The Impact of Economic Uncertainty on Fertility Cycles: The Case of
the Post WWII Baby Boom
Bastien Chabé-Ferret
†‡
∗
Paula Gobbi
§
February 1, 2016
Abstract
Using the US Census waves 1940-1990 and CPS 1990-2010, we look at how economic uncertainty
affected fertility cycles over the course of the XXth century. We use cross-state and cross-cohort
variation in the volatility of income per capita to identify the causal link running from uncertainty
to fertility. We find that economic uncertainty, measured by the standard deviation of either the
cyclical component of income per capita or of income growth during fertile years, has a large and
robust negative effect on completed fertility. We hypothesize that a greater economic uncertainty
increases the risk of large consumption swings, which individuals mitigate by postponing fertility
and ultimately decreasing their completed fertility. Accounting for differences in volatility over time
accounts for at least 20% of the one child variation observed during the post WWII baby boom.
∗
†
FNRS and IRES, Université catholique de Louvain, email: [email protected]
CCPR, University of California Los Angeles, email: [email protected]
§
FNRS and IRES, Université catholique de Louvain, email: [email protected]
‡
1
Introduction
“In sum, economic growth as such played an inconsistent and at most a marginal role in the story of
the baby boom.” Van Bavel and Reher (2013)
The past two centuries have witnessed drastic demographic changes in developed economies. Overall
completed fertility declined from high fertility levels (over five children per woman in 1800 in the US)
to levels around the replacement rate (two children per woman, level reached in the US by cohorts
born in the 1950s) in a phenomenon known as the demographic transition (Greenwood and Seshadri
2002). It is generally accepted that this demographic transition went hand-in-hand with the rise in
living standards over the same period (although the causal link between the two series remains widely
debated) as show in Figure 1.
Figure 1: Completed fertility and GDP per capita per cohort in the US
Source: US Census & Clio-Infra, www.clio-infra.eu
This transition was overall rather smooth, except for a marked spike of about one child per woman,
which occurred (in the 1950s and 1960s) for cohorts born between 1930 and 1940. This marked spike,
also known as the (post-war) baby-boom and illustrated in Figure 2, has drawn a lot of attention from
demographers and economists. In particular, scholars have focused on the effect of fertility cycles on
the age structure: a tinier cohort brought about by a drop in the fertility rate is said to yield first a
demographic dividend as the active population outgrows the number of dependents (Bloom, Canning,
and Sevilla 2003), while it puts pressure on Pay-As-You-Go systems as the larger older cohort ages
(Weil 2006).
1
The determinants of fertility cycles are still hotly debated among demographers. In turn, only a handful
of recent economic papers deal with the issue (we cover this literature in Section 2). Notwithstanding,
uncovering the roots of such swings in fertility would allow to more accurately predict changes in the
size of populations, their age structure and, incidentally, the robustness of Pay-As-You-Go systems to
these changes.
Figure 2: Completed fertility in the US
In this paper, we argue that the episode of soaring economic uncertainty that took place during the
1930s, due to the Great Depression and the on-set of WWII has negatively affected the completed
fertility of cohorts born around 1910 even further than what economic development alone would have
predicted. In this view, the post-WWII baby boom can be interpreted as a conjunction between
return to trend values of fertility and a period of very low economic uncertainty. Uncertainty about
future streams of income has been found to cause birth postponement. Our aim here is to show
that an episode of prolonged economic uncertainty, particularly in a context where instruments to
insure against adverse shocks or smooth consumption across fertile and non fertile periods were rare,
ultimately affected completed fertility substantially.
To this end, we exploit cross-state and cross-cohort variation in the volatility of the cyclical component
of income per capita in the US to identify the effect of economic uncertainty on completed fertility.
