The Effect of Labor‐Saving Technology on Longitudinal Fertility Changes

Abstract

In many subsistence communities worldwide, labor-saving technologies such as mechanized water pumps and grain mills are introduced during the first stages of economic acculturation. These technological changes affect the efficiency with which resources are acquired and processed. If women can produce the same amount of a good in less time and if the time savings is reallocated to less energetically costly activities, the positive shift in women's energy balance can affect fertility in a number of ways. Reproductive histories collected over the past ten years show that since the introduction of labor-saving technology Xculoc Maya women begin childbearing at a significantly younger age and have a greater annual probability of giving birth and higher age-specific and completed fertility. Access to wage labor, education, and the market economy remains limited. When a change in labor efficiency takes place under these conditions, parents appear to reallocate the saved time to reproduction rather to production. Such conditions may have significant implications for the rapid population growth occurring in many small-scale societies during the early stage of economic acculturation.

Full text

165
The Effect of Labor-Saving Technology on
Longitudinal Fertility Changes
Karen L. Kramer and Garnett P. McMillan
Department of Anthropology, Stony Brook University,
Stony Brook, NY 11794-4364, U.S.A. (karen.kramer@
sunysb.edu)/Behavioral Health Research Center of the
Southwest, Albuquerque, NM 87102, U.S.A. (gmcmil-
lan@bhrcs.org) 23 IX 05
In many subsistence communities worldwide, labor-saving
technologies such as mechanized water pumps and grain mills
are introduced during the first stages of economic accultur-
ation. These technological changes affect the efficiency with
which resources are acquired and processed. If women can
produce the same amount of a good in less time and if the
time savings is reallocated to less energetically costly activities,
the positive shift in women’s energy balance can affect fertility
in a number of ways. Reproductive histories collected over
the past ten years show that since the introduction of labor-
saving technology Xculoc Maya women begin childbearing at
a significantly younger age and have a greater annual prob-
ability of giving birth and higher age-specific and completed
fertility. Access to wage labor, education, and the market econ-
omy remains limited. When a change in labor efficiency takes
place under these conditions, parents appear to reallocate the
saved time to reproduction rather to production. Such con-
ditions may have significant implications for the rapid pop-
ulation growth occurring in many small-scale societies during
the early stage of economic acculturation.
The timing of the fertility and mortality decline characteristic
of the modern demographic transition varies among different
populations. National aggregate data indicate that a reduction
in fertility is well under way in most countries, but a very
different picture emerges when data from small-scale societies
in unindustrialized parts of the world are considered. For
example, national census data for Mexico report a decrease
in total fertility rate from 6.8 to 2.8 between 1960 and 2000
(Wilkie 2002). In contrast, during the same period of time,
fertility remained high in the rural Maya village we have stud-
ied in Yucatan, with mothers having a median completed
family size of 7.3 in 2000. This discrepancy between national
and village trends suggests that fertility processes through the
transition may be far more variable than aggregate data
suggest.
The classic demographic-transition model depicts a de-
crease in mortality, primarily through a reduction in child
䉷 2006 by The Wenner-Gren Foundation for Anthropological Research.
All rights reserved 0011-3204/2006/4701-0007$10.00
mortality, followed by a decline in fertility. It is during the
lag time between the drop in mortality and the drop in fertility
that natural increase is at its greatest and population growth
rates may reach unprecedented highs—a demographic situ-
ation common in parts of Africa, Latin America, the Middle
East, Asia, and elsewhere. The fertility reduction during the
later stages of the transition has been extensively documented
and its causes widely debated (Borgerhoff Mulder 1998; Cald-
well 1976; Kaplan and Lancaster 2000; Low 2000). Consid-
erably less is known about the earlier stages of the transition
and the factors that condition the lag between the decline in
mortality and the decline in fertility. The common explana-
tion has been a generational delay in fertility adjustment as
parents maintain the same level of fertility but more children
survive. This explanation, however, presupposes that smaller
families are universally desired. In this paper we address under
what economic conditions fertility might remain high or even
increase.
