,...,V .. V... 1. ....V _ . _..._V .VTTTV ,1 .3. V V 7 ..V.V ‘ .V . _ .. . . . . . . A
w . V ‘ V .V . . . . V . ‘V .V . V , . . V a . ~
, V V I . V. . V .V, . V , V

V v A V V V V ‘» . . . . x. . ‘ ,V V . . V V « ,

MICH|GAN STATE UNIVERSITY LIBRARIES
l

Ill/{IllIll/lllll’l/llllMl!I lll/l/I/Hllill l

l
l
3 1293 00895 2487

 

 

This is to certify that the

thesis entitled

INTERGENERATIONAL SUPPORT NETWORKS

AND FERTILITY IN RWANDA

presented by
JANE EILEEN VANDER HAAR-

has been accepted towards fulfillment
of the requirements for

 

MoA. degree in SOCIOLOGY

Major professor

 

Dategflw‘ ’1: '6’ cl 0

0-7639 MS U is an Winona: Action/Equal Opportunity Institution

 

 

 

F LIBRARY
Michigan State
1 University

L A J

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

J

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—7

MSU In An Affirmative ActiorVEquel Opportunity lndltution
cmmut

 

 

 

 

INTERGENERATIONAL SUPPORT NETWORKS
AND FERTILITY IN RWANDA

BY

Jane Eileen Vander Haar

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

fli'MASTER OF ARTS

Department of Sociology

1990

445).. v 4:; '0

ABSTRACT

INTERGENERATIONAL SUPPORT NETWORKS
AND FERTILITY IN RWANDA

BY

Jane Eileen Vander Haar

The contributions that children make to the social and
economic well-being of their parents is commonly cited as an
explanation for why parents in Third World settings so often
seek to have large familiesw This paper provides an empirical
test of this assumption using random sample survey data from
1,019 farm households in Rwanda. The key determinants of the
total support received by households were the age of the head
of household, the number of children residing in and out of
the household, and the proximity of children in relation to
the household. This study confirms the economic value of
children, but goes a step farther by specifying how and to
whom this support is provided. In Rwanda, where wealth flows
in the direction of parents, where social security is almost
non-existent, and where upward mobility is limited, more

children may indeed be a sensible survival strategy.

Copyright by
JANE EILEEN VANDER HAAR
1990

To my parents, Del and Trudy Vander Hear
and my husband, Rowland D. Van Es, Jr.

iv

ACKNOWLEDGEMENTS

Special thanks to my committee for all of their assistance:
to David Wiley, my major advisor and committee chair, for his
guidance and counsel throughout my masters program as well as his
careful reading and insightful commentary; to Dan Clay for use of
his data, his helpful hints on statistical analysis, and.his useful

editorial comments; and to Jay Artis for his reading of my thesis
and his thought provoking questions.

TABLE OF CONTENTS

INTRODUCTION . . . . . . . . . . . . . . . . . . . .
THEORETICAL FRAMEWORK . . . . . . . . . . . . . . . .
Parents, Children and Old Age Security . . . . .
Family Size and Social Mobility . . . . . . . .
Wealth Flows and Fertility . . . . . . . . . . .
ANALYTICAL OBJECTIVES AND GUIDING HYPOTHESES . . . .
BACKGROUND ON RWANDA . . . . . . . . . . . . . . . .
DATA AND METHODS . . . . . . . . . . . . . . . . . .
RESULTS AND DISCUSSION . . . . . . . . . . . . . . .
Participation in Family Support Systems . . . .
Support Provided by Children . . . . . . . . . .
Characteristics of Children Providing Support .
Characteristics of Households Receiving Support
Total Support Received by Households . . . . . .
Multi-variate Model of Intergenerational Support
SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . .

BIBLIOGRAPHY O O O O O O O O O O O O O O O O O O O 0

vi

15

16

19

26

30

32

33

34

39

41

48

54

58

TABLE 1:

TABLE 2:

TABLE 3:

TABLE 4:

LIST OF TABLES

FREQUENCY OF SUPPORT FROM CHILDREN . . . .
CHARACTERISTICS OF CHILDREN AND SUPPORT. .
CHARACTERISTICS OF HOUSEHOLDS AND SUPPORT.

CORRELATION MATRIX OF KEY STUDY VARIABLES.

vii

33

35

40

43

LIST OF FIGURES

FIGURE 1: MULTI-VARIATE MODEL OF SUPPORT. . . . . . . . 49

viii

INTRODUCTION

The contributions that children make to the social and
economic well-being of their parents is commonly cited as an
explanation for why parents in Third World settings so often
seek to have large families. Indeed, the premise that
children have high socioeconomic value to parents has become
fundamental to at least three important avenues of
contemporary fertility theory and research.

The first is the theory that high fertility is contingent
upon parents’ needs for social and economic security in their
old age (DeVos, 1985; Entwisle and Winegarden, 1984; Hohm,
1975; Mouton, 1975; Nugent, 1985; Wildasin, 1980; and.Willis,
1980). A second theory has developed around the idea that
poor families can pull themselves out of poverty and achieve
upward social mobility through the economic returns generated
by a large number of working-aged children (Cain, 1982, 1983,
and 1985; Greenhalgh, 1985 and 1988; Johnson, 1984; Stark and
Lucas, 1988; and Stokes and Schutjer, 1984). More recently,
the focus of attention has turned to the theory of "wealth
flows," the proposition that fertility decline can occur only
after children are seen by their parents as a socioeconomic

liability, rather than as contributors to household earnings

2
and welfare (See Caldwell, 1976, 1977, and 1982; Handwerker,
1986a and 1986b; and Willis, 1982).

Though integral to these as well as to other fertility
research themes, the socioeconomic value of children is a
notion that has been largely taken for granted, receiving
little empirical examination by demographers and others. In
short, it is one of those "anthropological assumptions" that
demographers have widely used and for which.they have not been
held empirically accountable (Caldwell et al., 1987).

Because demographers studying Third World fertility tend
to rely heavily on data gathered through censuses and
large-scale surveys, such as the World Fertility Survey, it
is not surprising that they are so often unable to explore
patterns of behavior that require the kinds of in-depth
information that such survey instruments cannot easily
capture. Specifically, Caldwell (1985) observes that
large-scale surveys are flawed by their "inability to deal
with the family, the kinship network, the neighbors, the
community, and usually the couple" (p. 48). This is often the
case because limited time and funding restrict the scope of
these surveys to»one respondent.per household, usually'a woman
of child-bearing age.

This paper provides an empirical test of one important
dimension of the broad assumption that children contribute to
the social and economic well-being of their parents. The

focus of this study is farm households in Rwanda, where

3
fertility rates are among the highest in Africa, and on the
levels of economic support that Rwandan children give their
parents once they leave the parental household. Limitations
of the data do not permit an analysis of support provided by
children still living with their parents. Of special interest
are variations in intergenerational support as a function of
the number of children living within and outside of the
household, the distance between households, and individual

characteristics of parents and their children.

THEORETICAL FRAMEWORK

Parents, Children and Old Age Security

The Old Age Security theory of fertility suggests that
in Societies where the aged are dependent on their adult
children for support, fertility will remain high in order to
guarantee enough surviving children to meet this need.

Research in many different regions of the world.has shown
that parents who believe that children.provide old age support
do in fact have higher fertility. For example, in a study
conducted in southern Nigeria, Fapohunda and Todaro (1988)
found that parents who expected to receive old age support
from their children were less likely to be "demographic
innovators" and to have smaller families. Similarly, DeVos
(1985) found that in the Philippines and Taiwan, parents with
high expectations for future support exhibited higher
fertility preferences than did parents who expected little old
age support from their children.