We look at 5-year cohorts of women who have most likely completed their lifecycle fertility (age 40 to
59) using the US Census waves 1940-1990 and the June fertility supplement of the CPS 1990-2010. We
compute for each of these women the economic conditions during fertile years1 in her state of residence
from income per capita data. The information on income per capita by state comes from the Bureau of
1
We take age 21-29 as a benchmark for fertile years. We study the sensitivity of the results to a change in that
assumption in Appendix X.
2
Economic Analysis (BEA) for the years 1938 to 1990, enriched by Fishback and Kachanovskaya (2010)
for the period 1919 to 1938. We compute the uncertainty measure in two ways. First we extract the
cyclical component of income per capita by taking the residual variation when a state-specific log-linear
trend is removed. Then we assign to each woman the standard deviation of the cyclical component
over her fertile years. Alternatively, we compute the year-to-year growth rate in income per capita and
use the standard deviation of this growth rate over fertile years. Using 5-year cohort times state fixed
effects, we find a large negative effect of economic uncertainty that can account for at least 20% of the
one child variation in completed fertility between the cohort born in 1910 and that born in 1935. This
effect is robust to the inclusion of controls for alternative mechanisms, such as infant and maternal
mortality, income relative to aspirations, labor market conditions for women and electrification.
We believe that the effect of economic uncertainty on fertility that we identity actually captures a lower
bound of the true effect. Indeed, our strategy of using 5-year birth cohort times state fixed effect,
if most convincing in terms of identification, is likely to downward bias our coefficient of interest as
a substantial part of the variations in economic uncertainty actually occurred nation-wide, which is
netted out by the fixed effects. This is why, in a second step, we plan to build a model that can
reproduce the demographic transition as well as fertility cycles. A calibration exercise will be used in
order to assess the quantitative relevance of the mechanism we propose.
We also plan to explore why equivalent variations in economic uncertainty may not have had the
same effect on completed fertility in more recent years. We hypothesize that the medically induced
expansion of the fertility window as well as the emergence of better instruments to insure against job
losses and other income shocks through social transfers and deeper financial development, led to a
decrease in the effect of economic uncertainty on completed fertility.
The paper is organized as follows: first, in Section 2 we review the literature that has studied both
trends and cycles in fertility rates over time as well as the literature looking at the relationship between
economic uncertainty and fertility rates. We then expose a theoretical model that explicitly shows how
uncertainty might affect fertility in Section 4. We identify the causal effect of uncertainty on fertility
in Section 5.
2
Literature review
We first provide a literature review on fertility over time in the United States, distinguishing between
studies that looked at the trend in fertility from those who looked at the cyclical movements of
fertility rates. We then discuss the literature that has looked at the relationship between uncertainty
and fertility.
3
2.1
Demographic transition and the Baby-Boom
To date, there are few papers dealing at the same time with the demographic transition and the
baby boom. Demographers generally divide the demographic transitions into two separate parts. The
first, pre-WWII, demographic transition is caused by declining infant mortality rates, urbanization
and secularization (Thompson 1929; Landry 1934; Notestein 1945; Davis 1945). The second, post1970, demographic transition, arises from the diffusion of individualistic norms and the spread of birth
control methods (Lesthaeghe 1983; van de Kaa 1987). The roots of the baby-boom that happened in
between are still hotly debated among demographers (Van Bavel and Reher 2013), but also among
the few economists who studied the question.
The first attempt to theorize the baby boom was made by Easterlin (1966) who suggested that fertility
cycles were driven by the relative economic conditions young adults face with respect to those in vigor
when they were children. The idea is that if the income that adults of childbearing age dispose of is
large relative to their material aspirations, proxied by the income that was available when they were
children, then fertility tends to increase. This hypothesis gave rise to a large literature trying to test
it, which was summarized by Macunovich (1998). Hill (2015) recently offered a rigorous test using
cross-state variation in the US and found that the Easterlin hypothesis can account for 12% of the
post WWII baby boom. Contributing to a better understanding of the relationship between economic
conditions and fertility cycles, Jones and Schoonbroodt (2014) provide a theoretical framework to
account for the pro-cyclicality of fertility rates. Such model allows to explain the bust of fertility rates
during the Great Depression and the boom for the next generation.
