Journal of Economic Literature 2011, 49:3, 589–614

http:[Link]/[Link]?doi=10.1257/jel.49.3.589

The Historical Fertility Transition:

A Guide for Economists
Timothy W. Guinnane*

The historical fertility transition is the process by which much of Europe and North
America went from high to low fertility in the nineteenth and early twentieth centu-
ries. This transformation is central to recent accounts of long-run economic growth.
Prior to the transition, women bore as many as eight children each, and the elasticity
of fertility with respect to incomes was positive. Today, many women have no children
at all, and the elasticity of fertility with respect to incomes is zero or even negative.
This paper discusses the large literature on the historical fertility transition, focusing
on what we do and do not know about the process. I stress some possible misun-
derstandings of the demographic literature, and discuss an agenda for future work.
( JEL I12, J13, N30)

1.  Introduction transition forms a major part of the process

that brought the European, and later other,

P rior to the historical fertility transition,

which took place in most of Europe
and North America between the late eigh-
economies from slow to rapid and sustained
growth. This article discusses the empirical
literature on the historical fertility transition:
teenth and early twentieth centuries, married what we know and what we still need to learn.
women could expect eight or more births. Current interest in the broad area of
Today many women do not have children demographic behavior and long-run growth
at all. For centuries, the elasticity of fertil- reflects several intellectual influences, includ-
ity with respect to incomes was positive. ing Robert E. Lucas’s lectures on growth
Today it is zero or even negative. The fertility (Lucas 2002). Early efforts in this area usually

* Yale University. My research on this and related topics Ogilvie, Barbara Okun, Cormac Ó Gráda, Jean-Laurent
has been generously funded by the National Institutes of Rosenthal, Mark Rosenzweig, Paul Schultz, William
Health (U.S.), the Leverhulme Trust (U.K.), and the Eco- Sundstrom, Nathan Sussman, James Trussell, Christopher
nomic and Social Research Council (U.K.). This paper has Udry, and David R. Weir. For comments and suggestions
been presented in seminars at Brown University, the Uni- on this paper, I thank the editor and three referees, as
versity of Colorado (Boulder), the University of Santiago well as Robert Allen, Stanley Engerman, James Fenske,
de Compostela, and the Autonomous University of Barce- David Mitch, Tommy Murphy, and Marianne Wanamaker.
lona. I have learned much about these issues from work- My interest in this subject is due to Paul David, and any
ing with and talking to John Brown, Christoph Buchheim, (good) ideas here probably reflect his influence more than
Oded Galor, Carolyn Moehling, Thomas Mroz, Sheilagh is reflected in the citations.

589
590 Journal of Economic Literature, Vol. XLIX (September 2011)

embed the microeconomic model of fertility stresses the implications of ­ demographic

decisions developed by Gary S. Becker in a change for economic growth and develop-
framework that allows feedbacks from the ment. That issue lies beyond the scope of
economy to fertility decisions. Perhaps the this paper. Second, economists typically
leading paper in this area is Becker, Kevin M. think of fertility decline as the reduction in
Murphy, and Robert Tamura (1990). A more the number of children born to a woman
recent body of research focuses explicitly on or to a couple. Demographers and oth-
how economies transition from a “Malthusian” ers stress heterogeneity in the way fertil-
economy of high fertility and little growth in ity declines, for example, whether couples
per capita incomes to one in which fertility reduce the number of surviving offspring by
is much lower and per capita incomes grow spacing their childbearing or ending child-
rapidly. This “Unified Growth Theory” (here- bearing before that is biologically necessary.
after UGT), developed by Oded Galor, David I discuss this issue only where important to
N. Weil, and others, integrates a microeco- understanding the evidence on the fertil-
nomic model of the demand for children with ity transition. Finally, the fertility transition
feedbacks from a model of growth. Other, in developing countries since World War
related literature either considers fertility and II has been studied far more intensively
growth in a different way, or examines related than its historical counterpart. This paper
­questions such as the role of children in the focuses on the earlier, historical episode,
labor market. which is most relevant to what theorists
These discussions have provoked renewed have in mind when modeling the Industrial
interest in the empirics of the historical fer- Revolution.
tility transition: when and why it took place,
and how rapidly it created the low fertil-
2.  The Basic Contours of the Historical
ity we now see in most wealthy countries.
Fertility Transition
Growth theorists are to be commended
for trying to explain this historical fact and Figure 1 reports fertility experience for
for highlighting the importance of issues the period 1800–1970 for five major coun-
that were previously of interest primarily tries: France, England and Wales, Germany,
to economic historians and to others, such the United States, and Italy. The measure
as demographers, outside the economics reported in figure 1 is the Crude Birth Rate
profession. Unfortunately, some recent dis- (CBR), defined as the number of births per
cussions misconstrue the demographic lit- thousand per annum. This paper focuses
erature. The use of certain theoretical ideas on the first four countries; Italy is included
has also led some to believe that the rele- in figure 1 only to suggest the heteroge-
vant propositions have actually been tested neity of historical experience. Figure 2
in historical contexts, when that is unfortu- focuses on the single case of Germany to
nately often not the case. Fully understand- highlight the relationship between fertil-
ing the fertility transition will require much ity and mortality decline.1 Ignoring the
more empirical research. This paper has heterogeneity in figure 1 for the moment,
two goals. I provide an overview of the his- we see fertility declining starting in the
torical fertility transition, and then discuss
the main hypotheses that are current in the
economics and economic history literature. 1 Deaths in figure 2 are reported as the Crude Death
Throughout, I set aside three other issues. Rate (CDR), defined analogously to CBR. The Crude Rate
First, most current e­conomic research of Natural Increase (CRNI) is defined as CBR – CDR.
 Guinnane: The Historical Fertility Transition 591

50

40
Births per 1000 population

30

20

10

England France Germany Italy USA

0
1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970
Figure 1. Crude Birth Rates, Selected Countries, 1820–1970
Note: For the United States, values before 1909 are linear interpolations between decennial census years.
Source: Crude birth rates as reported in Mitchell (1980).

eighteenth or nineteenth century, and this to sustain population growth through natural
decline accelerates in the second half of increase alone.2 The relationship between
the nineteenth century. The two world wars fertility and mortality declines differs across
produced ­dramatic, t­ emporary reductions in these countries, but the German experience
­fertility, and the post–World War II period
saw a “Baby Boom.” By the 1970s, most of
2 Hans-Peter Kohler, Francesco C. Billari, and José
Western Europe experienced the emer-
Antonio Ortega (2002) discuss the emergence of very low
gence of “low–low” fertility. Today, some fertility in the 1990s; Joshua R. Goldstein, Tomáš Sobotka,
OECD countries have fertility rates too low and Aiva Jasilioniene (2009) discuss a recent, partial reversal.
592 Journal of Economic Literature, Vol. XLIX (September 2011)

45

40

35

30

25

20

15

10

57
47

87
37

67
47

57

77

27

37

77

97
17

27

87

97

17
07

67
19
19

19
18

18
18

18

18

19

19

19

19
18

18

18

18

19
19

19
Figure 2. Fertility and Mortality in Germany
(Number of events per thousand population)
Source: Crude birth rates as reported in Mitchell (1980).

detailed in figure 2 is fairly typical. For view that the fertility transition took place
most of the nineteenth century, birth rates all at once across Europe. Most economists
exceeded death rates and population grew would not admire the data and methods
(even with, as in the German case, extensive upon which that assertion is based.3 The
emigration). This is no longer true; for much ­all-at-once view has important implications
of the late twentieth century, German rates for causality. Some scholars invoke the claim
of natural increase were negative. of ­simultaneous fertility transitions to sup-
Most countries outside Europe and port their view that economics has little to
North America did not experience fertility do with the fertility transition:4 “Clearly the
declines until after World War II. Timing simultaneity and speed of the European
within Europe and North America is less transition makes it highly doubtful that
clear. Some economists have accepted the

3 Galor (2005a, footnote 33) accepts this view, for exam- and Susan Cotts Watkins (1986) for definitions. Ig does
ple. Galor is citing the results of a large project under- not perform as claimed in Monte Carlo studies (Timothy
taken at Princeton University in the 1960s and 1970s. W. Guinnane, Barbara S. Okun, and James Trussell 1994).
The Princeton conclusion reflects serious problems with This index’s performance is crucial to the claim that fertility
sources, measures, and econometrics. The Princeton proj- started to decline everywhere at once. John C. Brown and
ect devised a series of indices of overall fertility, marital Guinnane (2007) critique the statistical exercises reported
fertility, illegitimate fertility, and the contribution of mar- in that project’s publications.
riage patterns to fertility. The index of marital fertility, Ig, 4 Thus it is puzzling to find an economic historian claim-
is scaled such that it would equal one if the population in ing, counter to the evidence, that “.  .  . the timing of the
question had fertility equal to the very high level of the demographic transition in Europe and the United States
Hutterites in the 1920s. See appendix B to Ansley J. Coale places it circa 1890” (Gregory Clark 2007, p. 225).
 Guinnane: The Historical Fertility Transition 593

any economic force could be found which ­ easure ­

m available or, as in my figure 1,
was powerful enough to offer a reasonable the only measure available that is defined
explanation” (John Cleland and Christopher consistently across a set of countries for
Wilson 1987, p. 18). the right period. Yet it has serious weak-
Figures 1 and 2 report national aggre- nesses, especially applied to this period of
gates. Several studies document the exis- rapid economic and demographic change.
tence of fertility control among small The CBR does not account for age-struc-
groups as early as the seventeenth century. ture or marriage patterns, both of which
These “forerunners” were usually urban changed significantly in some European
elites or members of minority groups such populations during this period. Consider
as Jews (Massimo Livi-Bacci 1986). More England and Wales, where from 1871 to
generally, research based on either sub-
­ 1911 there was a disproportionate increase
national aggregates or micro data often find in the number of women of child-bearing
earlier fertility declines than in national age. The overall population increased by
data. The Princeton studies report earlier about 60 percent, while the female popula-
fertility declines in cities, for example, but tion aged 15–44 increased by 77 percent.
stress the behavior of national populations. The shift in age-structure in the absence of
John E. Knodel (1988) reports that couples a fertility reduction would have increased
married 1800–24 in five of the fourteen the CBR by 11 percent. By failing to take
German villages he studied exhibited sig- into account the changing population age-
nificant control of marital fertility. structure, the CBR understates the change
There are more general reasons for skep- in behavior.6 Some European societies
ticism about simultaneous fertility transi- experienced significant changes in mar-
tions. Most scholars agree that France’s riage patterns during this period that make
fertility decline started in the early nine- the CBR a poor guide to marital fertility.
teenth century at the latest. David R. Weir’s For example, between 1871 and 1911, the
(1994, table B3) careful construction of the proportion of English women who had
Princeton indices shows that the decline in never married by ages 25–29 rose from
French fertility began in the late eighteenth 0.37 to 0.44 (John Hajnal 1953, table 3).
century. In the early nineteenth century, Mean age at marriage for English males
U.S. birthrates were higher than elsewhere, rose over the same period by 1.2 years (to
but falling rapidly. By the end of the nine- 27.6), while at the same time in Germany it
teenth century, U.S. fertility fell below that fell by nearly 1 year (to 27.9). The changes
of much of Western Europe. France and the for women are similar but slightly less
United States are not esoteric exceptions one pronounced (Josef Ehmer 1991, table 1).7
can ignore.5
Another concern reflects the CBR so
often relied upon in economics research. 6 This precise effect is not present in every population
In many circumstances, it is the only represented in figure 1. This is the point. The figures in
question come from the decennial censuses of England
and Wales for 1871 and 1911, and the Registrar-General’s
reports on births, deaths, and marriages. I have taken them
5 J. David Hacker (2003) questions whether the United from B. R. Mitchell (1980, series B1, B2, B5, and B6).
States really had an early fertility decline. He is right to 7 Cross-sectional differences in marriage patterns at the
stress the weakness of the sources for the early nineteenth time were especially large. In 1900, England and Ireland
century, but the methods he prefers are fragile in their own had about the same CBR, but the latter was achieved by
way, and even he suggests a U.S. fertility transition well relatively few couples having very large familes (Guinnane
before the late nineteenth century. 1997).
594 Journal of Economic Literature, Vol. XLIX (September 2011)