While these studies provide some support for the old age
security hypothesis, they do not elaborate on the specific
circumstances under which children can be an important source
of old age support. Nugent (1985) has identified several of
factors which increase the importance of children including:
1) undeveloped capital markets, 2) parental uncertainty about
the accumulation of assets necessary for old age, 3) the

absence or inefficiency of public or private old age and

5

disability programs, 4) the confidence parents have in the
loyalty of their children, 5) the absence of well-developed
labor ‘markets for 'women, andh children, 6) ‘underdeveloped
markets for the goods and services that elderly' people
consume, 7) the absence of a spouse who is much younger and
could care for the person in old age, and 8) the perception
of old age as a considerable portion of the life cycle (p.
76) 0

Throughout much of Africa, especially in the poorest
countries, growth is restrained by a number factors including
undeveloped infrastructures, markets, and financial
institutions. Goldthorpe (1984) points out that the capital
markets and investment opportunities that do exist in
developing countries are limited to the formal sector.

In third world countries, then, many needs are met

(however inadequately) in the traditional subsistence

sector without the intervention of money. By contrast,

the modern cash economy is represented by such
institutions as mines, factories, banks, and commercial
offices, pervaded by a thoroughly business outlook and
the fullest use of money, credit, and international

exchange (p. 82).

The majority of the elderly in rural areas then do not
have access to capital markets and savings institutions. In
addition, Goldthorpe indicates that much of the labor
performed in the traditional subsistence sector is not
included in the cash economy. Since most of this labor is

performed by women and children, the market for their labor

remains undeveloped.

6

Furthermore, Treas and Logue (1986) note that despite
the fact that life expectancies at birth are much lower in
developing countries, once an individual reaches sixty years
of age, he/she will live nearly as many additional years in
either a developed or developing country. It is perhaps
reasonable to assume that even in the developing world old
age is perceived as a lengthy period in the life cycle.

In addition, public and private old age security programs
are inadequate. While a few African nations have some form
of social welfare program for the aged, they generally have
limited budgets and provide incomplete coverage. The average
annual expenditure on all public and private social security
benefits for ten African countries was only 2.1% of GDP.1
Furthermore, old age security programs represent only a small
portion of this total expenditure. (Wildasin, 1980) In many
countries these programs only cover those working in the
formal sector. Large segments of the population working in
subsistence agriculture or the informal sector have no old.age
benefits. Mouton (1975, p. 9) states,

The system of protection has in fact been extended

for the benefit of limited classes of the

population, and this is one of the most

characteristic features of African social security.

In the absence of state and private sector programs, the

family remains an extremely important form of support for the

aged. In evaluating social welfare for the aged in Ethiopia,

 

1 These countries included: Ethiopia, Libya, Malawi, Morocco,
Mauritius, Senegal, Sierra Leone, Togo, Tunisia, and Zambia.

7
Schenk and Schenk (1987, p. 14) conclude, "The welfare of the
aged in Ethiopia is considered to be primarily a family
responsibility." Similar conclusions are drawn by Bakayoko
and Ehouman (1987) in the Ivory Coast, Gethaiga and Williams
(1987) in Kenya, and Sanda (1987) in Nigeria.

All of these factors indicate that the old age security
motive for bearing children is likely to be quite strong. It
is less evident, however, that parents have confidence in the
loyalty of their children. Nugent (1985) has identified
several factors that affect the commitment children feel
toward their parents, a precondition for the actual provision
of support.

First, Nugent suggests that a close conjugal relationship
between husband and wife has a negative effect on loyalty to
parents. In addition, he argues that formal education and the
influence of mass media will also negatively affect loyalty.
Furthermore, identifying landholding as an important factor
in the loyalty of children, he asserts that the more
agricultural land the family operates and, therefore, the more
directly parents are seen as sources of employment
opportunities, the more likely it is that.children will remain
loyal. Similarly, he argues that the fewer the outside
employment opportunities available, the greater the importance
of the family holdings to a child's livelihood. This he
suggests also leads to a greater loyalty among children

(regardless of landholding). Finally, Nugent argues that the

8
more land and other property that is potentially inheritable
the more loyal children are likely to be. This factor is also
discussed in studies by Ben-Porath (1980) and Parsons (1984).

While much of the traditional family system remains,
especially in rural areas, changes are taking place. As more
children are educated, leave the farm, and move to urban
areas, the loyalty of children may decrease. An examination
of the support of the elderly in Africa shows that there is
not only an absence of other old age support mechanisms but
also that there may be a low return on investment in children
or at the very least an uncertain return” This is illustrated
by Sanda (1987, p. 173).

Old age in Nigeria is still considered to be

deserving of the care of the young and the

attention of government. However, the pressures of
urbanization, the demands of other responsibilities

of government, and the increasing disintegration of

the extended family, which previously provided

social security for the aged, all have made the

welfare of the aged very precarious.

Changes in the traditional family structure do indeed
place the elderly in a position of insecurity. However, just
because the loyalty of children is low, this does not
necessarily mean that the old age security motive for
fertility is low. In fact, a low rate of return on investment
in children may increase fertility in the absence of other
mechanisms for old age support (Nugent 1985). Where loyalty

is high, one or two children may meet old age security needs.

Where loyalty is low, an unlimited number of children may be

9
desired in order to meet minimal needs in old age. This
observation is also made by Cain (1983).

Others such.as'Vlassoff and Vlassoff (1980) conclude that
the returns from children in old age security are too low to
provide any real motivation for fertility. This conclusion
was based on a survey of Indian men and their sons. Treas and
Logue (1986), however, found that in less developed countries
the 'majority' of the elderly population is predominantly
female. Many of these women were widows with little
education, limited.experience in paid labor, and few resources
other than children. Nugent (1985) suggests that women have
a much higher need for children as a source of old age support
because they lack access to land and other income sources that
are available to men. In addition, much of the support given
to the elderly, such as the preparation of meals and care
during periods of illness, are traditionally tasks performed
by women. It is apparent then that Vlassoff and Vlassoff may
have obtained very different results if they had interviewed

women and their daughters or daughters in law.

Family Size and Social Mbbility

In addition to the old age security motive for bearing
children, a second theory, the social mobility motive for
fertility, rests on the assumption that children contribute
to the social and economic well-being of their parents.

According to this theory, children are seen as a means to

10

increase family wealth and social status. One way in which
children are seen to increase a families wealth is through
the labor they provide for family enterprises (generally
subsistence farming). It is also argued that in traditional
societies large families lead to both physical security and
political influence. In these ways having many children can
lead to upward social mobility. This argument is summarized
by Harrison.

Throughout these culture areas the extended family

is a kind of private corporation through which the

needs and ambitions of its members are fulfilled,

and usually, as in the business world, in

competition with other corporations. In the

traditional struggle for land and power, the bigger
corporations stand a better chance of winning. In

most of traditional Africa a man with three or four

sons is allocated more land.to work. (1984, p. 222).

The relationship between the mechanisms of land
distribution, the amount of land operated, and the economic
return of children's labor has been explored by the other
researchers as well. Two effects have been identified.
First, where land is communally owned, children who have not
migrated away increase the amount of land distributed to the
family which in turn increases the family wealth (Johnson,
1984). Second, as landholding increase, whether communally
or privately owned, the productivity of children increases.
Therefore, the value of children and child bearing are both

increased with increased landholding (Schutjer and Stokes,

1984; Johnson, 1984).

11

While there has been much debate over the causal
direction of this relationship between access to land and
fertility (Schutjer and Stokes, 1984; Cain, 1985; Stokes et
al., 1986), Cain (1978) has also argued that labor supplied
by children can provide greater income and an opportunity for
upward social mobility; Greenhalgh (1985) has found a
positive relationship between complexity of family structure
and income. This she attributes in part to the ability of
complex families to "diversify the family economy and.disperse
workers to new economic niches." (p. 575)

It has been suggested that an important way in which
families diversify is by preparing children to work in the
urban sector (Lucas and Stark, 1985). Urban migrants then
send remittances in cash and kind to their families in the
village. If these remittances are substantial, this can be
an important motivation to raise more children than can be
supported on the family farm.