Greenwood, Seshadri, and Vandenbroucke (2005) provide an explanation for both the secular decreasing pattern of fertility and the baby boom. For them, decreasing fertility is caused by raises in real
wages, which increased the opportunity cost of being out of the labor market and therefore of having children. The baby boom is the response of an atypical increase in technological progress in the
household sector, which lowered the cost of having children. This theory has however been rejected
by Bailey (2011), who uses two arguments: (i) electrification and fertility at the county levels are
negatively correlated; (ii) the Amish community also experienced a baby boom, despite their limited
use of modern household technologies (see also Greenwood, Seshadri, and Vandenbroucke (2015)).
Another possible explanation of the baby boom is the improvement in maternal health. Albanesi
and Olivetti (2014) support this thesis using cross-state variation in the magnitude of the decline in
pregnancy-related mortality. The rise in fertility rates then comes from a decline in the risk of dying
during delivery while the following decline of fertility rates, at the end of the baby boom, is due to an
increase in female human capital. More generally, Simon and Tamura (2009) stress the importance of
declining adult mortality for the secular declining fertility trend.
Doepke, Hazan, and Maoz (2015) suggest that WWII increased the demand for female labor and
this had a persistent effect on female labor force participation after the war. These women who
remained in the labor force after the war, made the competition for jobs tougher for younger women.
This asymmetric effect on the female labor market between the older and the younger cohorts was
4
responsible for the baby-boom, generated by these younger women who remained out of the labor
market. Using a similar argument, Bellou and Cardia (2014) show that the Great Depression induced
the cohort of women born between 1896 and 1910 (the “D-cohort”) to enter the labor force. This
pushed wages down and reduced labor opportunities for younger cohorts. These younger cohorts
faced a low opportunity cost in terms of foregone labor income and stayed at home and had children
(generating the baby-boom). Using commercial failures per state as a proxy for the impact of the great
Depression, they show that the baby boom was larger where the Great Depression hit the harder. Once
the D-cohort retired and freed female labor positions, then younger cohorts entered the labor force
and this increased the opportunity cost to childrearing and hence produced a bust in fertility rates.
The drawback of this explanation is that the female labor force participation shows a steady increasing
pattern over cohorts and therefore cannot explain all the fertility cycles of the XXth century.
2.2
Uncertainty and fertility
A number of studies have looked at the impact of economic uncertainty on fertility outcomes. Using
German panel data, Hofmann and Hohmeyer (2013) show that fertility is negatively affected by strong
economic concerns. Kind and Kleibrink (2013), also using German Data, show that uncertainty at
the individual level leads to a postponement of the first birth in Germany for the years 2001-2011.
On the contrary, they show that uncertainty at the macro level leads women to anticipate their first
birth. Schneider (2015) looks at the effect of state-level economic conditions on birth rates in the US
during the Great Recession. He finds that the negative effect of the crisis on fertility goes through both
increased economic hardship as well as higher economic uncertainty. Ranjan (1999) also suggests that
the low fertility rates in the former Soviet Republics and Eastern European countries can be explained
by the high uncertainty that followed the transition from a controlled economy to a market economy.
Hondroyiannis (2010) uses European panel data and shows that economic uncertainty has a negative
impact on fertility. Another important contribution is provided by Pommeret and Smith (2005), who
incorporate the fertility choice into a stochastic growth model. They find that if the inter-temporal
elasticity of substitution in consumption is small (lower than one), a higher level of uncertainty should
affect fertility negatively.