4
Cohort fertility rate

1
England France Germany Italy USA

0
0

0
0

0
0

18 890

91 0

26 0

31 5
11 5

21 5

5
5

51 5

61 5

71 5

81 5

01 5
18 186

18 190

19 191

19 194
18 188

19 192
84

85

18 187

18 189

19 193

19 193
19 191

19 192

94
84

18 185

18 186

18 187

18 188

19 190
–1

–1

–1

–1
–1

–
–

–
–

–
–

–
–

–
31

41

86

06

16

36
36

46

56

66

76

96
18

18
18

18

Figure 3. Cohort Fertility Rates, 1831–1945

Notes: The cohort fertility rate is the mean number of children born to women belonging to the birth cohorts
on the horizontal axis. The overlapping years are in the source. The precise birth cohorts vary slightly across
countries.
Sources: Festy (1979): for England, p. 262; for France, pp. 266–67; for Italy, p. 283; for the United States,
p. 290; and for Germany, p. 222. Marschalck (1984), table 3.6, for Germany for the years 1901–1945.

Figure 3 reports a better measure, the But figure 3 suffices to suggest that the
Cohort Fertility Rate (CFR) for the five experience of specific cohorts born in the
countries in figure 1. CFR is the average nineteenth century is not well-captured by
number of children born to women in a CBR.8 What looks like constant fertility (as
given cohort, and thus requires age-specific measured by the CBR) may well reflect two
fertility for a cohort’s entire reproductive offsetting trends. Figure 4 uses the excep-
life. Its cross-section analogue, the Total tional French data to illustrate just this
Fertility Rate (TFR), is the sum of current point: France’s precocious marital fertility
age-specific fertilities, and thus reports the decline was partially offset by a mid-century
number of children born to a woman who marriage boom.
experiences current age-specific fertility
throughout her life. Reliable data needed
8 TFR estimates for the United States start at 7.04 in
to compute the CFR do not go back very
1800 and fall to 3.56 by 1900. These figures, just like the
far into the nineteenth century, and are CBR in figure 1, are for the white population only (Michael
not available even then for most countries. R. Haines 2000a, table 4.3).
 Guinnane: The Historical Fertility Transition 595

0.9000

0.8000

0.7000

0.6000
Indices

0.5000

0.4000

0.3000

0.2000
Fertility Marital fertility Extramarital fertility Marriage
0.1000

0.0000
1740 1760 1780 1800 1820 1840 1860 1880 1900
Year

Figure 4. French Fertility, 1740–1910

(Indices, max = 1.0)
Note: “Fertility” is the Princeton index of overall fertility If ; marital fertility is the Princeton index Ig; extra-
marital fertility is the Princeton index Ih; and “marriage” is the Princeton index Im, which measures the contri-
bution of changing marriage patterns to overall fertility.
Source: Weir (1994).

3.  From Malthus to Becker the product of capital accumulation and

­technological change, no longer translated
Most of the literature uses Malthusian into higher fertility.
models to understand the relationship Two distinct approaches to the Malthusian
between population and the economy before world exist in the literature. An older ver-
the fertility transition, and views the fertility sion stresses Malthus’s own argument, that
transition as an escape from the Malthusian is, that the regulation of births reflected
world. Figure 5 shows that escape: until the the regulation of marriage. In this model,
early nineteenth century, there was a posi- ­marital f­ ertility depends only on age at mar-
tive relationship between fertility and the riage and the proportions who marry. The
real wage in England and Wales. Then the lifetime fertility of any given woman is a
relationship breaks down: higher wages, stochastic function of her age at marriage.
596 Journal of Economic Literature, Vol. XLIX (September 2011)

Central to this version of the Malthusian s­ hort-run models confirm the central role of
model is the “European marriage pattern” nuptiality in regulating fertility.11
in which young adults deferred marriage
until well after puberty, often into their 3.1 The Demand for Children
middle and late twenties, and as much as
ten to twenty percent of some cohorts never Virtually all economic analysis of fertil-
married at all. Couples could not marry ity today starts from Becker’s model of the
until they could support themselves and demand for children.12 Becker’s insight was
their offspring, implying that marriage deci- to analyze the demand for children using the
sions depended on the real wage in young tools of consumer choice. The model yields
adulthood.9 In this version of the Malthusian important insights for the fertility transition.
model, the fertility transition reflects a shift Observers have long noted that fertility tends
from controlling marriage to controlling to be negatively correlated with income in
­fertility within marriage. More recently, the cross-section, and, since the beginnings
many economists (and much of the recent of the fertility transition, over time. Becker’s
growth theory) set aside the marriage issue model implies that this is a standard substi-
and thus model fertility without concern for tution effect, that children are not inferior
its underlying determinants.10 goods: wealthier couples have higher oppor-
Estimating long-run versions of the tunity costs of time, and time is a major
Malthusian model poses serious challenges. cost of child-rearing. The simple version of
It has three equations and three endoge- Becker’s model starts with a household utility
nous variables, and because of echo effects function U = U(n, Z), where n is the number
due to past population shocks, the demo- of children and Z is a vector of all other com-
graphic fluctuations of today may represent modities. The household maximizes this util-
the effects of past shocks. Another litera- ity subject to a standard budget constraint.
ture avoids these problems by estimating Increases in child costs induce substitution
the short-run elasticities of births, mar-
riages, and deaths to real-wage shocks. The
11 The age-structure change in England noted above
is one such echo effect. Ronald D. Lee and Michael
Anderson (2002) rely on a state-space representation to
contend with the integration and endogeneity problems
9 Most studies find a substantial proportion of preg- implied by the model. This paper presents a clear overview
nant brides, but children actually born outside of mar- of earlier literature as well as the modeling problems at
riage rarely accounted for more than 5 percent of all issue. More recently, Esteban A. Nicolini (2007) relies on
births. For the European marriage pattern, see Hajnal a simple VAR model that he identifies by an assumption
(1965, 1982). Most scholars accept Hajnal’s European of “contemporaneous stickiness.” Niels Framroze Møller
Marriage Pattern as a stylized account, but it is not and Paul Sharp (2008) are unusual in the recent literature
clear which parts of Europe it describes. E. A. Wrigley in explicitly modeling marriage rates. There are also data
and R. S. Schofield’s (1981) book is an extended argu- reasons to prefer the short-run model: often for historical
ment that England’s demographic system functioned situations we know the number of events (births, deaths,
in a particularly benign way because of the strength marriages) but not the population size, and thus cannot
of the relationship between the real wage and mar- compute demographic rates. Lee (1981, 1985) explains
riage patterns. Earlier studies had stressed mortal- the logic of the short-run models. Weir (1984) used this
ity changes as the driving force in English population approach to challenge Wrigley and Schofield’s interpre-
history. Malthusian models in Chinese population his- tation of their English evidence. Guinnane and Sheilagh
tory are more controversial, partly because Chinese Ogilvie (2008) apply this approach to some German vil-
marriage patterns appear to have been different. lages for the period 1634–1870 and provide references to
James Z. Lee and Wang Feng (1999) provide an over- other efforts of this sort.
view of the debate and references to the literature. 12 The important paper references are Becker (1960)
10 This vision underlies Galor and Weil (2000), as well as and Becker and H. Gregg Lewis (1973). Becker (1981,
most other UGT models. chapter 5) is a more elegant and expansive exposition.
 Guinnane: The Historical Fertility Transition 597

England, 1541–1871
9
40

Wage 11-year MA
CBR 11-year MA

35

30
6

CBR 11-year MA Wage 11-year MA

25 5

1500 1600 1700 1800 1900

Year

Figure 5. Crude Birth Rates and Real Wages

Notes: This figure differs from Wrigley and Schofield (1981, figure 10.1) in two ways. I plot the CBR, not the
Gross Rate of Reproduction. The real wage index here is Robert Allen’s “labourers” index, rather than the
Phelps Brown-Hopkins index. The series plotted are centered eleven-year moving averages. Allen’s index can
be found at: [Link]
Source: Based on Wrigley and Schofield (1981), figure 10.6.