Several studies explore the nature and importance of
remittances to rural areas. In a study of rural-urban
migration in Ghana, Caldwell (1969) found that one-third of
all households received cash remittances. Most remittances
were used to maintain the migrant's family in the rural area,
particularly parents. In addition, he found that the rural-
urban link is likely to become weak over time due to the
establishment.of permanent residence in the urban area and the

death of parents in the rural area.

12

Johnson and Whitelaw (1974) found that 89% of urban
migrants in Kenya regularly sent money back to their villages.
Furthermore, urban-rural remittances represented about 20% of
urban wages. They conclude that such a high percentage of
remittances implies a significant increase in rural welfare.
They also conclude that the number and closeness of relatives
working in the high wage urban sector significantly determines
a rural individual’s welfare.

In another study in Kenya, Knowles and Anker (1981)
focused on rural-rural remittances as well as rural-urban
remittances. They found that these remittances are usually
passed from husbands to wives, children to parents, and to
other close relatives such as brothers. In addition, they
found that urban-rural remittances account for slightly more
than half of all remittances. In contrast to Johnson and
Whitelaw, Knowles and Anker conclude that remittances have
little effect on the overall urban-rural wage differential.

Knowles and Anker also identify a cycle of remittances
from urban-rural migrants. When migrants first arrive in
urban areas, they receive support from their families in rural
areas. After they have found work in urban areas, migrants
become heavy remitters. Finally, once migrants become well-
established in an urban area, they are less likely to send

remittances.

13

Based on an in-depth study of remittances Stark (1981)
has developed a theory regarding the impact of rural-urban
migration on rural development. He states,

In developing societies, grown children as rural to

urban migrants may assume the unique role of

financial intermediaries in an economy in which the

introduction of modern agricultural methods is

constrained by inadequate institutional (as well as

non-institutional) sources of credit and a high

aversion to risk (p. 671).

Stark and Lucas (1988) further elaborate this theory.
They describe a cycle of remittances from the urban migrant
as well. Stark and Lucas suggest that this cycle is part of
a contractual agreement between the migrant and the family of
origin. This differs greatly from Caldwell (1969) and Knowles
and Anker (1981) who see this cycle as part of the process of
becoming a permanent resident in the urban destination.

According to Stark and Lucas, during the initial period
of migration, the migrant is insured by the family of origin
against the high degree of risk involved in urban migration.
At this point the migrant receives support from the family of
origin. Once the migrant becomes self supporting, the migrant
begins to send remittances to his/her family of origin. At
this point, Stark and Lucas suggest, the migrant acts as an
insurer against the high risk involved in the introduction of
change on the family farm. The fact that the migrant acts as

an insurer allows the family head to introduce improvements

on the farm that would.have been impossible without insurance.

14
In this way urban-rural remittances from children play an
important role in increasing the wealth of the family.

Cain (1978, 1982, 1983) has also emphasized the role of
children as a form of insurance against risk and in the
prevention of downward mobility in a variety of settings.
Cain (1982) focuses on factors outside of the family which
necessitate high fertility as an aversion to riskn He states,

Unlike child labor and old age security,
it is possible to see how the insurance
value of children could economically
justify large numbers of children in high
risk settings: the combination of
uncertain child survival, diverse sources
of risk, a premium on sons, and the
probability of great loss in their
absence could argue for such a
reproductive strategy (p. 167).

In addition, Cain (1983) suggests that laterally
extended kin networks can serve as a substitute for the
security from risk provided by children. In the absence of
these lateral support systems, an individual becomes more
dependent on lineal kin. Furthermore, he states that the
degree to which an individual is dependent on lineal kin is
affected by the kind of risk in the environment, the
effectiveness of children in averting risk and the

availability of other forms of insurance against risk (Cain,

1982, p.166).

15
Wealth Flows and Fertility

Wealth Flows theory is the third theory that rests on
the assumption that children contribute to the social and
economic well-being of their parents. This theory
incorporates and expands upon both the old age security
theory and the social mobility theory discussed above.
Research in this area has been initiated by Caldwell (1976,
1982) in an attempt to explain continued high fertility.
Caldwell argues that such high fertility is not adequately
explained by traditional demographic transition theories
because this theory assumes that high fertility is irrational
and economic development will automatically be followed by a
decrease in fertility.

In contrast, Caldwell stresses that all fertility
behavior is economically rational. He suggests that high
fertility is economically rational in societies where the
flow of wealth is from the younger to older generation.
Caldwell defines wealth flows as "all money, goods, services,
and guarantees that one generation provides for another"
(1976, p. 333).

Caldwell identifies six economic benefits from children
in traditional societies that demonstrate how high fertility
can increase wealth flows to the older generation. They
include: disproportionate gains from high fertility to
patriarchal males; child labor both in the household and on

the farm which provide goods and services to adults; gifts

16
and labor inputs of adult children; the proportion of the
family contribution to ceremonial occasions which is provided
by adult children; care of aged parents provided by adult
children; and the ability of parents to increase the return
on their investment in children through training or education
(1976, p. 343).

Fertility will continue to be economically rational as
long as intergenerational wealth flows are from the younger
to older generation. It is only when these intergenerational
wealth flows have been reversed and are from parent to child
that it becomes economically rational to limit fertility. In
fact, once wealth flows have reversed, any fertility is
economically irrational, but parents are willing to pay the
price of child rearing for the psychic benefits received from
children (Caldwell, 1982). In order for a reversal of
intergenerational wealth flows to occur it is necessary that
there be an emotional and economic nucleation of family
structure. This, argues Caldwell (1976), can occur
independently of economic development and is most likely to
come about. through the influence of mass Western style

education and mass media.

ANALYTICAL OBJECTIVES AND GUIDING HYPOTHESES
Underlying these theories of old age security, social
mobility, and wealth flows is the assumption that there is a

strong intergenerational support system being utilized by

17

households in developing countries. The old age security
motive for fertility lies on the assumption that children do
in fact provide assistance to their elderly parents. The
relationship between social mobility and fertility is
dependent on the assumption that children do provide labor
and gifts that can create upward mobility and prevent
downward mobility through insurance against risk. Caldwell's
very definition of wealth flows incorporates the concept of
intergenerational support. Despite the importance of this
underlying' assumption, little has actually' been. done to
empirically measure the degree to which intergenerational
support systems are utilized and factors which influence
participation in these intergenerational support systems.

Based on the research reviewed above, we can derive a
number of hypotheses regarding the level of intergenerational
support and factors that influence intergenerational support
in Rwanda. First, the logic of Wealth Flows Theory leads to
the hypothesis that the high level of fertility in Rwanda is
accompanied by an intergenerational support system in which
wealth flows from the younger to older generation. We might
also anticipate that if there is a weak lateral kinship
support system then there will be a relatively strong lineal
support system, given a lack of other forms of insurance
against risk in Rwanda.

Second, emerging from prior research on fertility and

social mobility it can be hypothesized that larger household,

18

being either more complex (Greenhalgh 1985) or having more
children (Cain 1978), will have greater wealth and higher
social status. One appropriate measure of wealth in Rwanda,
where most people are involved in subsistence agriculture, is
operational landholding. Therefore, we can expect that those
who are farming more land, having a greater income, will have
more children.

Finally, based on the theories regarding the old age
security motive of fertility, we can predict that older
household heads will receive more support. We can also
predict that households will provide support to their elderly
parents. Nugent (1985) identified a number of factors that
effect the loyalty of children. These include the amount of
land worked by the household and level of the children's
education. Therefore, we can hypothesize that the larger the
plot of land worked the more support a household will
receive. Also, the more formal education that a child has
the less likely they will be to provide support.

Other factors that we can expect will effect the support
received by a household include the number and location of
family members residing outside of the household. Therefore,
the more children living outside of the household, the more
support the household will receive (Caldwell 1969).
Likewise, it can be expected that children living near to
their household of origin will provide more support to their

parents than those living far away. Therefore, as the

19
average distance between 'the children and. the household
decreases, the level of support received will increase
(Caldwell 1969).