One plausible mechanism through which uncertainty affects fertility is due to the irreversibly expenditure associated with the birth of a child (Ranjan 1999). Parents tend to postpone their fertility
in order to have children when uncertainty is solved. Fertility postponement might also imply lower
fertility through two other mechanisms. First, women who become mothers later have accumulated
human capital and work experience. Therefore, the opportunity cost for raising children in terms of
foregone income at a later age is higher and the desired fertility might then be reduced compare to
what would have been the case if these women would have started their maternity earlier in life. The
second mechanism is due to the biological decline in women’s fecundity as they grow older, which
could prevent them to reach their desired fertility. With either mechanism at play, a cohort that faced
high uncertainty during their early reproductive years end up having less children.
5
The following section shows preliminary descriptive relationships between uncertainty and fertility
over time. Section 4 provides a theory that accounts for the main mechanisms that can account for
a relationship between economic uncertainty and completed fertility. Section 5 gives evidence for a
causal relationship running from uncertainty to fertility.
3
3.1
Fertility and uncertainty in the data
Facts
Fertility rates show a steadily decline since the midst of the XIXth century in a phenomenon known
as the fertility transition and then the second demographic transition, as discussed in the previous
section. Along this long run trend, there are several deviations from it.
The first deviation from the trend that we observe in Figure ?? is caused by the Great Depression:
a long period of high unemployment rates pushes fertility down (and female labor force participation
up). On the one hand, a negative income effect (having and raising children cost a lot of money, people
can’t afford to have many when facing high unemployment risk) and on the other hand an opportunity
cost of female time effect (relative to that of men) that goes in the same direction as they enter the
labor force to make up for the risk of husband’s job loss. This likely caused an even faster decline in
birth rates during the 1930s and until 1943 due to the large negative income shock due to the high
rates of job losses during the Great Depression for the cycle.
The subsequent boom comes from the return to full employment around the end of WWII. [WRITE
MORE]
And the bust of the years 1700-1800’s is the return on the long-term decreasing trend, coming from
gender biased technological change (and well captured by the decrease in the wage and education
gender gap over the 20th century).
Jones and Schoonbroodt (2014) also claim it’s baby bust, boom then bust again, which is caused
by the Great Depression and then cyclical behavior of fertility (it’s a macro model, using dynastic
altruism, with some room for fertility to overshoot when returning to the long-run trend). We differ
in that we bring micro evidence and place the story in the centennial decreasing fertility transition,
so we don’t rely on cyclical behavior.
Figure ?? of Appendix ?? shows the same figure using a political uncertainty instead of the economic
uncertainty measure.
4
Model
Ut = ln Ct + β ln nt
6
where
Ct = W t P t − n t θ
and
Pt ∼ B(p)
θ is the goods cost of having one child, W the wage rate, P is the stochastic process that determines
if you are employed or not. It is a Bernoulli process with probability p of being employed, which may
vary with economic conditions. Potentially, it can be extended to multiple periods, not sure we need
that though, and it’s probably robust to fancier stochastic processes. The idea is that then C becomes
a random variable. I note C̄ its mean. What we do then with Paolo [the whole thing is inspired by
Kalemli-Ozcan] is to approximate the value of utility around the mean of the stochastic variable, using
a Taylor approximation (the so-called Delta method):
C − C̄
U (C̄, n, C) = ln C̄ + β ln n + C − C̄ UC (C̄) +
2
2
UCC (C̄)
And then we take the expectation of that:
E(C − C̄)
=0
by definition
2
E[(C − C̄) ] = V (C) by definition
so we get:
EU (C̄, n, C) = ln C̄ + β ln n −
V (C)
C2
where
V (C) = W 2 p(1 − p)
So clearly, if p > 1/2 (meaning unemployment rate lower than 1/2, sounds credible) and decreases
because there is a crisis, then the variance term increases (which is bad for expected utility it enters
utility negatively). People react by increasing consumption (because it decreases the variance term),
which means they necessary substitute for fewer children.
5
Identification
This section first discusses the data sources for the economic conditions and fertility rates, by state
over time. Then, we discuss the identification of the causal relationship of economic uncertainty on
fertility rates.