Most of the world’s subsistence populations, whether they
be hunter-gatherers, pastoralists, or agriculturalists, are be-
ginning to undergo economic changes as they move toward
inclusion in the labor market and a national economy. Eco-
nomic development is often linked to fertility reduction be-
cause the increase in the cost of children leads to a decline
in the demand for them (Bulatao and Lee 1983), but it can
also have unanticipated effects on local demography. Our
earlier research among the Maya of the Puuc region of the
Yucatan Peninsula, Mexico (Kramer and McMillan 1994,
1998, 1999), showed that the introduction of modern labor-
saving technology significantly lowered the age at first birth
sufficiently to allow women, if they chose, to bear an addi-
tional child over the course of their reproductive careers. Age
at first birth is a robust predictor of family size in natural-
fertility populations (Bongaarts 2002; Campbell and Wood
1998; Smallwood 2002), and now that some of the women
in the first study have completed their reproductive careers
we have an opportunity to test this effect. This study draws
on recent census and reproductive-history data to reassess the
question of the interaction between labor-saving technology,
economic development, and fertility patterns.
The Population and Data Collection
The Maya of Xculoc, a small, remote village on the Yucatan
Peninsula, are subsistence maize farmers, participating min-
imally in wage labor and the cash economy. The village con-
ditions described here are as they were during the 1992–93
study, when the village had no electricity, running water, or
motorized transportation. A gas-powered water pump and
maize mill were installed through a government development
program in the late 1970s, but little else changed in the village
in the next 20 years. Because of long distances to market
towns and the lack of motorized transportation, villagers re-
main removed from much involvement in wage labor or the

166 Current Anthropology Volume 47, Number 1, February 2006
market economy. Each household grows most of the food it
consumes, though in some households men occasionally leave
the village for several days at a time to engage in construction
work to finance the purchase of basic household items—cloth,
thread, machetes, medicine, and the like. Women and children
are not involved in wage labor or any other income-producing
enterprise, and in 25% of the 52 village households men never
participate in wage labor. Some households collect honey for
sale or occasionally exchange small quantities of maize for
goods such as vegetable oil, eggs, or candles; otherwise no
crops are grown for cash or trade.
Young men and women live, work, and eat in their parents’
households until they marry (mean age at marriage 21.6 for
males, 18.7 for females). Marriages are stable and monoga-
mous. Of adults living in the village, none reported “divorced”
as his or her marital status. Widows and widowers are usually
well into their 50s and 60s when they lose their spouses and
rarely remarry. Except for one widow, all mothers still in their
childbearing years are married. Of village residents over the
age of 30, 8% (3) of women and 9% (4) of men are unmarried.
Couples do not have access to modern contraception, and a
child is usually born within the first year of marriage. During
the first six months of life infants are breast-fed on demand.
Sometime after they are six months old, they begin to be fed
a supplemental maize gruel (mothers do not use store-bought
baby food or formula). Children are fully weaned at about
two years, when mothers are often pregnant with another
child.
The demographic data used here were collected in 1992–93,
2001, and 2003 (Kramer 2005; Kramer and McMillan 1998).
Village members were interviewed and asked their ages, their
birth dates, and the ages and birth dates of their parents,
siblings, and children. In the case of deceased relatives, age
at death and death date were requested. Constructing reliable
reproductive histories depends on accurate age data and a
precise count of all children born to a mother. Because few
villagers have written birth certificates issued at the time of
birth, a number of steps were taken to improve the accuracy
of the age estimates (Hill and Hurtado 1996; Howell 1979;
Pennington and Harpending 1993). First, all members of the
household were interviewed. Second, individuals were asked
for both their ages and their birth dates. And third, partici-
pants were asked to rank siblings and children by birth or-
der—a valuable technique for cross-checking age estimates.
The Maya of Xculoc are forthcoming in talking about de-
ceased family members and included deceased children and
siblings in their ranked birth orders. Only live births were
included in the tabulation of children ever born. Infants who
died very young were counted, but miscarriages were not.
1
1. The distribution of the population by age and sex was as follows:
males, 0–4 years, 31 (9.8%); 5–19 years, 60 (19%); 20–44, 48 (15.2%);
45⫹ years, 23 (7.3%); females, 0–4 years, 24 (7.6%); 5–19 years, 62
(19.6%); 20–44 years, 47 (17%); 45⫹ years, 25 (6%). The corresponding
figures for 2001 were as follows: males, 0–4 years, 35 (8%); 5–19 years,
85 (21%); 20–44 years, 53 (14.5%); 45⫹ years, 30 (7%); females, 0–4
The 1992–93 census and reproductive-history data show
that the incidence of child mortality was markedly low in this
community. Of the 284 births recorded for 57 living mothers,
12 deaths of children under the age of 16 were reported. (Four
of those deaths were due to Duchenne muscular dystrophy,
which afflicts one extended family.) Though the sample is
small, this indicates that 95% of children survived to age 16.