One final factor that may influence the frequency of
support received by the household is the gender of the head
of household. Likewise, the gender of the contributing child
may influence the frequency of support given to the

household.

BACKGROUND ON RWANDA
Rwanda is a small, landlocked country in the East
African highlands. It is bordered by Uganda to the North,
Burundi to the South, Tanzania to the East, and the Republic
of Zaire to the West. With an area of only 26,000 square
kilometers it is one of the smallest countries in Africa.
There are three ethnic groups in Rwanda, the largest of which
are Hutu (83%). The Tutsi are a large minority group, which
make up 16% of the population. Smallest in number are the
Twa who comprise little more than 1% of the population.
These ethnic groups share one cultural system and one
language, Kinyarwanda. The Courier (1982) states that Rwanda
is unique in that
It is a recognized territory, it has a
people, a language and a cultural
community and this means it has assets

which few African and not all European
countries enjoy.

20
allowed Hutu peasants to maintain cattle but not to own them.
These cattle were considered to be the property of the Tutsi
rulers.

In the late 1800’s Rwanda was colonized by Germany.
During the colonial period, the traditional system of Tutsi
rule was maintained. Rwanda was occupied by Belgian troops
during the First World War. After the war, Belgium was given
mandate over Rwanda and Burundi by the League of Nations.
Under Belgian rule, Tutsi supremacy was maintained and even
strengthened through a selective educational system which
provided greater opportunities for the Tutsi (The Courier,
1982).

In November of 1959, there was a Hutu uprising which led
to the fall of the Tutsi rule. At this time many Tutsi fled
to neighboring countries. The independence of Rwanda as a
sovereign and Democratic Republic was formally recognized on
July 1, 1962. Gregoire Kayibanda was appointed the first
President. He was re-elected in 1965 and 1969. In July of
1973, President Kayibanda was overthrown in a bloodless coup.
Former defence minister General Juvenal Habyarimana became
President.

Often called the "land of a thousand hills," Rwanda
ranges in elevation from about 1,000 to 2,000 meters. The
eastern part of Rwanda is made up of gentle slopes hills
while the western provinces are more mountainous. Because of

the higher elevation Rwanda is cooler than many other

21
countries at the same latitude. There is little seasonal
variation in temperature. The difference between daytime
highs and. nighttime lows are greater than the seasonal
variation in temperature. The mean temperature is around 66
degrees Fahrenheit, though this varies with elevation--the
east is warmer and the west somewhat cooler.

Throughout Rwanda there are two rainy seasons. The
first begins around March and lasts through May. The second
begins in October and lasts through December. The length of
the rainy season and the amount of rainfall as well as the
type of vegetation found in a region also vary with
elevation. For these reasons d'Hertefelt (1965) divided the
country into three homogeneous zones according to altitude:
the eastern region, where altitude varies from 3900 to 5100
feet; the central region, which is a plateau with an average
elevation of 5100 feet; and the northern and western regions,
with an altitude of 6000 feet or more.

The climate in Rwanda is unique in that variations in
elevation allow for the growth of crops that ‘would be
traditionally thought of as tropical such as coffee and
bananas as well as temperate crops such as peas and potatoes
(Nwafor, 1977). The range of crops grown also includes
beans, sorghum, millet, maize, sweet potatoes, rice, cassava,
groundnuts, wheat, and tea.

Despite such agricultural diversity, Rwanda is a poor

country. The GNP per capita is only $300 (World Bank, 1989).

22

There is little industry in Rwanda. Only 16% of the Gross
Domestic Product comes from manufacturing. Agricultural
products and minerals are the main exports. Coffee is the
largest export earner. Between 20,000 and 30,000 tons of
arabica coffee are exported each year. Minerals, such as
cassiterite and ferberite, are the second largest export
product. Tea is third with an annual export of nearly 10,000
tons yearly. Other major export products include pyrethrum
and tin ore (The Courier, 1982 and Africa Research Bulletin,
1985).

Subsistence agriculture is the basis of the Rwandan
people’ s livelihood, and Rwanda is a predominantly rural
country. Approximately 95% of the population are farmers,
and 87% of all cultivated land is planted with subsistence
food crops (Nwafor, 1977). More recent figures show that
Rwanda continues to be predominantly rural. In 1987, only 7%
of the population were living in urban areas (The World Bank,

1989). Those residing in the countryside do not live in

villages. Instead, each household lives in the center of
their landholding. These farming households are called
"urugo." There are about 880,000 thousand of these with an

average of five persons per household scattered across Rwanda
(Arid Lands Information Center, 1981). These households
generally consist of a nuclear family, and this forms the
basic unit of production and economic cooperation

(d’Hertefelt 1965).

23

The lack of centralized settlements has made it
expensive and difficult for the government to provide basic
services such as health care. The population per physician
is 34,680. This is more than twice the average for other low
income countries. The population per nurse is 3,650 (World
Bank, 1989). None-the-less, governmental health care
facilities are located within 25 kilometers of most people
(Arid Lands Information Center, 1981). Despite this effort,
the Infant Mortality Rate is 122 per 1,000. Life expectancy
at birth is only 49 years (World Bank, 1989). Nutritional
diseases, malaria, and trypanosomiasis (sleeping sickness)
pose serious threats to public health.

Another area of governmental expenditure is education.
Two-thirds of the primary school aged children are enrolled
in school. There is an equality of enrollment for boys and
girls at the primary school level. However, only 1% of the
secondary school aged girls are enrolled in school compared
to 3% for boys.

Rwanda is the most densely populated country in Africa.
With a total population of 6.4 million, it is estimated that
the population density is 245 persons per square kilometer.
However, due to the lack of urbanization and variations in
the arability of land, the actual density on agricultural
land is much higher. Some estimate that in the northern and

western regions, where there are rich volcanic soils,

24
population density is as high as 440 person per square
kilometer (Africa Research Bulletin, 1985).

Despite the scarcity of land the population continues to
grow at an alarming 3.7% (World Bank, 1989). Furthermore,
the growth rate is likely to remain high. Forty-three
percent of the women are of child bearing age (World Bank,
1989), and 56% of the population is under twenty years of age
(Nwafor, 1979).

In a poor country where virtually all of the arable land
is already under cultivation and the average household
possesses only 1.2 hectares of arable land (SESA, 1987,
p.34), a continued high rate of population growth can have
grave consequences.

One of the first concerns is with the ability of the
people to provide their basic food needs. At one point,
growth in agriculture was keeping pace with the growth in
population (Africa Research Bulletin, 1985). However, this
was due mostly to an increase in the amount of land under
cultivation. This type of expansion is no longer possible.
The most recent World Bank figures indicate that the average
annual growth rate in agriculture between 1980 and 1987 was
only 1.1%. In addition, in 1987 Rwanda imported 11,000
metric tons of cereal and received 16,000 metric tons of
cereal food aid. Finally, the daily calorie supply per
capita is 1,830. This is only 87% of the recommended daily

intake of 2100 calories.

25

Not only has production been unable to keep pace with
population growth, it has actually declined on a per capita
basis. Food production per capita between 1985 and 1987 had
declined by 13% as compared to similar figures for the period
between 1979 and 1981 (The World Bank, 1989). Much of this
decline has occurred in subsistence crops. For example, in
1976 227,572 tons of beans were produced. In 1979 only
180,660 tons of this staple crop were produced (The Courier
1982). This decline in subsistence crops is due in part to
a reduction in the size of family farm holdings. The size of
family holdings will continue to decrease as existing farms
are divided among the male children of existing household
heads. In Ginsenyi Prefecture, the average family farm
holdings is as small as 0.80 hectares, which is marginally
sufficient for subsistence (SESA, 1985, p. 69).

Despite the small size of family holdings, households
still try to plant a variety of crops in order to be as
diversified and self-sufficient in food production as
possible. This has led to overcropping in many areas. In
addition, each household tries to maintain some livestock,
which are still symbols of wealth and status. Consequently,
overgrazing also occurs on these small plots. If left
uncontrolled, overgrazing and overcropping will lead to the
second main concern caused by rapid population growth--

environmental degradation.