5.1
Data
Economic conditions
7
We use state-level income per capita data provided by the Bureau of Economic Analysis (BEA) and
complemented by Fishback and Kachanovskaya (2010). We first separate the trend from the cyclical
component. This allows to separate the effect of economic development from that of the business
cycle.
We assume that, in each state s, income per capita yst fluctuates around a balanced growth path, with
a (constant) growth rate γ such as,
yst = ȳs0 eγs t + µst ,
(1)
where ȳs0 is a constant, t are years between 1919 and 1999, and µ is the cyclical component. We
estimate ȳs0 and γs by minimizing the sum of square residuals
1999
X
(ysi − µsi )2 .
i=1919
We then compute the mean of the cyclical component µ and its standard deviation over 9-year periods.
For notation purposes, let us define the following:
trendst = ȳs0 eγs t
t+4
1 X
µsi
9
i=t−4
v
u t+4
X
1u
= t
(µsi − µ̄st )2
3
cyclest = E(µst )|t∈[t−4,t+4] =
volatilityst = σ(µst )|t∈[t−4,t+4]
i=t−4
Figure 3 illustrates how we compute these measures for a given state (here California):
Figure 3: Trend / cycle decomposition of the income per capita series
The left panel shows the income per capita series for California and its log-linear fit. For the cohort
born in 1950, for example, we assign to each woman the economic conditions around 1975 when these
8
women were 25 years old. The circle on the balanced growth path denotes the trendCA,1975 , while the
distance to the actual series is µCA,1975 . The right panel provides a zoom around 1975 in order to
show how we compute the mean of µCA,t from 1971 to 1979 to obtain cycleCA,1975 (the black square)
and its standard deviation to get volatilityCA,1975 (the black circle).
Fertility
We compute completed fertility rates by birth cohort from US Census waves 1940-1990 and the June
fertility supplement of the CPS 1990-2010. This data is taken from Integrated Public Use Microdata
Series (IPUMS). We restrict the sample to women between 40 and 59 years old for whom we have
information on economic conditions when they were fertile. We therefore obtain a sample of 3,617,102
women who were born between 1898 and 1970. There is a small discontinuity in the sample selection
as the children ever born question was asked only to ever-married females in the 1940 to 1960 Census
waves2 .
5.2
Causal effect of uncertainty on fertility
The relationship between the number of children ever born n to a woman i, residing in state s, born
in year b, and surveyed at age a, and the economic uncertainty she confronted during her fertile years
21-29 is given by:
nisba = β0 + β1 volatilitysb + β2 Xi + β3 Xsb + ∆s × Φc + Γa + isba
(2)
We estimate the coefficients by OLS, and compute standard errors clustered at the state level. Our
coefficient of interest is β1 , which measures the effect of volatility of the cyclical component of income
per capita during fertile years on fertility. The key to our identification strategy comes from the
inclusion of a set of fixed effects for every 5-year birth cohort times state cell. Indeed, using only
variation within such a small cell only makes us confident that our variable of interest is uncorrelated
to the error term. To improve our confidence in the causal interpretation of β1 , we include Xi , a set
of individual level controls, here mainly a 5-level categorical variable for educational attainment, and
Xsb , a vector of year of birth and state-specific controls, such as economic conditions (trend and cycle)
as well as infant and maternal mortality. We also include fixed effects for age at survey, which aim
at netting out any time-invariant systematic variation in children ever born due to the age at which
women were surveyed (40 to 59)3 . Baseline results are shown in Table 1.
In all specifications, the coefficient on volatility remains significant and negative. The magnitude of the
effect does not seem to change with the inclusion of further controls. We can illustrate the magnitude
of the effect by using the difference between average volatility for the cohort born in 1910 (the lowest
point before the baby boom) and that for the cohort born in 1935 (the peak of the baby boom). This
2
This issue though is likely to be very minor as a large proportion of women had been ever married by age 40 up to
1960.