away from children and toward the Zs. A number of children and their quality, usually
pure increase in income raises the number called the Quantity/Quality or “Q–Q” model.
of children demanded, as we expect. But (Child quality is usually taken to mean child’s
if that higher income reflects rising wages, health or education.) The Q–Q model starts
then that increased wage may show up as from a household utility function of the form
an opportunity cost of having children, and U(n, q, Z), where q is the quality of each
reduce the number of children demanded child. Becker (1981, pp. 107–08) divides
via the substitution effect.13 child costs into three categories. Some costs
Later interest in Becker’s model focuses depend only on the number of children:
on the possible trade-off between the an example of this, pn, would be the costs
associated with the mother’s pregnancy and
delivery.14 Another cost is related to child
13 In a number of articles often called collectively the
“Easterlin synthesis,” Richard A. Easterlin integrated
Becker’s approach with a better appreciation of the costs of
fertility control as well as the biological limits on reproduc- 14 This discussion follows Becker (1981) and uses his
tion. See especially Easterlin (1978). notation.
598 Journal of Economic Literature, Vol. XLIX (September 2011)

quality, but does not depend on the number finding seems consistent with the central
of children, as it goes to purchase household trade-off in the Q–Q model. Hoyt Bleakley
public goods: examples of pq include books and Fabian Lange (2009) take a different
that children could share. A final cost, pc, approach. They examine a program that
is the cost of augmenting the quality of any largely eradicated intestinal worms among
child. The household’s budget constraint is children in the early twentieth-century U.S.
then: South. They find that the reduction in the
prevalence of worms, which they interpret as
pq q  +  pc nq + πz Z  =  I, a shock to the cost of child quality, increased
school enrollment rates and reduced fertil-
ity. This again is consistent with the Q–Q
where I is household income and πz is the model.15
price of the Zs. The marginal rate of sub-
stitution between quantity and quality now
4.  Explanations and Evidence
depends on the ratios of fixed to variable
costs for quantity and quality respectively, as We can group the many economic expla-
well as on the ratio of marginal variable cost nations offered for the historical fertility
to the average variable cost of quality. The transition under six headings. The first is an
substitution effects between quantity and exogenous decline in infant and child mortal-
quality are stronger in the Q–Q model than ity. The second turns on innovations in the
in the original Becker model. Consider an technology of contraception, or more wide-
increase in pn. The household will substitute spread availability of contraceptive devices.
away from numbers to both child quality and The third looks for increases in the direct
Zs, as one would expect. But because of that cost of childbearing. The fourth explana-
interaction term pc nq, the shadow cost of n tion is based on changes in the opportunity
depends on q, so a reduction in child num- costs of child-bearing. The fifth looks for a
bers raises the shadow cost of numbers even net increase in returns to child quality. The
more, inducing more substitution of quality sixth argument assumes that children were
for quantity. an important way to ensure against risk and
This Q–Q model has considerable appeal to provide for old age, and that the rise of
in historical circumstances. We observe state social insurance as well as private insur-
sharp fertility declines that seem to reflect ance and savings vehicles led households to
small changes in the economic environ- substitute out of children. I consider these
ment. For example, as Becker noted himself, explanations in ceteris paribus fashion.
within this model a modest reduction in the
4.1 “Demographic Transition Theory” and
cost of contraception could induce a shift
the Role of Mortality Decline
from n to q, which seems consistent with the
historical evidence. Several recent papers A long tradition assigned to mortality
use historical data to test the Q–Q model. decline a causal role in the fertility transition.
Sascha O. Becker, Francesco Cinnirella, and
Ludger Woessmann (2009) use toughening
of compulsory schooling laws in Prussia in 15 The Q–Q model may get more attention than it war-
the late 1840s to study the fertility effects rants. Mark R. Rosenzweig and Kenneth I. Wolpin (1980)
of a reduction in the price of child quality. show that Becker’s original model (with pc nq constrained
to zero) generates nearly all of the testable implications
In their district-level data, areas with higher that are identified with the Q–Q model. See also T. Paul
enrollment rates had lower fertility. This Schultz (1981, pp.166–69).
 Guinnane: The Historical Fertility Transition 599

Frank Notestein (1945) argued that couples and a five-year-old girl, 57.4 years. A twenty
in high-mortality societies have a lot of births year-old woman could expect a further 40.7
to ensure a surviving brood of the desired years, not even a full year more than in 1850.17
size. An exogenous mortality decline induces The Notestein argument does not fit the
couples to have fewer children because they timing of the historical declines in fertility
do not need so many “spares.” Notestein’s and mortality. Fertility in the United States
account was motivated by the experience declined for decades before any noticeable
of developing countries after World War decline in mortality. The TFRs reported
II, where public-health interventions first by Haines (2000b, table 8.2) decline from
reduced infant and child mortality, and in the early nineteenth century; there is no
some places those developments were fol- sustained fall in the infant mortality rate,
lowed by declines in fertility. Most European on the other hand, until the 1890s. French
countries also experienced a significant mor- experience was similar, with a fertility tran-
tality decline in the nineteenth century. sition preceding mortality declines. In other
Historians and others still debate the causes places, such as Germany (figure 2), the fer-
of the historical mortality decline, but most tility and mortality declines took place at
scholars stress some combination of bet- roughly the same time. This does not rule out
ter food supplies, improvements in public a role for exogenous changes in mortality as
health systems (such as clean water sup- a causal force, of course, but it suggests that
plies and food-safety measures), and modest Notestein-style account explains only part
results from medical interventions (such as of the change. The total fertility rate in the
vaccines against smallpox).16 Some of these United States in the early nineteenth century
developments reflect local decisions about was about seven. Even if thirty percent of
public-good investments, but it is plausible children then died in infancy or childhood,
to view them as largely exogenous to any this implies that households wanted a surviv-
couple’s decision making. ing brood of four or five. By the end of the
This mortality decline was concentrated in nineteenth century, in contrast, white, urban
the early years of life. A woman born in the women in the United States were increas-
United States in 1850 had an expectation of ingly having just two children (Paul A. David
life of 39.4 years. A five-year-old girl in that and Warren C. Sanderson 1987).
year had an expectation of life of 50.8 years, “Demographic transition theory” also has
and a twenty year-old could expect to live two theoretical flaws. Part of the decline in
an additional 39.8 years. In 1910, these fig- infant and child mortality is endogenous to
ures were nearly the same for those who had the fertility decline. There are several lines
already survived the dangerous early years. A of argument here, all of which assume that
newborn girl could expect to live 54.7 years, parents can assert some influence on their
children’s mortality risks by providing health-
enhancing resources. In a historical con-
16 Thomas McKeown (1976) argued that prior to about text, these resources include breast-feeding
1900, medical science had done little to increase human (which costs mother’s time, but isolates an
longevity, and concluded that the observed mortality infant from possibly contaminated water and
declines to that point reflected direct and indirect effects
of better nutrition brought about by higher incomes and
better food supplies. Robert William Fogel (2004)’s more
nuanced account also stresses the role of nutrition. For 17 Historical Statistics of the United States series Ab657,
an introduction and overview to this issue see, Angus 659, and 661. Mortality figures for the United States are
Deaton’s review of Fogel’s book in the Journal of Economic complicated by the lack of complete death registration sta-
Literature, 44(1): 106–14. tistics, but the basic patterns noted in the text are robust.
600 Journal of Economic Literature, Vol. XLIX (September 2011)

food ­supplies), other nutrition, and protection Assessing this explanation is frustrated by lack
from danger such as hearth fires.18 In the Q–Q of direct evidence on contraceptive practice
model, reduced infant and child mortality for most of the period. The best we can do is
could also reflect changes in other costs. For to provide a broad characterization of those
example, improved contraceptive technology technologies and the constraints they put on
(discussed next) could allow parents to more couples. Indirect evidence shows that until
tightly control the link between actual and the second half of the nineteenth century,
desired fertility. Parents might have a smaller most fertility control relied on a combina-
number of children and care for them more tion of withdrawal (coitus interruptus) and
intensively, in effect not relying on high mor- abstinence from sexual relations.20 The first
tality to cull their brood to its desired size.19 “modern” methods appeared in the second
The second theoretical problem with the half of the nineteenth century. These tech-
“demographic transition” account is that niques relied partly on advances in medical
even a fully exogenous reduction in infant understanding, but also on the invention of
mortality would have two, countervailing vulcanized rubber (in 1844). Applied to the
effects. An exogenous mortality decline production of condoms (in 1855), this new
reduces pn and thus makes child numbers industrial process allowed couples to replace
cheaper relative to both child quality and Zs. expensive and relatively unreliable condoms
An exogenous mortality decline could actu- made from natural materials. (Most studies,
ally raise fertility net of child mortality. in fact, conclude that prior to vulcanization,
­condoms were intended to prevent the spread
4.2 Innovations in Contraceptive Methods
of disease, not ­pregnancy). Depending on the
A second explanation for the fertility transi- place, condoms were widely available in bar-
tion implies that couples long wanted smaller ber shops, drug stores and other retail outlets.
families, and improvements in contraceptive Vulcanized rubber was also the basis for the
methods made that goal easier to achieve. introduction of the diaphragm and similar bar-
rier methods in the later nineteenth century.
Thus we have a new set of technologies,
18 Historical research has not always taken this issue
which would make contraception easier and
seriously. Papers that do instrument for infant mortal- more reliable, introduced not long before we
ity in econometric analysis generally find that it dramati- observe the actual fertility decline in some
cally reduces the apparent impact of mortality in an OLS countries (but well after France and the
framework. See Patrick R. Galloway, Lee, and Eugene A.
Hammel (1998a), which also surveys earlier literature;
Brown and Guinnane (2002); Tommy E. Murphy (2010a).
19 If this claim seems extreme, consider the practice of 20 Angus McLaren (1978, pp. 25–27) among others
wet-nursing, which was extensive in France into the late notes that frequent condemnations of withdrawal in the
nineteenth century. In many cases, one woman would take eighteenth century suggest that the practice was already
several urban babies soon after their birth, and travel with used as a form of contraception. James Woycke (1988,
them by train to the location where they would be cared p. 11) concludes that throughout the nineteenth century,
for. Some babies would die en route, from cold or hunger, “. . . it was coitus interruptus that remained the most com-
and others would die from neglect at their destination. The mon contraceptive practice.” Gigi Santow (1995) provides
practice was widespread and not limited to urban middle the best recent account of this issue. The nineteenth cen-
and upper classes. Catherine Rollet (1982, table 1) esti- tury also witnessed the spread of “marriage manuals,”
mates that about ten percent of all French newborns, and a euphemism for guides to sexuality, sexual health, and
about thirty percent of those born in the Paris region, were contraception. These guides appeared as early as the eigh-
sent to a wet nurse in the late nineteenth century. Anne teenth century, although their circulation was at first lim-
Martin-Fugier (1978, pp. 26–27) quotes a thirty percent ited (McLaren 1978, pp. 26–30). The usefulness of such
mortality rate for wet-nursed infants in the Paris region guides in limiting fertility of course depends on their infor-
earlier in the century. This is roughly twice the infant mor- mation being more accurate than what couples already
tality rate for France as a whole at the time. knew from other sources.
 Guinnane: The Historical Fertility Transition 601

