per capita .02 .01 .003 .06
Distance to Wittenberg in km 290.24 151.43 30.15 708.50

Number of observations: 156.

Data for Prussian medium-sized and large towns from the 1816 Census.
 Table A.3: Descriptive Statistics: Prussian Counties, 1871
Mean StdDev Min Max
(1) (2) (3) (4)

% Females among the literate 48.45 1.16 43.82 49.89

% Protestants 64.18 37.83 .26 99.89
% Missing education info 1.69 1.10 .00 6.72
% Age below 10 24.71 2.48 15.33 29.87
% Jews 1.14 1.33 .00 12.87
% Females 51.00 1.51 43.97 54.63
% Born in municipality 58.97 12.39 32.01 87.23
% Of Prussian origin 99.07 1.97 74.22 100.00
Average household size 4.79 .34 3.83 5.86
ln(Population size) 10.80 .41 9.36 13.62
Popul. growth 1867-1871 (in %) 1.60 4.93 -7.76 33.83
% Blind .09 .03 .03 .24
% Deaf-mute .10 .05 .02 .42
% Dumb .23 .17 .02 1.56
Distance to Wittenberg in km 326.19 148.77 .00 731.46

Number of observations: 452.

Data for Prussian counties from the 1871 Population Census; see main text and appendix for details.