3
The idea is that some woman might have had children after being surveyed (though this bias is likely to be rather
small), while some others might have not been surveyed because they died before.
9
The dependent variable is children ever born
(1)
(2)
(3)
(4)
(5)
volatility
trend
cycle
-2.623∗∗∗
(0.401)
-0.128∗∗
(0.048)
-0.201∗∗∗
(0.037)
-2.587∗∗∗
(0.401)
-0.090∗
(0.050)
-0.211∗∗∗
(0.037)
-2.861∗∗∗
(0.369)
-0.093∗
(0.050)
-0.212∗∗∗
(0.038)
-0.001∗∗∗
(0.000)
-2.565∗∗∗
(0.345)
-0.071
(0.050)
-0.228∗∗∗
(0.039)
-0.000
(0.000)
-0.008∗∗∗
(0.001)
-2.422∗∗
(1.056)
-0.077
(0.049)
-0.056
(0.057)
-0.000
(0.000)
-0.008∗∗∗
(0.001)
-0.371∗∗∗
(0.084)
-2.057
(6.129)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
3617102
0.065
3617102
0.101
3608338
0.101
3608338
0.101
3608338
0.101
infant mortality
maternal mortality
level2
volatility2
(state of residence × cohort) F.E.
age at survey F.E.
education
Observations
R2
∗
p < 0.10,
∗∗
p < 0.05,
∗∗∗
p < 0.01. S.E. clustered at the state level in parenthesis.
Table 1: Relationship between fertility and economic uncertainty, using cyclical component
difference is of about 0.08, which multiplied by a coefficient of 2.5 gives a 0.2 variation in children ever
born, so about 20% of the overall variation.
In Table 2, we show the same exercise but using the mean and standard deviation of the growth rate
in income per capita over fertile years as measures of cycle and volatility respectively. Results are very
consistent with the alternative specification.
10
The dependent variable is children ever born
(1)
(2)
(3)
(4)
(5)
volatility
cycle
trend
-0.383∗∗∗
(0.062)
-0.090∗
(0.050)
-0.146∗∗∗
(0.050)
-0.347∗∗∗
(0.064)
-0.074
(0.049)
-0.104∗
(0.053)
-0.362∗∗∗
(0.065)
-0.108∗∗
(0.052)
-0.114∗∗
(0.053)
-0.001∗∗∗
(0.000)
-0.346∗∗∗
(0.065)
-0.152∗∗∗
(0.049)
-0.104∗
(0.054)
-0.000
(0.000)
-0.008∗∗∗
(0.001)
-0.463∗
(0.241)
0.108∗∗
(0.050)
-0.124∗∗
(0.054)
-0.000
(0.000)
-0.006∗∗∗
(0.001)
-0.941∗∗∗
(0.155)
-0.365
(0.501)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
3622404
0.064
3622404
0.100
3612440
0.100
3612440
0.101
3612440
0.101
infant mortality
maternal mortality
cycle2
volatility2
(state of residence × cohort) F.E.
age at survey F.E.
education
Observations
R2
∗
p < 0.10,
∗∗
p < 0.05,
∗∗∗
p < 0.01. S.E. clustered at the state level in parenthesis.
Table 2: Relationship between fertility and economic uncertainty, using growth
References
Albanesi, Stefania, and Claudia Olivetti. 2014. “Maternal health and the baby boom.” Quantitative
Economics 5 (2): 225–269 (July).
Bailey, Martha. 2011. “Did improvements in household technology cause the baby boom? Evidence from electrification, appliance diffusion, and the Amish.” American Economic Journal:
Macroeconomics 3 (2): 189–217.
Bellou, Andriana, and Emanuela Cardia. 2014. “Baby-Boom, Baby-Bust and the Great Depression.”
Unpublished.