United States). Robert T. Michael and Robert and practical enforcement of such laws were
J. Willis (1976) first integrated the costs of uneven, and in some cases, celebrated court
averting unwanted births into the micro- cases probably advanced public knowledge of
economic model. Their model assumes that contraceptive methods. The policies could also
couples can affect births, which is a random be self-contradictory. Germany’s Lex Heinze
variable, using methods that imply various (1900), for example, made illegal any public
utility and money costs. The couple’s opti- display or advertisement of objects intended
mization takes into account both the costs of for “obscene” use. But retailers could still sell
contraception and the utility costs of having condoms and other devices. At the same time,
“too few” or “too many” children. (Thus the two other important German institutions, the
Q–Q trade-off enters their thinking indirectly; Army and the Sickness Funds, were doing
a couple with too many children may, because their best to encourage the use of condoms to
of the budget constraint, be forced to choose stop the spread of venereal disease. 21
a lower level of quality than preferred). Any The effects of these restrictions on the
contraceptive method implies both fixed costs fertility transition are not really known.
(which must be “paid” to use the method at Demographers today tend to argue that the
all) and a marginal cost (which depends on availability of contraceptives and contracep-
the number of births averted). A couple that tive information is the most important barrier
wanted to avert three of an expected eight to fertility decline in developing countries.22
births would be happier with a relatively high Economists are more skeptical, stressing
marginal cost approach than would a cou- the incentives to reduce family sizes. In any
ple that wanted to avert all but two of eight case, three sets of useful surveys confirm that
expected births. throughout the nineteenth and early twen-
The United States and many European tieth century, withdrawal and abstinence
countries at first made concerted efforts to remained the primary approaches used by
limit the spread of contraceptive knowledge married couples.23 Since these technologies
and technologies. In the United States, the had been available, essentially, throughout
“Comstock Laws,” a collection of state and human history, it is unlikely that the condom
federal statutes, made it illegal to disseminate and similar new methods played a strong role
both marriage manuals and contraceptive in the fertility transition. Legal restrictions
devices such as condoms. Similar measures probably mattered less than cost. Rubber
were enacted in England, Germany, and condoms were at first expensive. Brown
many other European countries. The focus (2009a, pp. 15–16) estimates that in the early

21 Martha J. Bailey (2010) is one of the few careful view. See also John Bongaarts (1991). The publications of
empirical studies of the Comstock Laws, but deals with a the Alan Guttmacher Institute, on the other hand, argue
period, the 1960s and later, that is well after the U.S. fertil- that there is considerable “unmet need for contraception”
ity transition. She concludes (p. 122) that in 1965, with- today. See, for example, “Facts on Satisfying the Need for
out the bans in place, marital fertility in the affected states Contraception in Developing Countries,” April 2010.
would have been eight percent lower. 23 Three sets of surveys, some of doctors and some of
22 This view underlies the literature on the so-called women or couples directly, all conclude that the primary
KAP-gap. In many surveys in developing countries, a sig- techniques used by married couples were withdrawal and
nificant number of women report that they do not want any abstinence. See Brown (2009a, table 1) for a summary of
more children but are not using contraception, or that they three German surveys from the early twentieth century;
had more children than they actually wanted. The pre- David and Sanderson (1986, pp. 317–28) discuss the
cise reasons for this are debated, but many demographers Mosher survey of American women born in the 1850s and
think this fact reflects lack of access to contraceptives. 1860s, as well as later U.S. surveys; Robert Jütte (2003, p.
Charles F. Westoff (1988) is an early, somewhat skeptical 220) discusses a survey of French doctors from the 1890s.
602 Journal of Economic Literature, Vol. XLIX (September 2011)

twentieth century, a year’s supply of condoms ­ roduce voluntary reductions of the magni-
p
cost a Berlin worker the equivalent of ten to tude we observe in the nineteenth and early
fifteen days of wages. Other barrier methods twentieth centuries.
such as the diaphragm required the attention
4.3 Increases in the Direct Costs of
of a trained medical professional.24 Cost may
Children
account in part for Santow’s (1993) finding
that coitus interruptus remained widely used In the Q–Q model, increases in pn induce
well into the twentieth century, even in coun- substitution toward both child quality and
tries where alternatives were available. David the Z goods directly, and toward quality
and Sanderson (1986) develop a model of a through the interaction term in the budget
couple’s lifetime fertility as a renewal process, constraint. One logical possibility to explain
and use it to derive estimates of the number the fertility transition is that the direct costs
of live births a couple would experience in a of child-bearing changed in ways that induced
twenty-year marriage under various assump- couples to have smaller families. The problem
tions about coital frequency and contracep- is that most costs did not change, over the
tive failure rates. Their baseline couple (no relevant period, in ways that would produce
contraception) would have about nine births the observed fertility decline. Most house-
if they had sexual intercourse, on average, holds in this period devoted the bulk of their
five times per 24-day cycle. If this couple expenditure to food, clothing, and housing.
used a method with a 12.5 percent failure The real price of clothing dropped dramati-
rate, and failed to use it about 10 percent of cally following the technological innovations
the time, they would have only three births of the Industrial Revolution, many of which
in twenty years. This “method” approximates were in textiles. Food prices varied over time
what we know about the use of coitus inter- and place, and protective tariffs on agricul-
ruptus in modern populations. Conscientious tural goods could raise the price of food in
use of condoms would get the couple below one country above its counterpart in others.
one birth.25 The methods ­available even prior But in general, food prices declined, which
to the fertility transition were sufficient to at a crude level would imply a reduction
in pn.

24 The contraceptive pill used today dates from the

1950s, with its first widespread use taking place in the assume that only stopping is legitimate evidence of fertility
1960s. Historical sources also refer to efforts to induce control; evidence of smaller families achieved via spacing
abortion. Given social and later legal views of this practice, is attributed to motivations unrelated to family size. The
we cannot hope to know how common abortion really was reasons for this assumption lie beyond the scope of this
during the period of the transition. paper, but amount to a concern about identification (see
25 Their model is based on Mindel C. Sheps and Jane Louis Henry 1961). Many historical demographers also
A. Menken (1973) and assumes that couples follow the view rudimentary fertility control techniques as unsuited
same strategy over their entire lifetime. Michael and to a spacing strategy; dynamic models of family-building,
Willis (1976, table 2) report a similar exercise using dif- on the other hand, imply that risk-averse couples using
ferent parameter values. The unwary economist may fall unreliable contraceptive methods will prefer spacing
into a trap created by the way most demographers think to stopping (David, Thomas A. Mroz, and Kenneth W.
of fertility control. To most demographers, the term “fam- Wachter 1987). Several empirical studies show that spacing
ily limitation” does not mean a reduction in family size, it was widespread in the early stages of the fertility transition
means a reduction in family size achieved in a particular (see Lee L. Bean, Geraldine P. Mineau, and Douglas L.
way. A couple seeking to have only N children can adopt a Anderton 1990 and Mroz and Weir 1990). The develop-
“stopping” strategy (have N children before initiating any ment of more reliable fertility control techniques in the
effort to restrict fertility) or a “spacing” strategy (reduce nineteenth century could well have caused stopping to
the probability of a birth right from the start of marriage, replace spacing. Works relying on the demographer’s defi-
and thus spread out births throughout the fertility years nition would identify a shift to stopping, holding number
of the marriage). Most studies in historical demography of children constant, as the onset of the fertility transition.
 Guinnane: The Historical Fertility Transition 603

The only significant increases in direct costs e­ xplanation for the fertility decline in rural
took place because of urbanization. Most America rests on the rising costs of farm-
European countries as well as the United land as an area was settled. Suppose a farm
States experienced rapid urbanization during couple wanted to establish each child on a
the nineteenth century. About six percent of farm similar to their own. As the price of
the U.S. population lived in an urban place local farmland rose, parents either had to
in 1800; in 1900 that was nearly forty per- send their children further west, where land
cent (Haines 2000a, table 4.2). England was was cheaper, or had to have fewer children.
already very urban in 1801 (34 percent), and Easterlin argues that parents preferred to
became even more so over the nineteenth have fewer children and be able to settle
century. By 1911, 79 percent of the English them locally. He dates the beginning of the
population lived in an urban center (Robert decline in New York State to 1805 and even
Woods 1996, table 3). France started out the Iowa, much further west, to 1835.27 Later
period less urban, and while its cities did research focuses on Easterlin’s assump-
grow, it remained less urban than England tion that parents wanted to give each child
or Germany. Urbanization in Germany was a fixed bequest. William A. Sundstrom and
especially rapid in the late nineteenth cen- David (1988), for example, motivate their
tury. Germans living in places with fewer regression analysis using a bargaining model
than 2,000 people fell from 64 to 40 percent that presumes that a primary motivation for
of the population between 1871 and 1910 child-rearing is support in old age. In equi-
(Hans-Ulrich Wehler 1995, table 71). In librium, children can drive a harder bargain
urban areas, housing costs exceeded costs in with their parents if they can point to bet-
rural areas, but of course the decision to live ter, off-farm opportunities. Cross-sectional
in a city was up to a couple. Most studies find regressions for U.S. states in 1840 show that
that urban fertility was lower than rural fer- fertility is negatively correlated with mea-
tility in the nineteenth century, although the sures of nonfarm labor-market opportuni-
precise causation has not been established. ties. Once such proxies are introduced, land
Knodel (1974, table 3.2) reports that marital prices have no influence on fertility.28
fertility in Berlin in 1867–68 was about 87 Child labor raises another source of
percent of that in rural Prussia; by 1905–06, variation in direct costs. In many societ-
Berlin’s fertility was half of rural Prussia’s and ies children offset some of the direct costs
Prussian cities overall had fertility about 75 to their parents by working either in paren-
percent of rural Prussia’s. Haines (1989, table tal income-generation activities (such as a
2) estimates TFR in the urban United States farm) or by working in the labor market. Any
in 1905–10 at about 2.7, with rural nonfarm change in children’s earnings would clearly
TFR at 4.0 and rural farm TFR at 6.0. Once
the fertility transition began, f­ertility usually
fell first in urban areas, with rural areas then 27 His fertility measure is the child–woman ratio, or the
catching up.26 number of children age 0–9 per thousand women 16–44.
A second type of direct costs underlies This measure is sensitive to in- and outmigration of both
a literature that started with the U.S. fer- children and adults, as well as to variations in infant and
child mortality.
tility transition. Easterlin’s (1976) famous 28 Susan B. Carter, Roger L. Ransom, and Richard
Sutch (2004) provide additional evidence on this debate.
This type of argument illustrates the problems with defin-
ing child quality. One could argue that Easterlin’s explana-
26 Alan Sharlin (1986) surveys urban–rural differences tion is one where parents reduce child numbers when the
in European fertility, using the Princeton project’s data. costs of quality increased.
604 Journal of Economic Literature, Vol. XLIX (September 2011)