26

Rwanda’s fragile soils are quickly becoming depleted by
more intensive cropping and shorter fallow periods. The
Courier (1982) has reported, "The diminishing quantity and
quality of land is a major concern for the government." This
is likely to become an even more serious concern as
population densities increase. As more and more people
settle on the slopes of the many hills of Rwanda, erosion
also becomes a serious problem. It has been estimated that
as much as 35-40% of useful land may be lost due to the
gullying of the slopes (The Courier 1982).

In addition, the increasing demand for wood for fuel and
construction is causing rapid deforestation. A firewood
deficit of 6.0 million cubic meters by the year 2000 has been
predicted (Arid Lands Information Center, 1981). This

depletion of wooded areas intensifies the problem of erosion.

DATA AND METHODS

The Non-farm Strategies Survey was implemented by
Rwanda's Agricultural Surveys and Statistics Service (SESA)
as one component of the Agricultural Surveys and Policy
Analysis Project (ASPAP). This project is jointly funded by
the Ministry of Agriculture, Livestock and Forests and the
U.S. Agency for International Development. Survey
questionnaires were administered over a three-month period
beginning in July, 1988, to a stratified random sample of

1,078 farm households in Rwanda. An experienced team of

27
SESA's field staff supervisors was engaged to carry out the
interviews.

The questionnaires were designed to obtain information
from various members of the households, including husbands,
wives and adult children. A variety of topics were addressed
in the questionnaire. In addition to family support networks
information was gathered on: demographic characteristics of
all household members and migrant children; non-farm and
off-farm employment of all household members; permanent and
temporary migration patterns of selected household members:
fertility/family size behaviors, plans and preferences of all
adult household members; sources of household income;
physical characteristics of the farm and residence; hired
farm labor; plans, aspirations, and opinions of parents and
adult children regarding non-farm training and employment and
the future for young people in farming.

In all, interviews required an average of approximately
one and a half hours to administer and usually required
multiple visits to meet the various respondents from each
household. Adult children were the most difficult to locate,
though the interviewers did succeed in interviewing a large
number of these young people. In addition to the data on
adult children within the household, information was gathered
on 1,480 children living’ away from ‘their households of

origin.

28

An unusual, yet important feature of this study is that
the households selected for study are actually a sample of
households that SESA had been studying since 1986 as a part
of its ongoing survey program. Consequently, the great
volume of information already collected on these households,
in areas such as farm size and fragmentation, crop and
livestock production, and market transactions, can be drawn
upon to enhance the analyses of non-farm strategies. Since
1986, a small number of households in the initial sample have
moved away or have been otherwise dropped from SESA's current
sample. The total number of households on which data are
available at all points in time is 1,019; these are the
households for which data are presented in the following
section. The results reported are appropriately weighted to
adjust for differential sampling fractions used in the
stratified sample design.

Questions in the Non-farm Strategies Survey referring to
family support networks were addressed to the head of
household. He/she was asked to declare the frequency of
support received from various family members living outside
of the household. In addition, he/she was asked to declare
the frequency of support going from the household to his/her
mother and/or father. Three categories of support were
recorded in all cases. These categories were support in

money, support in labor, and support in kind. Further

29
questions were asked about the location of the family members
who could possibly be involved in the family support network.

It is important to emphasize that the information
gathered on the support given to the household only included
those family members living away from the household. The
data do not include any measure of the labor, money, or gifts
in kind provided to the household by children or other family
members residing within the household. Furthermore, while
the data include information about support given by the
household to parents residing away from the household, it
does not include information about support given to children
or other family members residing away from the household.
The data are also limited to information about the frequency
of support and do not include any measure of the value of
support provided by family members residing away from the
household.

Unless specified otherwise, the figures reported in the
results section will be based on the number of households in
the appropriate sub-sample. For example, when referring to
the support received from children, the figures will be based
on only those households with children living away from the
household. Likewise, when referring to the overall frequency
of support given, all households were included except those
that did not have any of the appropriate family members from
which the measure was created. For example, if both the

mother and father of the household head resided within the

30
household or were no longer living, that household would not
be included in ‘the figures concerning support. given. to

parents.

RESULTS AND DISCUSSION

The presentation of findings focuses mainly on two
groups involved in support networks: the parental household
and the adult children of the head of household who have
moved away from the household. Looking first at the parental
household (N=1,019), we find the mean household size is 5.5
persons. Men head 76% of these households. The remaining
24% of households are headed by women, the majority of whom
are either widowed or part of a polygamous union but living
separately from their husband. Most marriages are
monogamous, but 18% are polygamous. The mean age of the head
of household is 47.3 years old. More than half of the heads
of household (56%) have no formal education. In general,
younger household heads have more education.2 The operational
landholding of sampled households is 1.20 hectares. The
landholding tend to increase up through age 59, peaking at
1.61 hectares for those between 50 and 59. After age 60 the
amount of land worked decreases slightly. The amount of land
worked also tends to be greater among more educated heads of

household, ranging from 1.11 hectares for those with no

 

’ All findings reported are significant at p g .01 unless
otherwise specified.

31

formal schooling to 1.74 hectares for those who have had
schooling beyond the primary level. The mean number of
surviving children of the head of household is 4.8. Finally,
the mean number of children of the head of household
increases with the amount of land worked. Those with less
than 0.5 hectares have a mean of 3.7 children while those
with more than 2.0 hectares have a mean of 5.8 children.

Turning to the characteristics of children living away
from the parental household (N=1,480), we find the mean age
of these children is 29.5. Fifty-four percent of the
children living away from the household are female. Forty-
three percent of the children have had no formal schooling,
while 34% have attended some primary school. An examination
of their residence patterns shows that two-thirds of the
children (67%) are living in the same commune as their
parents or neighboring commune. Only 7% of the children are
living in urban centers. More than half of the children see
their families a few times a year or more (58%): 21% of the
children have daily contact with their families. A full 80%
of the children are farmers, 4% are students, and 3% are
civil servants. The majority of the children are married

(74%).

32
Participation in Family Support Systems

On the whole, family support systems in Rwanda appear to
be fairly strong. Sixty percent of the households report
that they receive some form of support from family members
outside of the household. Seventy-three percent of the
households report that they give some form of assistance to
their parents living outside of the household. Furthermore,
there is a clear indication that the lineal
(intergenerational) support system is much stronger than the
lateral support system. While only 25% of the households
reported that they received some form of assistance from
their brothers or sisters, 69% receive some form of
assistance from their children living outside of the
household.

The direction of support is from the younger generation
to the older generation as hypothesized. The figures
reported above confirm that the number of households giving
assistance to their parents is high, and the number of
households receiving support from their children is also
high. By comparison, only 26% of the households report that
they receive some form of assistance from their parents.

These initial findings appear to confirm the assumption
that children contribute to the social and economic
well-being of their parents. Intergenerational support

systems are in fact utilized by a large percent of households

33
'and the direction of support is from the younger to older

generation.

Support Provided by Children

Among children living outside of the household, 68%
provided some form of support to their household of origin.
The importance of this support can be further explored by
examining the type of assistance provided by these children.
While the few children help out their parents with gifts in
kind or of money (18% and 11% respectively), nearly half
(42%) provide some support in labor; .Furthermore, nearly
one tenth of children living outside of the household provide
labor a few times a month or more. The frequency of support
for the various types of support is presented below in Table
1.