2
All children who were alive in the 1992–93 census were ac-
counted for in the 2001 census, in which only 1 child death
was reported. Several facts may account for this low child
mortality. First and foremost, the drinking water is clean.
There is no running surface water in Yucatan, and drinking
water has historically been collected from natural limestone
basins and, for the past century, from closed wells (fig. 1). In
addition, the villagers have an adequate diet in terms of both
calories and macronutrients and are within the normal range
of height for weight (Cotter 2002). The community’s re-
moteness and the low population density in general reduce
the impact of contagious disease, though immunization pro-
grams have no doubt produced further gains in infant and
child survival.
Although child mortality is low and has been for at least
the past 60 years, fertility has remained high. Reflecting this,
between 1992–93 and 2001 the population increased from
316 to 405, giving an average annual growth rate of 3.1%
3
—
comparable to those of the fastest-growing populations in the
world. The Xculoc sample thus presents an ideal situation for
observing potential fertility fluctuations following the drop
in child mortality.
Labor Efficiency, Female Energy Balance, and Reproductive
Function
Energy availability is an underlying constraint to fertility and
key to an understanding of variation in human reproductive
function. Ever since debate arose over the critical-weight hy-
pothesis for explaining age at menarche (Frisch and Revelle
1970; see overview in Ellison 2001 and Jasienska 2001), energy
balance (calorie intake relative to energy expenditure) has
been linked to various aspects of fertility. The relationship
between energy expenditure and female reproductive function
has been studied with respect to both exercise and sports in
the developed world and physical activity in traditional pop-
ulations. Among female athletes, research has focused on
heightened physical activity, weight loss, and impaired repro-
ductive function (Cumming 1990; Ellison 1990; Rosetta 1993,
1995). Among traditional populations, an increase in female
activity levels has been associated with postpartum anovu-
lation (Panter-Brick 1991; Panter-Brick and Ellison 1994).
years, 24 (6%); 5–19 years, 83 (20%); 20–44 years, 70 (17%); 45⫹ years,
25 (6%).
2. By comparison, 98% of U.S. children survive to age 16 (U.S. Bureau
of Census 1992, 9).
3. Calculated using an exponential model. Ten deaths and 97 births
occurred in the eight-year interval.

167
Figure 1. Women drawing water from well, Xculoc, 1993.
Recent research further suggests that peaks in levels of sub-
sistence labor—in this case farm work—are associated with
suppressed ovarian function in adult women even when not
accompanied by negative energy balance (Jasienska and El-
lison 1998, 2004).
Other studies have considered the effects of a shift to a
positive energy balance. Research in a nutritionally stressed
population of Ethiopian women found that the period of
postpartum amenorrhea and birth-interval length were cor-
related with a decrease in physical activity due to women’s
access to water taps (Gibson and Mace 2002). While the re-
lationship between hormones and energy balance is complex
and not fully understood, differences in energy balance appear
to be associated with the pace of sexual maturation (Cum-
ming, Wheeler, and Harber 1994; Ellison 1990; Rosetta 1990;
Strassman 1996; Ulijaszek 1995). A recent comparison of the
maturation trajectories of urban and rural Zambian girls, for
example, suggests that the time to completion of sexual ma-
turity may be mediated by nutritional condition, in this case
by the preferential feeding of pubescent girls (Gillett-Netting,
Meloy, and Campbell 2004).
Our previous research considered the effect of an increase
in labor efficiency and a decrease in the time women spent
in energetically costly activities on age at first birth. Following
the introduction of a gas-powered water pump and mill that
markedly increased women’s labor efficiency and a positive
tip in the energy balance, age at first birth dropped signifi-
cantly, from 21.2 to 19.5 years, among women who became
mothers (Kramer and McMillan 1998, 1999; fig. 2).