alter the net costs to their parents. Two of 15 who had not attended school at least
important trends affected children’s earnings three months in the previous year (Carolyn
opportunities during the period in question. Moehling 1999, p. 74).
Most accounts imply that industrialization at There are two styles of explanation for the
first increased income-earning opportunities new child-labor laws. One is that a combi-
for children, because new technologies did nation of social-welfare concerns, along with
not require physical strength. By the 1830s, representatives of labor concerned about
large minorities of English children were competition with adult males, overwhelmed
working. Clark Nardinelli (1990, table 4.2) industrialists’ opposition to child-labor
reports that in most English counties, at least restrictions. The other explanation is that
one-quarter of children aged 10–14 were these measures were enacted when indus-
reported in the workforce. Some parts of the try no longer opposed them; either it had
textile sector depended heavily on children. become easy to substitute capital and other
One Parliamentary inquiry reported that sorts of labor for child labor, or the work-
in cotton textiles, half of all workers were force had already changed in ways that the
under 18, and 6.8 percent were under 10 new laws were not a binding constraint when
(Nardinelli 1990, table 5.3). Wehler (1996, passed.29
p. 254) notes that in German factories in this Child-labor restrictions potentially
period, children could be fifteen percent of reduce the incentive to have a large family,
the workforce. Estimates for industrializing but we have to bear in mind their limita-
New England run even higher. tions. Most such measures either did not
By the mid-nineteenth century, the use apply to agricultural work, or did so in a
of children, especially in industry, became more relaxed way. Wehler (1996) empha-
controversial. Governments imposed age sizes that the German restrictions did not
restrictions and other measures that reduced successfully limit the role of children in pro-
children’s earning opportunities. The British duction at home, which remained important
“Factory Acts,” starting in 1833, imposed throughout the nineteenth century. And in
restrictions on the ages of children who every case, the restrictions’ impact would
could work, and how many hours they could depend both on enforcement measures
work. But they started at a modest level; and parents’ desire to evade them. Finally,
the 1833 Act restricted children aged 9–12 if child-labor restrictions were introduced
to forty-eight hour weeks. Prussia intro- when they were mostly irrelevant, then they
duced similar measures in 1839, with other could not be a strong causal force in the fer-
German states soon following (Wehler 1996, tility transition.
p. 257). Many governments tied restrictions
4.4 Increases in the Opportunity Costs of
on child labor to an education requirement.
Childbearing
The English Factory Acts required that child
workers also be in school. In some cases, Industrialization usually changed the role
the factory had to set up its own school to of women in the workforce, although eco-
continue employing children (Nardinelli
­ nomic historians do not agree on just how.
1990, pp. 106–07). The Prussian 1839 Act Several studies show that women played
established a minimum work age of nine important roles in factory work early in the
years, and sixteen years for children who had
not yet had at least three years of schooling.
In Massachusetts as of 1837, manufactur- 29 Moehling (1999) argues the latter for the United
ers could not employ anyone under the age States and gives references to the debate.
 Guinnane: The Historical Fertility Transition 605

industrialization process, but became a less care for children, while a woman working in
important part of that labor force as time a textile factory could not. Some industries
went on. In a few industries such as textiles, refused to hire married women at all, thus giv-
women were certainly quite numerous into ing women an incentive to delay marriage.32
the twentieth century. Sara Horrell and Jane Industrialization thus raised the opportunity
Humphries (1995, table I) estimate wives’ cost of children in two ways.
labor-force participation rates of about 65 Several studies find that local employment
percent for England in the period 1787– opportunities for women lowered fertility. N.
1815, which corresponds to the height of the F. R. Crafts (1989) relies on the fertility and
Industrial Revolution. This figure appears to occupational information in the 1911 census
fall in the late nineteenth century. Most other of England and Wales. He finds a consistent,
accounts report women as a proportion of the negative correlation between women’s local
workforce in particular industries.30 In Britain labor-force opportunities and marital fertility,
in 1851, women constituted about thirty per- with elasticities ranging from –   0.13 to –   0.34.
cent of the country’s measured labor force, Studies such as Brown and Guinnane (2002),
and about forty percent of all employed which uses both textile mills and the struc-
women worked in textiles and clothing ture of local agriculture to proxy for women’s
(Duncan Bythell 1993, p. 35). Claudia earnings opportunities, find small, statistically
Goldin and Kenneth Sokoloff (1982, table 3) significant effects with the expected signs.
estimate that women comprised 20–30 per- Schultz (1985) uses a different approach that
cent of the workforce in New England tex- links women’s earnings opportunities to the
tile factories in 1820, and about twice that fertility transition per se. Using time series-
in 1832. Gary Saxonhouse and Gavin Wright cross section data on Swedish counties for
(1984, table 1) report that 57.6 percent of the the period 1860–1910, he shows that the ratio
workforce in cotton textiles in the American of women’s to men’s wages explains about a
South in 1880 were female. Wehler (1996, p. quarter of the decline in Swedish fertility.
254) reports that women could be half the He treats women’s earnings as endogenous,
workforce in some German factories in the and instruments for them using demand-side
1830s and 1840s; in 1875, according to the shocks to ­agricultural prices. Women’s earn-
census, women constituted nearly half of the ings depress fertility at virtually all ages, so
workforce in textile and clothing factories in this effect seems to work through more than
Germany, and about twenty percent of indus- delayed marriage.33
trial workers overall.31
Married women certainly worked prior to
the Industrial Revolution, but industrial work 32 The “Lowell system” used by some textile mills in
created new opportunities and trade-offs for New England before the Civil War recruited young women
women. It offered better-paying work that to work in the mill and live in a special company board-
could not be combined with c­ hild-minding; inghouse under the supervision of a “housemother.” The
point was to recruit farmers’ daughters who would other-
a woman spinning yarn at home could also wise be unwilling to undertake factory work (Saxonhouse
and Wright 1984, pp. 4–5).
33 Marianne H. Wanamaker (2010) uses the introduc-
tion of textile mills in South Carolina in the period 1880–
30 Historical censuses do a poor job of reporting wom- 1900 to study the impact of changes in opportunity costs on
en’s occupations, especially married women’s occupations. family sizes. The introduction of a textile mill reduced fer-
Horrell and Humphries (1995) base their study on family tility in the surrounding area by about 11 percent c. 1900.
budgets. The effect reflects differential migration of low-fertility
31 Jürgen Kocka (1990, table 16). This figure refers only couples and so does not reflect a shift in the opportunity
to establishments with five or more employees. cost of childbearing in the sense of Becker’s model.
606 Journal of Economic Literature, Vol. XLIX (September 2011)

4.5 Changes in the Costs of and Returns to about 60 percent of ­ students required to
Child Quality attend school actually did so, a figure rising
to 82 percent by 1846. Some German states,
Another appealing idea is that fertility such as Saxony, did better (Nipperdey 1994,
decline reflects increases in the net return to p. 463). Several U.S. states introduced free
child quality. There are two different ques- public elementary education starting in the
tions to ask. First, did the cost of child quality, 1840s and, for most of the nineteenth cen-
in the form of education, decline? Second, tury, the United States was an outlier in
did the return to child quality increase? A the proportion of young people in school.
positive answer to either question implies Free, compulsory education came later
a substitution away from numbers towards to Germany (1872), France (1882), and
child quality. Unfortunately we can say more England and Wales (1893) (Kristine Bruland
about the former than the latter. 2003, pp. 160–61). Easterlin (1981, appendix
Economic historians of education stress table 1) estimates that in 1850, there were
important distinctions in the types of eco- 1,800 children in primary schools per 10,000
nomically useful education. One could total population in the United States, com-
acquire basic literacy and numeracy at home pared to 1,600 in Germany, 930 in France,
(if the parents were literate) or in primary and 1,045 in the United Kingdom.
school. More advanced education or training Literacy is both a broadly comparable
required secondary schools, formal appren- measure of educational outcomes and the
ticeship, or on-the-job training. Tertiary most economically useful output of primary
education during the relevant period was schools. David F. Mitch (2004, tables 12.5A
restricted to a small elite, and while perhaps and 12.5B) reports that, in the time around
important for overall TFP growth, would not 1860, adult male illiteracy rates were about
figure heavily in demographic decisions. 35 percent in France, 30 percent in England,
The growth of literacy and its primary 30 percent in the United States (whites
cause, public elementary education, differs only), and 5 percent in Prussia. Female illit-
dramatically across the countries on which eracy rates in these countries were higher:
we focus. There are two broad classes of 45 percent for France, 37 percent for
important decisions: First, to make primary England, 10 percent for the United States
schooling universally available, and second, to (whites only), and 5 percent for Prussia. The
make it compulsory. Prussia led the way with rapid development of schools in both Britain
the 1763 requirement that all children aged and France dramatically increased literacy
five to thirteen attend primary schools. The rates by the end of the nineteenth century
schools were not free, but there was tuition (François Furet and Jacques Ozouf 1977, p.
assistance for the poor. Like many grand edu- 293).
cational reforms, this measure’s implementa- The creation of schools certainly reduced
tion was resisted by various interests, and in the cost of elementary education. But the
any case Prussia lacked sufficient teachers for opportunity cost of the time spent in school
all the children in the territory (James van remained, even when schools were free.
Horn Melton 1988, pp. 174–77).34 In 1816, Mitch (1992, p. 156) quotes Horace Mann’s

34 Economic historians of schooling stress that the qual- toward subjects such as religion. Primary school students
ity of schools varied dramatically across time and place. The learned to read and write, and some basic arithmetic, but
celebrated Prussian schools, for example, stressed the for- more advanced skills were acquired in other ways (Thomas
mation of Prussian citizens; thus instruction was weighted Nipperdey 1994, p. 462).
 Guinnane: The Historical Fertility Transition 607