TABLE 1: FREQUENCY OF SUPPORT FROM CHILDREN

 

 

Eercenr_£royiding_Snnncrt_in
Frequency Menegl .Laher_______Eind
Never 88.? 57.7 81.6
Once a Year or less 3.4 7.2 4.7
Few Times a Year 6.2 27.1 10.3
Few Times a Month or More 1.7 8.0 3.3
Total 100.0% 100.0% 100.0%

(N=1457)

34

Again, these results appear to confirm the assumption
that children contribute to the social and economic well-being
of their parents. Although only two-thirds of the children
living outside of the household provide some form of support
and much. of this is given infrequently, this does not
necessarily conflict with the old age security motive of
fertility. As suggested by Nugent (1985) and Cain (1983), a
low rate of return on children may indeed increase fertility.
Moreover, the fact that some children can and do regularly
provide their parents with labor, even after they have left
the household, appears to confirm the assumption underlying

the theory of social mobility and fertility.

Characteristics of Children Providing Support

Whether or not children residing away from the household
provide support and also the type of support provided varies
with the characteristics of these children. Some of the key
characteristics examined include the gender, level of
education, marital status, principal occupation, and the
residential location of the child. For each. of these
characteristics we examine how support to parents in general
is effected. This is based on whether or not children
provided any one of the three types of assistance. Also
examined is how the provision of each type of support varies
with these characteristics. These findings are presented in

Table 2.

TABLE 2:

 

Characteristic
Gender
Male 15.8
Female 7.4
Eta .145
Significance .000
Education
No Formal Educ. 7.8
Some Primary 10.3
Complete Primary 14.2
Post Primary 31.5
Eta .187
Significance .000
Marital Status
Single 19.8
Married 10.5
Free Union 2.9
Other 1.6
Eta .129
Significance .000
Occupation
Farmer 8.0
Civil Servant 54.7
Student 0.0
Other 25.9
Eta .300
Significance .000
Location
Same/Neighboring
Commune 7.6
Rural Migrant 14.7
Urban Migrant 34.3
Outside Rwanda 15.5
Eta .172
Significance .000

35

36.3
47.4

.104
.000

45.7
41.3
46.6
13.6

.161
.000

12.9
49.6
42.5
13.6

.275
.000

.290
.000

14.7
29.4
5.9
5.4

.302
.000

15.0
21.3

.076
.004

19.6
14.5
25.5
13.2

.079
.027

U10)N\l
O...
UQNU’

.164
.000

19.0
26.3

0.8
14.4

.098
I 000

18.2
21.5
19.6

7.6

.052
.271

CHARACTERISTICS OF CHILDREN AND SUPPORT

(674)
(783)

(615)
(501)
(242)

(94)

(241)
(1095)
(84)
(37)

(1171)
(44)
(54)

(185)

(977)
(322)
(94)
(59)

36

Looking first at the gender of the children we find that
a slightly higher percentage of female children are providing
some form of support (59% of females vs. 51% of males). The
influence of gender on the provision of support differs from
one type of support to another. Male children were more than
twice as likely to provide support in money; 16% of males
provided support in money as compared to 7% of females. This
difference is likely to be a reflection of the fact that the
cash earnings of men are higher than those of women. Female
children, however, were more likely to provide support in both
labor and kind.

When examining overall support it does not appear that
the level of education significantly differs for children who
do and do not provide support. However, different behaviors
are found for children with different levels of education when
the provision of each type of support is examined
individually. The proportion giving money to their parents
significantly increases along with level of education. In
contrast, the percentage providing their parents with labor
is nearly the same for those who have no formal education,
those with some primary education, and those who have
completed primary school. At the same time, support in labor
drops off significantly for those who have received post-
primary schooling. Finally, the percentage of those providing
support in kind does not significantly differ by level of

education.

37

Rather than decreasing the frequency of support, an
increase in education appears to merely change the type of
support that is provided. This finding seems to contradict
Nugent's suggestion that education will diminish children’s
loyalty to their parents and Caldwell's theory that education
will reverse the direction of the wealth flows.

The provision of support significantly differs with
marital status. Married children are most likely to give to
their parents in one form or another, followed by those in
free unions (traditional marriages not recognized by
government certificates), and finally single children.
However, this pattern differs for the ‘various types of
support. The highest percentage of those providing support
in money are single. In contrast, married children are more
likely to provide their parents with gifts in labor and kind.

This suggests that the type of support provided may be
affected by the child's stage in the life cycle. Single
children, who have not established their own households, are
perhaps more likely to be working for wages and, therefore,
can more readily send cash remittances. Married children, on
the other hand, who are more likely to have established their
own farms, more readily provide support in labor and kind.

When examining the three most common occupations--
farmers, students, and. civil servants--a significant
difference in the provision of support is found. Civil

Servants are the most likely to provide some form of support

38

(68%). Farmers are also quite likely to provide some form of
support (58%). Only 11% of students.provide their parents with
supportn The effect of occupation on the provision.of support
also differs from one type of support to another. Only 8% of
farmers provide support in money, compared with 55% of the
civil servants. A slightly lower percentage of farmers (19%)
provide support in kind as compared to civil servants (26%).
However, 50% of farmers provide support in labor, while only
6% of the civil servants provide labor support. Virtually no
students provide support in money or kind, but 11% of students
provide some support in labor.

When examining support in general we see that those
living closest provide significantly more support than those
living farther away. Of those living in the same or
neighboring commune 58% provide some form of support, while
only 23% of those living outside of Rwanda provide support.
When the different types of support are examined, variations
in the effect of location are seen. The location of children
does not significantly affect the provision of support in
kind. It does, however, significantly affect the provision
of support in money and labor. Those living in urban areas
are most likely to provide money support (34%) as contrasted
with 14% of those who have moved to other rural areas, 16% of
those living outside of Rwanda (many in rural Uganda), and
only 8% of those living on neighboring farms. Those living

on neighboring farms and in other rural areas are more likely

39
to provide support in labor, with 15% and 29% respectively.
Only 6% of those living in urban areas and 5% of those living
outside of Rwanda provided labor support.

This variation in the type of support provided. by
children with different characteristics lends support to the
idea that a household can increase its wealth and social
status by having more children. For example, it can diversify
its income by training one child to become a civil servant,
who will send cash remittances, and another child to be a

farmer, who will provide support in labor.

Characteristics of Households Receiving Support

The receipt of support varies with the characteristics
of the household. Some of the key household characteristics
examined include the age, gender, and level of education of
the head of household. These findings are shown in Table 3.

The mean of age of those receiving some form of support
(59.4) is significantly higher than of those receiving no
support (50.1). This same pattern is found for each of the
various types of support, providing strong support for the
old age security hypothesis.

In general, there is not a strong relationship between
the gender of the head.of household.and.supportn Seventy-five
percent of female-headed households and 64% of male headed
received some form of support (Sig.=.03) . When looking at the

gender of the head of household and the various types of

TABLE 3:

CHARACTERISTICS OF HOUSEHOLDS AND SUPPORT

Characteristic

Mean Age of Household Head
Receiving No Support
Receiving Some Support

Eta
Significance

Gender of Household Head

(Percent Providing Support)
Male

Female

Eta
Significance

Education Level of

Head of Household

(Percent Providing Support)
No Formal Education
Some Primary

Complete Primary or More

Eta
Significance

.___Jhnrrru1§hnnzuflr____1H1.
Money Labar____Kind____

55.3
60.9

.188
.000

21.3
29.3

.077
.111

24.9
20.6
24.3

.035
.737

52.2
59.9

.311
.000

54.0
66.9

.109
.021

62.3
55.7
31.8

.169
.001

54.4 (368)
61.6 (121)
.267
.000
26.0 (326)
38.0 (163)
.133
.005
33.7 (344)

24.0 (100)
18.3 (44)
.122

.027

41

support, a significantly higher percentage of female-headed
households receive support in kind (38% as compared to 26%).
A higher percentage of female-headed households also appear
to receive support in labor (Sig.=.02). There is not,
however, a real difference between the percentage of female-
headed and male headed households that receive support in
money.

Overall, household heads with more education are less
likely to receive some form of support. A similar pattern is
seen for support in labor and kind. The percentage of heads
of households receiving support in both labor (Sig.=.00) and
kind (Sig.=.03) decrease as the head of household’s level of
education increases. IHowever, there is no relationship

between level of education and the receipt of money support.