4
This
decrease was linked to physiological and behavioral factors
that also changed with the introduction of technology and
affected the age at which young women left home and mar-
ried. Time allocation data collected during 1992–93 were used
to compare how nulliparous reproductive-aged young women
reallocated their time when the labor-saving technology was
available. Individual time allocation budgets were gathered
using scan-sampling and focal-follow techniques (Altmann
1974; Borgerhoff Mulder and Caro 1985; Hames 1992). Scan
sampling, a direct observation method developed in animal
behavior research, produces an accurate estimate of the pro-
4. A multivariate logistic regression procedure was used to model the
probability of beginning a reproductive career as a function of the pres-
ence or absence of technology while controlling for the effects of a general
historical trend. Variables added to the model include a mother’s age at
risk, technology presence/absence, and year at risk (to measure for a
general historical effect) and the interaction effects between them. The
best-fit model from this set of predictor variables included age, quadratic
term for age, and technology presence/absence as significant (p p .0220)
in predicting the probability of a mother’s age at first birth. It is the
fitted probabilities from this model that are shown in figure 1. The year-
at-risk variable dropped out of the model as insignificant, suggesting that
there has not been an incremental decrease in women’s age at first birth
(Kramer and McMillan 1999, 509–11).

168 Current Anthropology Volume 47, Number 1, February 2006
portion of time an individual spends in each of a wide variety
of activities (Dunbar 1976). The results showed that when
labor-saving technology was available, young women (1) al-
located less time (70%) to collecting water and (2) did not
change the time they allocated to field work, domestic work,
or child care but (3) were almost twice as likely to spend time
in leisure activities (resting, sleeping, socializing, eating, re-
ligious participation, and personal maintenance). Thus, nul-
liparous young women reallocated the time spent in calorically
expensive activities to less energetically costly leisure activities,
potentially relaxing the physiological constraints on female
sexual maturation. When using the labor-saving technology,
women worked on the order of two and a half fewer hours
a day, a daily savings of an estimated 325 calories—an amount
sufficient, for example, to satisfy the increase in energy ex-
penditure required during the luteal phase of the menstrual
cycle (Strassman 1996).
When an increase in efficiency eases labor demand, the
marginal value (the value of a person’s labor when added to
the work already provided by other family members) of a
young woman’s help is also expected to diminish. In other
words, if a reproductive-aged daughter is less critical as an
economic contributor, from her parents’ point of view she
should be less constrained from leaving home and starting a
family of her own should the opportunity arise. From a young
woman’s point of view, conditions may become more favor-
able for initiating reproduction at a younger age if the change
in efficiency reduces the labor cost of supporting a family.
5
Although the predicted outcome of an earlier age at first
birth in natural-fertility populations is an increase in family
size, mothers may also reinvest the saved time in production
or leisure, lengthen birth intervals, or give birth to their last
child a little younger and have the same number of or fewer
children—changes characteristics of the demographic tran-
sition. Reproductive-history and census data collected in 2001
and 2003 were used to observe fertility trends since the drop
in age at first birth.
Fertility Trends
For two reasons, the sample used for the study includes
women who eventually become mothers. First, in the absence
of birth records, one of the shortcomings of retrospective
reproductive histories is that the true population of women
in the past cannot be known. For example, if a woman alive
5. Other things being equal, the same physiological and behavioral
factors (relaxed constraints on reproductive function and household labor
demands) that were hypothesized to effect a decrease in the age at first
birth following the increase in labor efficiency (Kramer and McMillan
1998) might be expected to produce shorter birth intervals. However,
the median birth interval before the introduction of labor-saving tech-
nology was 2.2 years (265 births to 57 women), which represents the
lower limit to birth-interval length without a dramatic increase in child
mortality (for an overview of studies, see Ellison 2001, 312). Consistent
with this, the introduction of labor-saving technology had no effect on
birth-interval length in the original study.
in Xculoc in the 1960s died or emigrated and had no relatives
alive during the 1992–93 census, she would not be counted
in the sample of women. This potentially underestimates the
true past population of women and biases population fertility
estimates. Second, we are interested in the effect of technology
on a woman’s continued fertility once she initiates repro-
duction. A woman may not have children for various phys-
iological or personal reasons. Among married Xculoc women
the level of childlessness is within the range of normal for
primary sterility reported in well-nourished natural-fertility
populations. Sterility is a primary cause of childlessness in
natural-fertility populations and, other things being equal, is
not expected to vary with respect to whether modern tech-
nology is available.
6
While including women who become
mothers reflects reproductive behavior once the decision has
been made to reproduce, fertility rates derived from this sam-
ple are likely to overestimate population parameters relative
to a sample that includes all reproductive-aged women.