comment in the British 1851 census of edu- 4.6 Social Insurance and Old-Age Support
cation: “It is not for sake of saving a penny
per week, but for the sake of gaining a shil- One particular return to child-rearing
ling or eighteenpence a week that a child is that receives considerable attention in the
transferred from the school to the factory.” literature is children’s role in insuring par-
Child-labor laws might have been more ents against the consequences of accidents,
important for encouraging schooling than ill-health, and old age. The most common
the schools themselves. argument is that children are a form of life-
What were the returns to education? cycle savings; parents invest when they are
Historical sources usually report only literacy young and healthy, and then expect their
status, not years of education, and occupation children to care for them in infirmity or
rather than income or wage. So we cannot old age. (The Sundstrom–David criticism
estimate returns to education as is in the mod- of Easterlin’s model, noted above, is one
ern literature on the economics of education. version of this argument.) We should also
Goldin and Lawrence F. Katz (2000) exploit consider insurance against accidents and
the unique Iowa census of 1915 to provide ill-health. Two versions of this argument
one of the only historical estimates available. have been advanced to explain the fertility
They find that the return to an additional year transition. One is that industrialization and
of high school or college then was, for males, the increased mobility it entails, especially
on the order of 11–12 percent (Goldin and rural-to-urban migration, made it harder for
Katz 2000). Mitch (1992, pp. 230–35) esti- parents to hold children to the intergenera-
mates the present value of acquiring literacy tional bargain. This “child default” argument
in Victorian Britain for a representative child. implies that the developing industrial econ-
The present value of the cost of acquiring lit- omy made children a less desirable vehicle
eracy would be about £4. At a wage premium for savings.35 A second version of the argu-
of 5 shillings per week for literacy, the pres- ment points to the development of substitute
ent value of the higher wages for a 35-year means of providing for old age, especially
work life would be over £200. social insurance and the welfare state.
Efforts to examine the relationship Both versions of the argument suffer from
between children’s education and their par- the problem that we know that economic ties
ents’ fertility also confront data problems. between parents and children varied dramat-
Sources usually do not include both ­parental ically across the societies in question before
fertility and children’s educational attain- the fertility transition. In some European
ment, which is what we need for direct tests regions, peasant households would draft for-
of the Q–Q trade-off. It would be tempting mal documents that turned a farmstead over
to interpret parental education as a proxy to the heir, and carefully specified the heir’s
for children’s education. But this kind of obligations to his parents (as well as to sib-
effect can be interpreted in many different lings to who had not yet received an inheri-
ways. The paper on Prussia discussed earlier tance). This Altenteil guaranteed specific
(Becker, Cinnirella, and Woessmann 2009) transfers to the retired couple, who often
takes the preferred approach, which is to lived on the farm in a special house reserved
estimate the impact of school enrollment
rates on fertility. Presumably this approach
35 This would also be consistent with Caldwell’s changes
could be replicated in other contexts, using
in “net intergenerational wealth flows.” Carter, Ransom,
changes in schooling and child-labor rules to and Sutch (2004) invoke the child default argument in dis-
measure the net returns to education. cussing the U.S. fertility transition.
608 Journal of Economic Literature, Vol. XLIX (September 2011)

for that purpose.36 At the other extreme, received poor relief in 1885. Social insur-
rural laborers’ children in England would, ance, on the other hand, generally operated
from at least the early-modern period, leave on insurance principles: covered individuals
home for good in their early to mid teens. received specified payments triggered by
The best evidence suggests these children specified events or conditions.37
had no further economic relationship with The literature divides social insurance
their parents (Alan MacFarlane 1986, pp. into four categories: sickness insurance,
83–84). Thus the extent to which parent– (workplace) accident insurance, old-age and
child ties changed during the nineteenth disability insurance, and unemployment
century varies a great deal. We should also insurance. The first broad social-insurance
note that the “child default” version of the system dates to 1883, when the German
argument resembles arguments about mor- government introduced compulsory sickness
tality (although in reverse): from the par- insurance (1883) and then accident insur-
ents’ viewpoint, rising child default is like ance (1884). Disability and old-age insurance
increased infant and child mortality. Parents’ were added in 1889, while unemployment
might actually invest in more, lower-quality insurance came later. The system’s intro-
children to ensure that at least some children duction was less discrete than these dates
remained faithful. imply. Some German workers were already
This argument faces a different kind of covered by schemes that were compul-
challenge, which is that the social-insurance sory for their industry, and that became the
systems introduced at the end of the nine- model for the system for which Bismarck is
teenth and early twentieth century were usu- always credited. The sickness and accident-
ally replacing earlier schemes. Thus there is insurance programs covered only workers in
no clear “before.” Prior to the introduction selected industries at first, although coverage
of social insurance, every society in question broadened rapidly.38 Participation in the dis-
here had some form of provision for the poor ability and old-age insurance program were
and those unable to provide for themselves limited by income. The United Kingdom’s
because of illness or age. These “poor relief” Old Age Pensions Act (1908) illustrates a
systems were locally financed and organized, different approach. This noncontributory
and were rarely as generous as the social- scheme was introduced for all persons who
insurance schemes that eventually replaced met a (mild) means test and had reached the
them. Another important difference was age of 70 years. France and other Continental
their logic: poor relief systems were intended
to relieve “need,” and so imposed asset tests.
37 Many workers in the late nineteenth century were
The numbers receiving relief at any one
also covered by voluntary, private schemes that provided
time were small, as one would expect. E. P. assistance in case of sickness, accident, or infirmity. Some
Hennock (2007, pp. 46–47) estimates that individual employers and labor groups also created insur-
6.6 percent of the English population and ance programs that covered parts of the workforce. In
Germany, some workers were obliged to join a particular
about 3.4 percent of the German population insurance organization long before the introduction of for-
mal social insurance. Guinnane and Jochen Streb (2011)
discusses Friendly Societies and the German organizations.
John E. Murray (2007) discusses these organizations more
36 The practice goes by many names. There are clear broadly and provides references to the broader literature.
references to it in Scandinavia and Ireland, but most would 38 David Khoudour-Castéras (2008) reports that the
associate it with German-speaking Central Europe. See health insurance law covered 21 percent of workers in
Guinnane (1997, pp. 149–151) for Ireland, Jon Gjerde 1885, rising to 44 percent in 1913. The accident insurance
(1985) for Norway, and David Gaunt (1983) for a survey of system at first covered 18 percent of workers, but by 1913
Northern and Central Europe. covered 94 percent (appendix tables 1 and 2).
 Guinnane: The Historical Fertility Transition 609

countries tended to emulate the German produced by demographers and others who
approach to social insurance, introducing sometimes use definitions that to economists
first sickness and accident insurance, usually are unfamiliar and perhaps surprising. These
in the 1890s or later. There were earlier old- differences reward care in consuming that
age pension systems, but they were volun- literature.
tary and did not receive state subsidies, and There are several different economic
were not widely used. The United States was explanations for the historical fertility tran-
an outlier: before the Social Security Act of sition. Two doubtless played a role, but are
1935, the United States had no social insur- easy to exaggerate. During the late nine-
ance per se. Local poor relief systems, war teenth century technological change intro-
pensions, and mother’s pensions filled some duced the first widespread use of “modern”
of the role met by poor relief and social insur- contraceptives such as condoms. Yet the
ance in Europe. The “workman’s compensa- imperfect evidence available suggests these
tion” system introduced in the United States methods were still expensive, and that most
in the early twentieth century filled the role couples continued to rely on “traditional”
of German-style accident insurance (Price V. methods such as abstinence and withdrawal.
Fishback and Shawn Everett Kantor 2000). Furthermore, simulation models show that
The timing of social insurance’s rise in traditional methods were sufficiently effec-
Europe hints that it is part of the story of tive to account for the fertility decline we
the fertility transition. We cannot identify observe. Another often-stressed explanation
a single date at which social insurance first for fertility decline is the decline of mor-
appeared, however. On the other hand, with tality, especially infant and child mortality.
care one might track the extension of such Mortality decline probably played a role in
programs to different parts of a national the historical fertility decline, but I have
population, and look for the impact on fertil- stressed two caveats. In some countries, fer-
ity of changes in social insurance over time. tility declined significantly before any real
The broad patterns also do not make it likely mortality decline. In addition, infant and
that social insurance alone is central to the child mortality are at least partially endog-
story. The two forerunners, France and the enous to fertility.
United States, were laggards in developing The other explanations discussed here all
social insurance work off relative price changes, as implied
by Becker’s demand-for-children model.
Several significant changes in the relevant
5.  Conclusions
period plausibly altered the costs of and
The historical fertility transition played returns to children in ways that would reduce
a central role in the making of modern fertility. These include housing costs due to
economies. This paper outlines the central urbanization, changes in child-labor law,
empirical patterns in a selection of impor- increases in the opportunity costs of child-
tant, wealthy countries, and then provides bearing because of better labor-force oppor-
an overview of the major economic explana- tunities for women, the introduction of free
tions for the fertility transition. I caution at or compulsory primary education, and the
several points that apparently minor differ- development of social-insurance systems.
ences in demographic measures can make an Despite at least one hundred years of
important difference to the patterns econo- academic and official interest in the decline
mists seek to explain. More generally, much of fertility, this question is not one for
of the relevant literature in this area was which economists have a clear, empirically
610 Journal of Economic Literature, Vol. XLIX (September 2011)

­ ell-founded explanation. (This is not to say

w Even when the sources are cross-sectional,
there are not many partial answers based on individual-level data supports tests of
particular times or places, or many theoreti- multivariate hypotheses. (With aggregate
cal efforts that capture some important part data, on the other hand, there are usually
of the process.) There are several reasons for ecological inference problems with any
the current state of the literature. The most multivariate hypothesis.) These empirical
important is extensive simultaneity: the rel- approaches require significant investment
evant period saw many significant economic of researcher time and other resources,
changes, and some of these changes were but hold out the promise of really under-
tied to other changes such as the introduc- standing, for example, the effect of women’s
tion of social insurance. Economists are labor-force opportunities on the demand
well aware (and more recently, working for children during the fertility transition.
hard on) the interrelations between fertil- The recent upsurge in theoretical inter-
ity and the economy. But there remains the est in the historical fertility transition will
problem of understanding the role of any pay an even greater dividend if it motivates
given change in a period when many other renewed interest in using the available data
important forces changed at the same time. to understand the economics of the histori-
This is of course partly a standard identifi- cal fertility transition.
cation problem, and partly an issue of being
able to invoke ceteris paribus reasoning.
References
Both require much more careful empirical
work than we presently have. Bailey, Martha J. 2010. “‘Momma’s Got the Pill’: How
Much more empirical work is both nec- Anthony Comstock and Griswold v. Connecticut
Shaped US Childbearing.” American Economic
essary and possible. The recent literature Review, 100(1): 98–129.
typically relies on national aggregates or, Barro, Robert J., and Gary S. Becker. 1989. “Fertility
in some cases, subnational aggregate data. Choice in a Model of Economic Growth.” Economet-
rica, 57(2): 481–501.
This approach sometimes reflects a reason- Bean, Lee L., Geraldine P. Mineau, and Douglas L.
able trade-off between the costs and benefits Anderton. 1990. Fertility Change on the American
of data collection. But for many historical Frontier: Adaptation and Innovation. Berkeley and
Los Angeles: University of California Press.
economies there remain significant possi-
­ Becker, Gary S. 1960. “An Economic Analysis of Fer-
bilities for using data much better-suited to tility.” In Demographic and Economic Changes in
testing the hypotheses described. The most Developed Countries, 209–30. Princeton: Princeton
University Press.
important unexploited sources are individ- Becker, Gary S. 1976. The Economic Approach to
ual-level data, especially individual-level Human Behavior. Chicago and London: University
data that include wealth and income, or rea- of Chicago Press.
Becker, Gary S. 1981. A Treatise on the Family. Cam-
sonable proxies for wealth and income (such bridge and London: Cambridge University Press.
as occupation). The latter point is crucial: we Becker, Gary S., and Robert J. Barro. 1988. “A Refor-
know relatively little about the historical fer- mulation of the Economic Theory of Fertility.” Quar-
terly Journal of Economics, 103(1): 1–25.
tility transition because much of the earlier Becker, Gary S., and H. Gregg Lewis. 1973. “On the
research relied on sources that do not contain Interaction between the Quantity and Quality of
useful economic information. Individual- Children.” Journal of Political Economy, 81(2):
S279–88.
level information is important for several dif- Becker, Gary S., Kevin M. Murphy, and Robert
ferent reasons. Sometimes data of this sort Tamura. 1990. “Human Capital, Fertility, and Eco-
allows the researcher to follow life-cycle fer- nomic Growth.” Journal of Political Economy, 98(5):
S12–37.
tility, which permits use of a broader range Becker, Sascha O., Francesco Cinnirella, and Ludger
of shocks to ­identify ­behavioral responses. Woessmann. 2009. “The Trade-Off between F ­ ertility
 Guinnane: The Historical Fertility Transition 611