Total Support Received by Households

The above findings explore how various characteristics
of both households and individuals effect whether or not
support is received (in the case of households) or given (in
the case of children). However, the question that remains
unanswered at ‘this stage asks. how ‘these: characteristics
interrelate and how they collectively account for variations
in intergenerational support.

In order to examine this question, an overall scale was
constructed on which households could be indexed according to

the total support received. The questionnaire in its original

42

form allowed the interviewers to specify the frequency of all
three types of support from various family members in the
following categories: 0) never, 1) once a year or less, 2) a
few times a year, 3) a few times a month, 4) a few times a
week, and 5) daily. To create the scale the original values
for each category were squared. Though frequent support was
given a higher scoring than infrequent support, the frequency
was not translated into a proportion based on frequency per
year (i.e. once a year=1/365 and daily=365/365). In other
words, the scale gave slightly more weight to infrequent
support. This not only provided a more linear fit but also
seemed justified in that daily support, while occurring more
frequently, is generally of lower economic value than some
other less frequently occurring forms of support.

Each child living outside of the household received a
score for each type of support (money, labor, and in-kind),
and.these scores were then summed.to obtain an overall measure
of support provided by each child. Total support received by
the parental household was simply a summation of the overall
support scores of all their children living outside of the
household.

Table 4 reports the zero-order correlations between the
total support received by the household and various
characteristics of the parents and their intergenerational

support networks.

43
TABLE 4:

CORRELATION MATRIX OF KEY STUDY VARIABLES

Support Education Children

Household from of House-

oe E ‘ 'V '. 1' q :1 11 9 1‘29
Age of Household

Head -.35* -.24*
Gender of Head

of Household .13* -.31*
Education of Level

Household Head -.09 1.00
Children Outside of

the Household .49* -.11*
Children Within the

Household -.32* .18
Average Distance

of Children -.13* .13*
Landholding of

Household -.01 .15*

*Significant at .01 or more.

Children
Within Outside
1°-.‘Q' 0. r‘._‘!.‘ 0.
-.45* .51*
-.29* .13*
.18* -.11*
-.39* 1.00
1.00 -.39*
.06 -.02
.28* -.04

44

First, total support is moderately correlated to the
average distance from the household of all children residing
outside of the household (r=-.14). As the average distance
increases support decreases. This is consistent with the
hypothesis and the findings for individual children.

Total support was also moderately correlated to the
gender of the head of household (r=.13), indicating that
female heads of households receive more frequent support.
Again, this is consistent with the earlier findings.

The total support received was very strongly correlated
to the number of children residing outside of the household
(r=.47). Although the number of children living outside of
the household appears to increase the frequency of support
received, perhaps the frequency of support received per child
would decrease as more and more children take up their share
of responsibility for their parents--in essence dividing some
hypothetical fixed cost of old age support. If this were the
case, once a sufficient number of children were born to
provide this fixed level of support, the cost of raising
additional children would far exceed any future return from
these children.

We find, however, that the per child contribution does
not significantly decrease as the number of children living
outside of the household increases. While limitations in the
data do not allow for an accurate estimate of the economic

value of the support received, it is clear that the more

45
children parents raise and send outside of the household, the
better off they will be.

Furthermore, the total support received was negatively
associated with the number of children residing within the
household (r=-.35). The more children living in the
household, the less frequent the support received by the
household.

Finally, the age of head. of household is strongly
correlated with the total support received as well (r=.40).
Older household heads receive more help from their children.
There are two possible ways, however, through which age could
affect the total support received. The first is a life cycle
effect--as the household moves through different stages of
life, the total support received may vary with the abilities
of the children to contribute and the changing needs of the
household. The second is a period or cohort effect--the
particular history that an entire age group has lived through
may affect the amount of support received by households in
that age group.

Furthermore, both the number of children residing in and
out of the household vary with the age of the head of
household (r=-.51 for children in the household and r=.50 for
children outside the household). In other words, younger
heads of households have more children living within the
household and older heads of households have more children

living outside of the household. Therefore, it is unclear to

46
what degree the relationship between support received and
number of children residing in and out of the household is a
function of the life cycle stage of the households The multi-
variate model presented in the following section will help to
sort out these direct and indirect effects.

At this point, the relationship between children outside
of the household, age of head of household, and the amount of
support received seem to indicate that there is an old-age
security motive for fertility in Rwanda. Clearly, more
support is received as the head of household ages. In fact,
the mean total support received by household heads between 60
and 69 is more than twice that received by household heads
between 40 and 49 (eta=.38).

We also find that households begin to become net
receivers of support as the head of household approaches fifty
years of age. At this stage, the household is receiving more
support from children living outside of the household than
they provide to the elderly parents of the head of household.

In addition, total support was associated with the level
of education attained by the head of household (r=-.11) but
was not significant at the .01 level. However, the mean total
support for households where the household head had no formal
education was nearly twice that of the households where the
household head had completed primary school or more (eta=-.13,
significance of .04). The number of children residing in

and out of the household also seem to be linked to the level

47

of education attained by the head of household. Education
level is negatively correlated to the number of children
outside the household (r=-.13) and positively correlated to
the number of children within the household (r=.21). Those
with more education have fewer children living outside of the
household and also have more children living within the
household. These correlations may, however, be influenced by
the relationship between age and level of education. Again,
this will be clarified in the multi-variate model below.

Furthermore, the level of education attained by the head
of household is also associated with how far his/her children
live away from the parental household (r=.14). The children
of more educated household heads on average live farther from
the household.than.children of less educated household.heads.

Another background variable, the amount of land worked,
which is not significantly related to total support, the
number of children residing outside of the household, or the
average distance of children from the household, is
significantly associated.with the number of children residing
within the household (r=.28). This appears to support our
hypothesis that larger households (having more children) have
more wealth--in this case wealth in land.

Surprisingly, the amount of land worked by the household
does not account for any significant variation in the level
of support received (r=-.01, not significant at .01).

Therefore, our hypothesis that larger plots of land will

48
induce children to be more loyal and, thereby, provide more

support to their parents was unconfirmed in this case.

Multi-variate Model of Intergenerational Support

In the bi-variate analysis above, a number of factors
were found to be directly correlated with the level of support
received by households. These factors included: the gender
and age of the head of household, the number of children
living outside of the household, the average distance of
children from the household, and the number of children
residing within the household. In addition, factors which
were found to be correlated with the number of children in and
out of the household and.the average distancerof children from
the household include: the age of the head of household, the
level of education of the head of household and the amount of
land worked.

While the bi-variate analysis suggests that children do
indeed contribute to the social and economic well-being of
their parents, the extent to which this support from children
can be accounted for by other factors such as the size of
landholding, the age and gender of the household head, or the
proximity of children to the parental household need to be

examined through a multi-variate model (See Figure 1).

49

60955:!— :.~ emu 26:20:90 nogclouou

”H gUHh

”Roam—am no H060: oucfl—ETHZEA

 

.28 no 8. a. uiuzaaa ..

   

of 55....
8.8.482 802.6

 

     

 
 

 

 

503.50 .5: h / 39146: «5 391.8: m,
t a 38 a .2: $2.; Ewe innumonlSZ «6 cl: Go A
8.3 . .33 5.38:3 (
Nm .
5656386:
2236: a.

31:609.

50

The model shows that the initially strong relationship
between age of head of household and frequency of support
(r=.40) is weakened when controlling for the other factors
(beta=.16). The effect of age on the number of children
residing outside of the household also diminishes slightly
but remains strong (beta=.32). Older household heads have
more children living outside of the household.

In addition, the relationship between the number of
children residing outside of the household and total support
received remains strong when controlling for the other factors
in the model (beta=.32). The more children there are residing
outside of the household. the 'more support. parents will
receive.

These findings indicate that while the age of the head
of household does have some direct effect on total support
received, much of the age effect is indirect, through the
number of children residing outside of the household and
elsewhere. The number of children residing outside of the
household is the most important determinant of the total
support received by the household.