To improve the quality of the age and birth-date data, we
included only women who were alive and therefore inter-
viewed in 1992 and had given birth to their first child after
1960. Women who were married but had not given birth
within five years (n p 3) and women who were married but
had given birth to their last child before the age of 26 were
excluded (n p 4). These women had either become widows
and were no longer at risk of conception or they or their
spouses’ physiological circumstances precluded their con-
ceiving again. Women who were interviewed in 1992 but had
since emigrated were excluded. Given these considerations, a
sample of 60 mothers was used in the following analyses.
Figure 3 plots age-specific fertility and the number of chil-
dren ever born to mothers 40 and older.
7
Mothers were strat-
ified by the decade in which they gave birth to their first child
because their reproductive careers in many cases spanned the
introduction of labor-saving technology in 1978. Women in
the 1960–69 cohort (n p 8) were past childbearing age in
2003 and had spent most of their reproductive careers prior
to the introduction of the technology. Women in the 1970–79
cohort (n p 5) were 40 and older in 2003 and had spent
most of their reproductive careers after the introduction of
the labor-saving technology. Women in the after-technology
cohort had a higher maximum number of children, signifi-
cantly higher age-specific fertility throughout their prime
childbearing years, and a higher completed family size (8.0
compared with 7.12, p p .0167) compared with the before-
6. Of the 55 women who have been married for more than five years,
5% (n p 3) have no children. This probable primary sterility is within
the range (3–5%) that Bongaarts and Potter (1982) report as normal for
a well-nourished natural-fertility population.
7. Total fertility rate is often used as an estimate of average completed
family size within a population and is calculated as a cross-sectional
measure by summing observed age-specific birthrates. Given the size of
the village, the sample of women in five-year age-groups is insufficient
to calculate a standard cross-sectional total fertility rate and children ever
born is used instead.

169
Figure 2. The fitted proportion of Xculoc Maya mothers ever having
given birth (n p 50), showing the median age at first birth at 0.5 prob-
ability and indicating, for example, that 25% of the women who started
childbearing after the introduction of labor-saving technology had given
birth to their first child by age 18 (after Kramer and McMillan 1999,
510).
Table 1. Results of the Best-Fit Model for the Annual Prob-
ability of a Xculoc Maya Mother’s Giving Birth
Variable Parameter Estimate p
Intercept ⫺0.4003 .0035
Age effect 0.0639
! .0001
Age
2
effect ⫺0.0011 ! .0001
Year-at-risk effect 0.0074 .0021
Age/technology
interaction
⫺0.0144 .0004
Presence/absence of
technology
0.3672 .0005
Cohort ⫺0.1035
! .0001
technology cohort. All of the women who started their re-
productive careers in the 1970s and were over the age of 40
in the 2003 census had 8–12 children.
Because most women who started childbearing after 1980
were still of childbearing age (89%, [n p 42] are 40 and
younger), to make the most of our data set and avoid some
of the problems of small sample sizes, we modeled recent
fertility trends with a logistic regression procedure to evaluate
the effect of labor-saving technology on the annual probability
of giving birth. Logistic regression is advantageous because it
maximizes sample size by treating each year a mother is at
risk of giving birth (ages 15 to 50) as the unit of observation.
The sample include 1,289 person-years at risk for the 60
mothers. Parameter and variance estimates from logistic re-
gression models are similar to those generated with standard
survival models (see Efron 1988 and Hill and Hurtado 1996
for application to reproductive events). A nested, or hierar-
chical, model-selection procedure is used to discriminate
among a set of predictor variables. We reduced the model by
comparing the drop in deviance from the full model to a chi-
square distribution with one degree of freedom (Collett 1991).
Predictor variables include age at risk, first-birth cohort, and
the presence or absence of technology for each year that a
mother was at risk of giving birth. A variable for the year at
risk was added to adjust for possible secular trends in the
probability of giving birth. Since mothers contributed more
than one person-year at risk, it was important to consider
the repeated-measures nature of the data. We modeled the
within-mother covariance structure using the generalized es-
timating approach (Fitzmaurice, Laird, and Ware 2004),
which reduces the bias in standard errors of parameter esti-
mates that is inherent in the use of classical methods for
repeated measures. For the purpose of this study, we assumed
a working within-mother covariance matrix that was order-
1 autoregressive. The best-fit model is given in table 1. The
adjusted odds ratios, derived by exponentiating parameter
estimates, are consistent with the patterns in figure 2. The
annual probability of giving birth is 1.5 times (e
.3672
p 1.44)
greater after the introduction of modern technology. Thus,
after adjusting for age at risk and a slight general historical
trend, there is a punctuated increase in the annual probability
of giving birth following the introduction of labor-saving
technology.