Bloom, David, David Canning, and Jaypee Sevilla. 2003. “The Demographic Dividend: A new
perspective on the ecocnomic consequences of population change.” Santa Monica, CA: RAND
Corporation.
Davis, Kingsley. 1945. “The World Demographic Transition.” The Annals of the American Academy
of Political and Social Science 237:1–11.
Doepke, Matthias, Moshe Hazan, and Yishay D. Maoz. 2015. “The Baby Boom and World War II:
A Macroeconomic Analysis.” Review of Economic Studies 82 (3): 1031–1073 (July).
Easterlin, Richard A. 1966. “On the relation of economic factors to recent and projected fertility
changes.” Demography 3 (1): 131–153.
Fishback, Price V, and Valentina Kachanovskaya. 2010. “In search of the multiplier for federal
11
spending in the states during the Great Depression.” Technical Report, National Bureau of
Economic Research.
Greenwood, J., A. Seshadri, and G. Vandenbroucke. 2015. “Measurement without theory, once
again.” Journal of Demographic Economics 81 (3): 317–329.
Greenwood, Jeremy, and Ananth Seshadri. 2002. “The US demographic transition.” The American
Economic Review 92 (2): 153.
Greenwood, Jeremy, Ananth Seshadri, and Guillaume Vandenbroucke. 2005. “The Baby Boom and
Baby Bust.” American Economic Review 95 (1): 183–207.
Hill, Matthew J. 2015. “Easterlin revisited: Relative income and the baby boom.” Explorations in
Economic History 56:71–85.
Hofmann, Barbara, and Katrin Hohmeyer. 2013. “Perceived economic uncertainty and fertility:
evidence from a labor market reform.” Journal of Marriage and the Family 75:503–521.
Hondroyiannis, George. 2010. “Fertility determinants and economic uncertainty: an assessment using
European panel data.” Journal of Family Issues 31:33–50.
Jones, Larry E., and Alice Schoonbroodt. 2014. “Baby Busts and Baby Booms: The fertility response
to shocks in dynastic models.” Unpublished.
Kind, Michael, and Jan Kleibrink. 2013. “Sooner or Later - Economic Insecurity and the Timing of
First Birth.” RUHR Economic Papers.
Landry, Adolphe. 1934. La révolution démographique : études et essais sur les problèmes de population. Paris, Ined (réédition 1982).
Lesthaeghe, Ron. 1983. “A Century of Demographic and Cultural Change in Western Europe: An
Exploration of Underlying Dimensions.” Population and Development Review 9 (3): 411–435.
Macunovich, Diane J. 1998. “Fertility and the Easterlin hypothesis: An assessment of the literature.”
Journal of Population Economics 11 (1): 53–111.
Notestein, Frank W. 1945. Pages 36–57 in Population – the long view. in Schultz (dir.), Food for the
World, Chicago, Chicago University Press.
Pommeret, Aude, and William T. Smith. 2005. “Fertility, Volatility, and Growth.” Economic Letters
87:347–353.
Ranjan, Priya. 1999. “Fertility behaviour under income uncertainty.” European Journal of Population
15:25–43.
Schneider, Daniel. 2015. “The Great Recession, Fertility, and Uncertainty: Evidence From the United
States.” Journal of Marriage and Family 77 (5): 1144–1156.
Simon, Curtis, and Robert Tamura. 2009. “Secular Fertility Declines, Baby Booms and Economic
Growth: International Evidence.” MPRA paper 41669.
Thompson, Warren S. 1929. “Population.” American Journal of Sociology 34 (6): 959975.
12
Van Bavel, Jan, and David Reher. 2013. “The Mid-20th Century Baby Boom in the West: What
We Know and What We Need to Know.” Population and Development Review 39 (2): 257–288.
van de Kaa, Dick J. 1987. “Europes second demographic transition.” Population Bulletin 42 (1):
1–59.
Weil, David N. 2006, April. “Population aging.” NBER Working Paper No. 12147.
13