and Education: Evidence from before the Demo- David, Paul A., and Warren C. Sanderson. 1986. “Rudi-
graphic Transition.” CESifo Working Paper 2775. mentary Contraceptive Methods and the American
Bleakley, Hoyt, and Fabian Lange. 2009. “Chronic Transition to Marital Fertility Control, 1855–1915.”
Disease Burden and the Interaction of Education, In Long-Term Factors in American Economic
Fertility, and Growth.” Review of Economics and Growth, ed. Stanley L. Engerman and Robert E.
Statistics, 91(1): 52–65. Gallman, 307–79. Chicago and London: University
Bongaarts, John. 1991. “The KAP-Gap and the Unmet of Chicago Press.
Need for Contraception.” Population and Develop- David, Paul A., and Warren C. Sanderson. 1987. “The
ment Review, 17(2): 293–313. Emergence of a Two-Child Norm among American
Brown, John C. 1988. “Coping with Crisis? The Dif- Birth-Controllers.” Population and Development
fusion of Waterworks in Late Nineteenth-Century Review, 13(1): 1–41.
German Towns.” Journal of Economic History, 48(2): David, Paul A., Thomas A. Mroz, and Kenneth W.
307–18. Wachter. 1987. “Rational Strategies of Birth-Spacing
Brown, John C. 2009a. “Fertility Control with Imper- and Fertility Regulation in Rural France during the
fect Methods: Strategies of Family Building and the Ancien Regime.” Stanford Project on the History of
Choice of Technique during the German Fertility Fertility Control Working Paper 14-R.
Transition, 1885–1915.” Unpublished. Deaton, Angus. 2006. “The Great Escape: A Review
Brown, John C. 2009b. “Household Choice of Tech- of Fogel’s The Escape from Hunger and Premature
nique in the Fog of Uncertainty: Strategies of Family Death, 1700–2100.” Journal of Economic Literature,
Building and Patterns of Choice during the German 44(1): 106–14.
Fertility Transition, 1885–1915.” Unpublished. Easterlin, Richard A. 1975. “An Economic Framework
Brown, John C., and Timothy W. Guinnane. 2002. for Fertility Analysis.” Studies in Family Planning,
“Fertility Transition in a Rural, Catholic Popula- 6(3): 54–63.
tion: Bavaria, 1880–1910.” Population Studies, 56(1): Easterlin, Richard A. 1976. “Population Change and
35–49. Farm Settlement in the Northern United States.”
Brown, John C., and Timothy W. Guinnane. 2007. Journal of Economic History, 36(1): 45–75.
“Regions and Time in the European Fertility Tran- Easterlin, Richard A. 1978. “The Economics and Soci-
sition: Problems in the Princeton Project’s Statisti- ology of Fertility: A Synthesis.” In Historical Stud-
cal Methodology.” Economic History Review, 60(3): ies of Changing Fertility, ed. Charles Tilly, 57–133.
574–95. Princeton: Princeton University Press.
Bruland, Kristine. 2003. “Education.” In The Oxford Easterlin, Richard A. 1981. “Why Isn’t the Whole
Encyclopedia of Economic History, Volume 2, ed. World Developed?” Journal of Economic History,
Joel Mokyr, 160–70. Oxford and New York: Oxford 41(1): 1–19.
University Press. Ehmer, Josef. 1991. Heiratsverhalten Sozialstruktur,
Bythell, Duncan. 1993. “Women in the Workforce.” In ökonomischer Wandel. Göttingen: Vandenhoeck &
The Industrial Revolution and British Society, ed. Ruprecht.
Patrick O’Brien and Roland Quinault, 31–53. Cam- Festy, Patrick. 1979. La fécondité des pays occiden-
bridge and New York: Cambridge University Press. taux de 1870 à 1970. Paris: Presses Universitaires de
Carter, Susan B., Roger L. Ransom, and Richard Sutch. France.
2004. “Family Matters: The Life-Cycle Transition Fishback, Price V., and Shawn Everett Kantor. 2000. A
and the Antebellum American Fertility Decline.” Prelude to the Welfare State: The Origins of Workers’
In History Matters: Essays on Economic Growth, Compensation. Chicago and London: University of
Technology, and Demographic Change, ed. Timothy Chicago Press.
W. Guinnane, William A. Sundstrom, and Warren Fogel, Robert William. 2004. The Escape from Hunger
C. Whatley, 271–327. Stanford: Stanford University and Premature Death, 1700–2100: Europe, America,
Press. and the Third World. Cambridge; New York and
Clark, Gregory. 2007. A Farewell to Alms: A Brief Eco- Melbourne: Cambridge University Press.
nomic History of the World. Princeton and Oxford: Furet, François, and Jacques Ozouf. 1977. Lire et
Princeton University Press. écrire: L’alphabétisation des français de Calvin à
Cleland, John, and Christopher Wilson. 1987. “Demand Jules Ferry. Paris: Editions de Minuit.
Theories of the Fertility Transition: An Iconoclastic Galloway, Patrick R., Eugene A. Hammel, and Ronald
View.” Population Studies, 41(1): 5–30. D. Lee. 1994. “Fertility Decline in Prussia, 1875–
Coale, Ansley J., and Susan Cotts Watkins, eds. 1986. 1910: A Pooled Cross-Section Time Series Analysis.”
The Decline of Fertility in Europe: The Revised Pro- Population Studies, 48(1): 135–58.
ceedings of a Conference on the Princeton European Galloway, Patrick R., Ronald D. Lee, and Eugene A.
Fertility Project. Princeton: Princeton University Hammel. 1998a. “Infant Mortality and the Fertility
Press. Transition: Macro Evidence from Europe and New
Crafts, N. F. R. 1989. “Duration of Marriage, Fertility Findings from Prussia.” In From Death to Birth:
and Women’s Employment Opportunities in Eng- Mortality Decline and Reproductive Health, ed.
land and Wales in 1911.” Population Studies, 43(2): Mark R. Montgomery and Barney Cohen, 182–226.
325–35. Washington, D.C.: National Academies Press.
612 Journal of Economic Literature, Vol. XLIX (September 2011)

Galloway, Patrick R., Ronald D. Lee, and Eugene A. of Marital Fertility in the United States.” Demogra-
Hammel. 1998b. “Urban versus Rural: Fertility phy, 40(4): 605–20.
Decline in the Cities and Rural Districts of Prus- Haines, Michael R. 1989. “American Fertility in Tran-
sia, 1875 to 1910.” European Journal of Population, sition: New Estimates of Birth Rates in the United
14(3): 209–64. States, 1900–1910.” Demography, 26(1): 137–48.
Galor, Oded. 2005a. “The Demographic Transition and Haines, Michael R. 2000a. “The Population of the
the Emergence of Sustained Economic Growth.” United States, 1790–1920.” In The Cambridge Eco-
Journal of the European Economic Association, nomic History of the United States, Volume 3: The
3(2–3): 494–504. Twentieth Century, ed. Stanley L. Engerman and
Galor, Oded. 2005b. “From Stagnation to Growth: Robert E. Gallman, 143–205. Cambridge; New York
Unified Growth Theory.” In Handbook of Economic and Melbourne: Cambridge University Press.
Growth, Volume 1A, ed. Philippe Aghion and Steven Haines, Michael R. 2000b. “The White Population of
N. Durlauf, 171–294. Amsterdam and San Diego: the United States, 1790–1920.” In A Population His-
Elsevier, North-Holland. tory of North America, ed. Michael R. Haines and
Galor, Oded, and David N. Weil. 2000. “Population, Richard H. Steckel, 305–69. Cambridge; New York
Technology, and Growth: From Malthusian Stagna- and Melbourne: Cambridge University Press.
tion to the Demographic Transition and Beyond.” Hajnal, John. 1953. “The Marriage Boom.” Population
American Economic Review, 90(4): 806–28. Index, 19(2): 80–101.
Gaunt, David. 1983. “The Property and Kin Relation- Hajnal, John. 1965. “European Marriage Patterns in
ships of Retired Farmers in Northern and Central Perspective.” In Population in History: Essays in
Europe.” In Family Forms in Historic Europe, ed. Historical Demography, ed. D. V. Glass and D. E. C.
Richard Wall, Jean Robin, and Peter Laslett, 249–80. Eversley, 101–43. London: Arnold.
Cambridge and New York: Cambridge University Hajnal, John. 1982. “Two Kinds of Preindustrial House-
Press. hold Formation System.” Population and Develop-
Gjerde, Jon. 1985. From Peasants to Farmers: The ment Review, 8(3): 449–94.
Migration from Balestrand, Norway, to the Upper Hennock, E. P. 2007. The Origin of the Welfare State
Middle West. Cambridge and New York: Cambridge in England and Germany, 1850–1914: Social Policies
University Press. Compared. Cambridge and New York: Cambridge
Goldin, Claudia, and Lawrence F. Katz. 2000. “Edu- University Press.
cation and Income in the Early Twentieth Century: Henry, Louis. 1961. “Some Data on Natural Fertility.”
Evidence from the Prairies.” Journal of Economic Eugenics Quarterly, 81(1): 81–91.
History, 60(3): 782–818. Horrell, Sara, and Jane Humphries. 1995. “Women’s
Goldin, Claudia, and Lawrence F. Katz. 2002. “The Labour Force Participation and the Transition to the
Power of the Pill: Oral Contraceptives and Women’s Male-Breadwinner Family, 1790–1865.” Economic
Career and Marriage Decisions.” Journal of Political History Review, 48(1): 89–117.
Economy, 110(4): 730–70. Jütte, Robert. 2003. Lust ohne Last: Geschichte der
Goldin, Claudia, and Kenneth Sokoloff. 1982. “Women, Empfängnisverhütung. Munich: Beck.
Children, and Industrialization in the Early Repub- Khoudour-Castéras, David. 2008. “Welfare State and
lic: Evidence from the Manufacturing Censuses.” Labor Mobility: The Impact of Bismarck’s Social
Journal of Economic History, 42(4): 741–74. Legislation on German Emigration before World
Goldstein, Joshua R., Tomáš Sobotka, and Aiva War I.” Journal of Economic History, 68(1): 211–43.
­Jasilioniene. 2009. “The End of ‘Lowest-Low’ Fer- Knodel, John E. 1974. The Decline of Fertility in Ger-
tility?” Population and Development Review, 35(4): many, 1871–1939. Princeton: Princeton University
663–99. Press.
Guinnane, Timothy W. 1997. The Vanishing Irish: Knodel, John E. 1988. Demographic Behavior in the
Households, Migration, and the Rural Economy in Past: A Study of Fourteen German Village Popula-
Ireland, 1850–1914. Princeton: Princeton University tions in the Eighteenth and Nineteenth Centuries.
Press. Cambridge and New York: Cambridge University
Guinnane, Timothy W., and Sheilagh Ogilvie. 2008. Press.
“Institutions and Demographic Responses to Shocks: Knodel, John E., and Etienne van de Walle. 1986. “Les-
Württemberg, 1634–1870.” Yale University Eco- sons from the Past: Policy Implications of Histori-
nomic Growth Center Discussion Paper 962. cal Fertility Studies.” In The Decline of Fertility in
Guinnane, Timothy W., Barbara S. Okun, and James Europe: The Revised Proceedings of a Conference on
Trussell. 1994. “What Do We Know About the Tim- the Princeton European Fertility Project, ed. Ansley
ing of Fertility Transitions in Europe?” Demography, J. Coale and Susan Cotts Watkins, 390–419. Princ-
31(1): 1–20. eton: Princeton University Press.
Guinnane, Timothy W., and Jochen Streb. 2011. “Moral Kocka, Jürgen. 1990. Arbeitsverhältnisse und Arbeiter-
Hazard in a Mutual Health-Insurance System: Ger- existenz: Grundlagen der Klassenbildung im 19. Jah-
man Knappschaften, 1867–1914.” Journal of Eco- rhundert. Bonn: Dietz.
nomic History, 71(1): 70–104. Kohler, Hans-Peter, Francesco C. Billari, and José Anto-
Hacker, J. David. 2003. “Rethinking the ‘Early’ Decline nio Ortega. 2002. “The Emergence of L ­ owest-Low
 Guinnane: The Historical Fertility Transition 613