In sum, age at first birth as a predictor of completed fertility
assumes that conditions that affect the pace of childbearing
remain stable throughout a woman’s childbearing years. This
appears to be the case with the introduction of labor-saving
technology. The younger age at first birth coupled with a

170 Current Anthropology Volume 47, Number 1, February 2006
Figure 3. Age-specific fertility (mean, confidence interval, and p value)
for children ever had for Xculoc Maya mothers stratified by first-birth
cohort. Circles, before technology; squares, after technology. (The fertility
of women over 45 in the after-technology cohort is slightly lower than
for women in the 40–44-year age-group because two women in that
group have 8–12 children. All the women in the 45⫹ age-group have 8
children.)
significant increase in completed fertility in the 1970s cohort
suggests that birth intervals did not appreciably lengthen and
mothers did not stop childbearing earlier. Analysis of the
annual probability of giving birth, which includes younger
women, indicates no appreciable decline in fertility. There is
some suggestion, however, that childbearing strategies may
be becoming more diversified. Of mothers between the ages
of 25 and 35 in 2003, while 31% had between five and eight
surviving children, 69% had four or fewer children. This is
in keeping with the spate of economic changes that began to
occur in Xculoc during the mid- and late 1990s. Although
still subsistence farmers, households had almost twice as much
land under cultivation in 2003 as they did in 1992. A junior
high school was built in 1995, and in 1998 lines were built
connecting the village to electricity. A few years ago women
began to organize economic cooperatives through bank and
government loans, and in 2002, after a devastating hurricane
and almost complete crop failure, some young women began
for the first time to travel to work in a maquiladora. No doubt
these changes affect future trade-offs in childbearing strategies
and the way individual women spend their time—whether in
agricultural intensification, in wage labor, in school, or sup-
porting children.
8
8. The minimum number of children born to a mother does not fall
below four, the expected lower limit in a natural-fertility population (see
Bentley, Goldberg, and Jasienska 1993 for overview).
Changing Economic Opportunities and Fertility Options
Small-scale population studies and individual data can offer
valuable insight into variation in the course of the demo-
graphic transition that large-scale aggregate data often ob-
scure. The Maya of Xculoc have sustained high levels of fer-
tility through several generations of low child mortality—a
fertility pattern that is not apparent in national aggregate data,
which indicate that Mexico is well into the demographic tran-
sition. Yet, this demographic situation may be common to
many communities in the early stages of the transition to a
market economy (Pennington and Harpending 1993). Un-
covering the mechanisms that condition the maintenance or
even increase of fertility following the decline in child mor-
tality is important because this is the sector of the world’s
population that is experiencing the greatest growth.
When there is a change in labor efficiency, time and re-
sources can be reallocated to different ends—an issue nested
in what economists and behavioral ecologists call the quality/
quantity trade-off (Becker 1981; Blurton Jones 1989; Caldwell
1983; Handwerker 1986). Given a finite time and resource
budget, each additional child diminishes the time and re-
sources available for other children. When economic devel-
opment introduces more efficient means of production, par-
ents may reallocate the saved time to the production of goods
and services that augment child quality (increased investment
in education, skill acquisition, health, nutrition, material
goods, training, or social status)—one explanation for the

171
fertility drop during the modern demographic transition
(Becker and Lewis 1973; Kaplan 1996). Alternatively, in the
absence of opportunity for or benefit from increasing child
quality, parents may maintain or even increase reproductive
effort.
In Xculoc, as in many rural indigenous communities world-
wide, basic mechanized technologies and rudimentary health
care have been introduced, while at the same time partici-
pation in wage labor, the cash economy, and education re-
mains limited. At the time of this study there was no local
opportunity to work in education-based jobs, and parents
benefited little from increasing investment in their children’s
education beyond the schooling they received in the village.
Their lack of transportation and proximity to markets limited
their ability to engage in wage labor or to transform saved
time into income-producing enterprises. Rather, the saved
time appears to have been allocated to reproductive effort.
Trade-offs in time allocation are crucial to the effect of a
change in labor efficiency on fertility. Further examination of
those trade-offs will be fruitful in shifting the focus in ex-
plaining fertility differences from mode of production or cul-
tural norms to those aspects of the economic development
and labor environment that affect fertility variation through
the transition.
Acknowledgments
We thank the Maya for their graciousness in allowing Karen
Kramer to live among them. The 1992–93 data collection was
supported by a National Science Foundation Dissertation Im-
provement Grant awarded to James Boone and Christopher
Dore (BNS-9115249). Additional funding was provided by
the Tinker-Mellon Foundation and the University of New
Mexico. We gratefully acknowledge the support of the Na-
tional Institute of Health (AG19044-01) and the Center for
the Economics and Demography of Aging for funding the
2001 and 2003 phases of this research.
References Cited
Altmann, J. 1974. Observational study of behavior: Sampling
methods. Behaviour 49:227–67.
Becker, G. S. 1981. A treatise on the family. Cambridge: Har-
vard University Press.
Becker, G. S., and H. G. Lewis. 1973. On the interaction
between the quantity and quality of children. Journal of
Political Economy 81:s279–88.
Bentley, G., T. Goldberg, and G. Jasienska. 1993. The fertility
of agricultural and non-agricultural traditional societies.
Population Studies 47:269–81.
Blurton Jones, N. G. 1989. The cost of children and the adap-
tive scheduling of births: Towards a sociobiological per-
spective on demography. In The sociobiology of sexual and
reproductive strategies, ed. A. E. Rasa, C. Vogel, and E. Vo-
land, 265–82. London: Chapman and Hall.
Bongaarts, J. 2002. The end of the fertility transition in the
developed world. Population and Development Review 28:
419–43.
Bongaarts, J., and R. G. Potter. 1982. Fertility, biology, and
behavior: An analysis of the proximate determinants. New
York: Academic Press.
Borgerhoff Mulder, M. 1998. The demographic transition: Are
we any closer to an evolutionary explanation? Trends in
Ecology and Evolution 13:266–70.
Borgerhoff Mulder, M., and T. M. Caro. 1985. The use of
quantitative observational techniques in anthropology.
Current Anthropology 26:323–35.
Bulatao, R. A., and R. D. Lee. 1983. Determinants of fertility
in developing countries. New York: Academic Press.
Caldwell, J. C. 1976. Toward a restatement of demographic
transition theory. Population and Development Review 2:
321–66.
———. 1983. Direct economic costs and benefits of children.
In Determinants of fertility in developing countries, ed. R. A.
Bulatao and R. D. Lee, 458–93. New York: Academic Press.
Campbell, K. L., and J. W. Wood. 1988. Fertility in traditional
societies. In Natural human fertility: Social and biological
determinants, ed. P. Diggory, M. Potts, and S. Teper, 39–61.
Houndmills, Basingstoke: Macmillan.
Collett, D. 1991. Modeling binary data. London: Chapman
and Hall.
Cotter, W. 2002. Growth and body size among preadolescents
in a modern Maya society. MS.
Cumming, D. C. 1990. Physical activity and control of the
hypothalamic-pituitary-gonadal axis. Seminars in Repro-
ductive Endocrinology 8:810–12.
Cumming, D. C., G. D. Wheeler, and J. T. Harber. 1994.
Physical activity, nutrition, and reproduction. Annals of the
New York Academy of Sciences 709:55–76.
Dunbar, R. I. M. 1976. Some aspects of research design and
their implications in the observational study of behaviour.
Behaviour 58:58–78.
Efron, B. 1988. Logistic repression, survival analysis, and the
Kaplan-Meier curve. Journal of the American Statistical As-
sociation 83:414–25.
Ellison, P. T. 1990. Human ovarian function and reproductive
ecology: New hypotheses. American Anthropologist 92:
933–52.
———. 2001. On fertile ground: A natural history of human
reproduction. Cambridge: Harvard University Press.
Fitzmaurice, G. M., N. M. Laird, and J. H. Ware. 2004. Applied
longitudinal analysis. Hoboken: Wiley Interscience.
Frisch, R. E., and R. Revelle. 1970. Height and weight at
menarche and a hypothesis of critical body weights and
adolescent events. Science 169:397–99.