Fertility in Europe during the 1990s.” Population Møller, Niels Framroze, and Paul Sharp. 2008.
and Development Review, 28(4): 641–80. “­Malthus in Cointegration Space: A New Look at
Lee, James Z., and Wang Feng. 1999. One Quarter of Living Standards and Population in Pre-industrial
Humanity: Malthusian Mythology and Chinese Real- England.” University of Copenhagen Department of
ities, 1700–2000. Cambridge and London: Harvard Economics Discussion Paper 08-16.
University Press. Mroz, Thomas A., and David R. Weir. 1990. “Structural
Lee, Ronald D. 1981. “Short-Term Variation: Vital Change in Life Cycle Fertility during the Fertility
Rates, Prices, and Weather.” In The Population His- Transition: France Before and After the Revolution
tory of England, 1541–1871: A Reconstruction, ed. of 1789.” Population Studies, 44(1): 61–87.
E. A. Wrigley and R. S. Schofield, 356–401. Cam- Murphy, Tommy E. 2010a. “Old Habits Die Hard
bridge and New York: Cambridge University Press. (Sometimes): Can Département Heterogeneity Tell
Lee, Ronald D. 1985. “Population Homeostasis and Us Something about the French Fertility Decline?”
English Demographic History.” Journal of Interdis- Bocconi University Innocenzo Gasparini Institute for
ciplinary History, 15(4): 635–60. Economic Research Working Paper 364.
Lee, Ronald D, and Michael Anderson. 2002. “Mal- Murphy, Tommy E. 2010b. “Persistence of Malthus
thus in State Space: Macro Economic–Demographic or Persistence in Malthus? Mortality, Income, and
Relations in English History, 1540–1870.” Journal of Marriage in the French Fertility Decline of the Long
Population Economics, 15(2): 195–220. Nineteenth Century.” Bocconi University Innocenzo
Livi-Bacci, Massimo. 1986. “Social Group Forerun- Gasparini Institute for Economic Research Working
ners of Fertility Control in Europe.” In The Decline Paper 363.
of Fertility in Europe: The Revised Proceedings of Murray, John E. 2007. Origins of American Health
a Conference on the Princeton European Fertility Insurance: A History of Industrial Sickness Funds.
Project, ed. Ansley J. Coale and Susan Cotts Wat- New Haven and London: Yale University Press.
kins, 182–200. Princeton: Princeton University Nardinelli, Clark. 1990. Child Labor and the Industrial
Press. Revolution. Bloomington: Indiana University Press.
Lucas, Robert E. 2002. Lectures on Economic Growth. Nicolini, Esteban A. 2007. “Was Malthus Right? A VAR
Cambridge and London: Harvard University Press. Analysis of Economic and Demographic Interactions
MacFarlane, Alan. 1986. Marriage and Love in Eng- in Pre-industrial England.” European Review of Eco-
land 1300–1840. New York: Blackwell. nomic History, 11(1): 99–121.
Marschalck, Peter. 1984. Bevölkerungsgeschichte Nipperdey, Thomas. 1994. Deutsche Geschichte 1800–
Deutschlands im 19. und 20. Frankfurt: Jahrhundert. 1866: Bürgerwelt und Starker Staat, Second edition.
Martin-Fugier, Anne. 1978. “La fin des nourrices.” Le Munich: Beck.
Mouvement Social, 105: 11–32. Notestein, Frank. 1945. “Population—The Long View.”
McKeown, Thomas. 1976. The Modern Rise of Popula- In Food for the World, ed. Theodore W. Schultz,
tion. London: Arnold. 36–57. Chicago: University of Chicago Press.
McLaren, Angus. 1978. Birth Control in Nineteenth- Rollet, Catherine. 1982. “Nourrices et nourrissons dans
Century England. New York: Holmes and Meier le département de la Seine et en France de 1880 à
Publishers. 1940.” Population, 37(3): 573–604.
Melton, James van Horn. 1988. Absolutism and the Rosenzweig, Mark R., and Kenneth I. Wolpin. 1980.
Eighteeenth-Century Origins of Compulsory School- “Testing the Quantity–Quality Fertility Model: The
ing in Prussia and Austria. Cambridge and New Use of Twins as a Natural Experiment.” Economet-
York: Cambridge University Press. rica, 48(1): 227–40.
Michael, Robert T., and Robert J. Willis. 1976. “Contra- Santow, Gigi. 1993. “Coitus Interruptus in the Twenti-
ception and Fertility: Household Production under eth Century.” Population and Development Review,
Uncertainty.” In Household Production and Con- 19(4): 767–92.
sumption, ed. Nestor E. Terleckyj, 27–93. New York: Saxonhouse, Gary, and Gavin Wright. 1984. “Two
National Bureau of Economic Research. Forms of Cheap Labor in Textile History.” Research
Mitch, David F. 1992. The Rise of Popular Literacy in in Economic History, Supplement 3: 3–31.
Victorian England: The Influence of Private Choice Schultz, T. Paul. 1981. Economics of Population. Read-
and Public Policy. Philadelphia: University of Penn- ing, Mass.: Addison-Wesley.
sylvania Press. Schultz, T. Paul. 1985. “Changing World Prices, Wom-
Mitch, David F. 2004. “Education and Skill of the Brit- en’s Wages, and the Fertility Transition: Sweden,
ish Labour Force.” In The Cambridge Economic 1860–1910.” Journal of Political Economy, 93(6):
History of Modern Britain, Volume 1, ed. Roderick 1126–54.
Floud and Paul Johnson, 332–56. Cambridge and Sharlin, Alan. 1986. “Urban–Rural Differences in Fer-
New York: Cambridge University Press. tility in Europe during the Demographic Transiton.”
Mitchell, B. R. 1980. European Historical Statistics In The Decline of Fertility in Europe: The Revised
1750–1975. New York: Facts on File. Proceedings of a Conference on the Princeton Euro-
Moehling, Carolyn M. 1999. “State Child Labor Laws pean Fertility Project, ed. Ansley J. Coale and Susan
and the Decline of Child Labor.” Explorations in Cotts Watkins, 234–60. Princeton: Princeton Univer-
Economic History, 36(1): 72–106. sity Press.
614 Journal of Economic Literature, Vol. XLIX (September 2011)

Sheps, Mindel C., and Jane A. Menken. 1973. Math- Marital Fertility in France, 1740–1911.” Population
ematical Models of Conception and Birth. Chicago: Studies, 48(2): 307–31.
University of Chicago Press. Westoff, Charles F. 1988. “Is the KAP-Gap Real?”
Sundstrom, William A., and Paul A. David. 1988. “Old- Population and Development Review, 14(2): 225–32.
Age Security Motives, Labor Markets, and Farm Woods, Robert. 1996. “The Population of Britain in
Family Fertility in Antebellum America.” Explora- the Nineteenth Century.” In British Population His-
tions in Economic History, 25(2): 164–97. tory: From the Black Death to the Present Day, ed.
Wanamaker, Marianne H. 2010. “Did Industrialization Michael Anderson, 281–358. Cambridge and New
Cause the American Fertility Decline? Evidence York: Cambridge University Press.
from South Carolina.” Unpublished. Woycke, James. 1988. Birth Control in Germany 1871–
Wehler, Hans-Ulrich. 1995. Deutsche Gesellschaftsge- 1933. London and New York: Routledge.
schichte 1849–1914. Munich: Beck. Wrigley, E. A., R. S. Davies, J. E. Oeppen, and R. S.
Wehler, Hans-Ulrich. 1996. Deutsche Gesellschafts- Schofield. 1997. English Population History from
geschichte 1815–1845/4. Zweiter Band (Dritte Family Reconstitution 1580–1837. Cambridge and
Auflage). New York: Cambridge University Press.
Weir, David R. 1984. “Life under Pressure: France and Wrigley, E. A., and R. S. Schofield. 1981. The Popu-
England, 1670–1870.” Journal of Economic History, lation History of England 1541–1871: A Recon-
44(1): 27–47. struction. Cambridge and New York: Cambridge
Weir, David R. 1994. “New Estimates of Nuptiality and University Press.
This article has been cited by:

1. Karen N. Eggleston,, Victor R. Fuchs. 2012. The New Demographic Transition: Most Gains in Life
Expectancy Now Realized Late in Life. Journal of Economic Perspectives 26:3, 137-156. [Abstract]
[View PDF article] [PDF with links]