Gibson, M., and R. Mace. 2002. Labor-saving technology and
fertility increase in rural Africa. Current Anthropology 43:
631–37.

172 Current Anthropology Volume 47, Number 1, February 2006
Gillett-Netting, R., M. Meloy, and B. Campbell. 2004. Catch-
up reproductive maturation in rural Tonga girls, Zambia.
American Journal of Human Biology 16(6):1–12.
Hames, R. 1992. Time allocation. In Evolutionary ecology and
human behavior, ed. E. A. Smith and B. Winterhalder,
203–35. New York: Aldine de Gruyter.
Handwerker, W. P. 1986. The modern demographic transition:
An analysis of subsistence choices and reproductive con-
sequences. American Anthropologist 88:400–417.
Hill, K., and H. M. Hurtado. 1996. Ache life history. New
York: Aldine de Gruyter.
Howell, N. 1979. Demography of the !Kung. New York: Aca-
demic Press.
Jasienska, G. 2001. Why energy expenditure causes repro-
ductive suppression in women. In Reproductive ecology and
human evolution, ed. P. T. Ellison. Hawthorne, N.Y.: Aldine
de Gruyter.
Jasienska, G., and P. T. Ellison. 1998. Physical work causes
suppression of ovarian function in women. Proceedings of
the Royal Society of London B 265:1847–51.
———. 2004. Energetic factors and seasonal changes in ovar-
ian function in women from rural Poland. American Jour-
nal of Human Biology 16:563–80.
Kaplan, H. 1996. A theory of fertility and parental investment
in traditional and modern human societies. Yearbook of
Physical Anthropology 39:91–135.
Kaplan, H., and J. Lancaster. 2000. The evolutionary eco-
nomics and psychology of the demographic transition to
low fertility. In Adaptation and human behavior: An an-
thropological perspective, ed. L. Cronk, N. Chagnon, and W.
Irons, 283–322. New York: Aldine de Gruyter.
Kramer, K. L. 2005. Maya children: Helpers at the farm. Cam-
bridge: Harvard University Press.
Kramer, K. L., and G. P. McMillan. 1994. Demographic
changes and technological shifts in a modern Maya village.
Paper presented at the 93d annual meeting of the American
Anthropological Association, Atlanta, Ga.
———. 1998. How Maya women respond to changing tech-
nology: The effect of helping behavior on initiating repro-
duction. Human Nature 9:205–23.
———. 1999. Women’s labor, fertility, and the introduction
of modern technology in a rural Maya village. Journal of
Anthropological Research 55:499–520.
Low, B. S. 2000. Why sex matters: A Darwinian look at human
behavior. Princeton: Princeton University Press.
Panter-Brick, C. 1991. Lactation, birth spacing, and maternal
work-loads among two castes in rural Nepal. Journal of
Biosocial Science 23:137–54.
Panter-Brick, C., and P. Ellison. 1994. Seasonality of work-
loads and ovarian function in Nepali women. Annals of the
New York Academy of Sciences 709:234–35.
Pennington, R., and H. Harpending. 1993. The structure of
an African pastoralist community: Demography, history, and
ecology of the Ngamiland Herero. Oxford: Oxford Science
Publications/Clarendon Press.
Rosetta, L. 1990. Biological aspects of fertility among Third
World populations. In Fertility and resources, ed. J. Landers
and V. Reynolds, 18–34. Cambridge: Cambridge University
Press.
———. 1993. Female reproductive dysfunction and intense
physical training. Oxford Reviews of Reproductive Biology
15:113–41.
———. 1995. Nutrition, physical workloads, and fertility reg-
ulation. In Human reproductive decisions, ed. R. I. M. Dun-
bar, 52–75. London: St. Martin’s Press and the Galton
Institute.
Smallwood, S. 2002. The effect of change in the timing of
childbearing on measuring fertility in England and Wales.
Population Trends 109:36–45.
Strassman, B. 1996. The evolution of endrometrial cycles and
menstruation. Quarterly Review of Biology 71:181–220.
Ulijaszek, S. 1995. Human energetics in biological anthropology.
Cambridge: Cambridge University Press.
U.S. Bureau of the Census. 1992. Statistical abstract of the
United States, 1992. Washington, D.C.
Wilkie, J. W., ed. 2002. Statistical abstract of Latin America.
Vol. 38. Los Angeles: University of California Press.