The relationship between age and the number of children
residing within the household is also reduced but remains
strong (beta=-.35). Recall that younger heads of households
have more children living within the household. The age of

the head of household, therefore, continues to be an important

51
determinant of the number of children residing within the
household.

However, theme is still a direct relationship between
the number of children residing within the household and the
total support received (beta=-.14). In other words, all else
being equal, households with more children at home receive
less total support. There are two possible explanations for
this. First, children living outside of the parental
household may be less likely to contribute to the household
while other children, their brothers and sisters, are still
living at home and are able to help out on a daily basis.
Second, children who come from household with more children
still residing at home are likely to be younger and,
therefore, may be less able to provide assistance.

There continues to be a moderate relationship between
the amount of land worked and the number of children within
the household (beta=.22). The more land a household works
the more children will be living within the household. The
Social Mobility theory would lead us to conclude that the
relationship between land and the number of children is caused
by the opportunities for increased wealth, including wealth
in land, created by children. However, it is possible that
the direction of causation is quite the opposite, that is,
more land leads to more children (Cain, 1985).

The relationships between level of education attained by

the household head and both the number of children residing

52

in and out of the household drop to near zero. The level of
education attained by the household head does not account for
any variation in either the number of children within the
household or the number of children outside of the household.
It seems likely that age explains much of the original
relationship we observed, since age and education are fairly
strongly correlated (r=-.26). Older household heads not only
have less formal education. but also have more children
residing outside of the household. This explains the spurious
relationship observed earlier between education and the number
of children residing outside of the parental household. The
age of the head of household similarly explains the
relationship between education and the number of children
within the household.

The relationship between level of education and the
average distance of children from the parental household
remains much the same and in fact increases slightly when
controlling for the other factors in the model (beta=-.14).
Children.of more educated household heads tend to live farther
from the household. We can surmise that this may be due to
the fact that highly educated household heads are more likely
to educate their children and that more educated children
migrate farther from home in pursuit of their education or in
search of work in the urban sector.

Not surprisingly, the gender of the head of household

affects the level of education obtained (beta=.-28) as well

53

as the amount of land worked (beta=-.15). The relationship
between theigender of head of household and the amount of land
worked may account for the disappearance of the relationship
between education and the amount of land worked. Because
women have less access to both education and land, in the bi-
variate analysis it appears that those with more education
have more land. In actuality, when controlling for gender, we
find that education does not influence the amount of land
worked.

In addition, the gender of the head of household is also
related to the number of children residing within the
household (beta=-.18) . Therefore, female household heads have
fewer children within the household. This may be due to a
lower level of fertility in the absence of a regular male
partner, or to the tendency for such households to be headed
by older, widowed women whose children have long since left
home.

Finally, the relationship between the gender of the head
of household and total support drops to near zero and becomes
insignificant. Again, the original bi-variate relationship
may have been due to the tendency of female heads of
households to be older widows. Therefore, when controlling

for age, the relationship disappears.

SUMMARY AND CONCLUSIONS

This paper has explored the assumption that children
contribute to the social and economic well-being of their
parents. Our findings support this assumption, although the
focus of attention was only on the frequency of support
provided by Rwandan children once they leave the parental
household. In Rwanda, as well as in many other African
countries, where wealth flows in the direction of parents,
where social security is almost non-existent, and where upward
mobility is limited, more children may indeed be a sensible
survival strategy.

In short, this study confirms the economic value of
children, but goes a step farther by specifying how and to
whom this support is provided. Special attention has been
given to the importance of intergenerational support, and the
degree to which it varies in relation to the number of
children living within and outside of the household, the
distance between households, and individual characteristics
of parents and their children.

The frequency and type of support provided by children
is closely linked to the individual characteristics of the
children, and thereby provides support for the social mobility
hypothesis. By having more children, a household can
diversify its sources of income and increase its wealth and

social status.

54

55

The key determinants of total support given to the
parental household were found to be 1) the number of children
residing outside of the household--the more children outside
of the household the more support received, 2) the number of
children residing inside of the household--the more children
in the household the less support received, 3) the proximity
of children to the household--the greater the distance the
less support received, and 4) the age of the head of
household--older heads of households received more frequent
support.

Although we found that the age of the head of household
was directly related to the total support received, this was
only a weak relationship. A stronger relationship was found
between the number children outside the household and total
support received. Therefore, the number of children residing
outside of the household was found to be the most important
determinant of the level of support received by the household.
These findings indicate that while children do provide more
frequent support to older heads of households, old age in
itself does not guarantee increased support. Successfully
rearing more children and sending them outside of the
household increases total support regardless of the age of
head of household.

Furthermore, the amount of support received increases
proportionally with each additional child living outside of

the household. In other words, there does not appear to be

56
a diminishing return to support received from each child as
the number of children contributing increases. This finding
is limited, however, by the fact.that the data do not estimate
the economic value of the support provided.and.because support
from children residing within the household is not measured.

The fact that children not only provide old-age support
but also serve to increase the well-being of the household
and provide a means of social mobility prior to this stage in
the life cycle lends support to the Wealth Flows hypothesis.
Wealth Flows theory in essence combines the ideas of old-age
support. and social mobility' by' suggesting' that. children
provide a flow of wealth to their parents throughout their
lifetimes. However, no support was found for Caldwell's
hypothesis that education reverses the flow of wealth. Given
the limited number of children who attain higher levels of
education in Rwanda, it is quite possible that the influence
of mass Western-style education has not, as yet, taken hold
in this case.

Finally, the findings of the study are further limited
by the fact that support was not measured during hard times-
-such as those caused.by drought or war; The findings reflect
the provision of economic maintenance during normal daily life
and not during times of emergency. However, given the
importance of children's maintenance support, it is hard to
imagine that the same would not hold under more difficult

conditions.

57

Because this study has examined only one aspect of the
premise that children have high socioeconomic value to
parents, and has focused on one small, high-fertility East
African population during relatively untroubled times,
additional research in this area is still needed. Studies
linking fertility to intergenerational support networks in
other sociocultural settings will enable us to begin to build

upon some of the findings presented here.

BIBLIOGRAPHY

BIBLIOGRAPHY

Africa Research Bulletin. 1985. "Rwanda: Budget (1985)."

Africa_Resaargh_Limited. December 15. 1984-Januery 14.
1985.

Arid Lands Information Center. 1981. Environmental Profile of
Ryanaa. Tucson: University of Arizona.

Bakayoko, Adama and Sylbestre Ehouman. 1987. "Ivory Coast."
pp. 69- 99, in Social Walfana in Afriga, John Dixon,

editor, London: Croom Helm.

Ben-Porath, Yoram. 1980. "The F-connection: Families,
Friends, and Firms and the Organization of Exchange."

Population and Developnant Review. 6(1):1-30.

Cain, Mead. 1978. "The Household Life Cycle and Economic

Mobility in Rural Bangladesh." Population and Development;
ngyign, 4(3):421-438.

----------- 1982. "Perspectives on Family and Fertility in
Developing Countries. " Popnlati an Studi as, 36(2) :152-175 .
----------- 1983. "Fertility' as an .Adjustment. to .Risk."
PQpulat10n_and_0e_elepme_r_Eeyiew. 9(4): 688- 702-
----------- 1985. "On the Relationship Between Landholding and
FertilitY-” Penulation_S_udies. 39: 5-15.
Caldwell, John C. 1969. A 'ca ral- ' ' n: T e
em t ’ . Canberra: Australian National

University Press.

----------------- 1976. "Toward a Restatement of Demographic
Transition Theory." Eopnlation and Devalgpmant Reviaw,
2:321-366.

----------- (ed.) 1977. e ist c o ' 'l't .
Canberra: Australian National University Press.

----------------- 1982. Tnapzy pf Fentility Dealine. London:

Academic Press.

Caldwell, John C., Pat Caldwell, and Bruce Caldwell. 1987.
”Anthropology and Demography: The Mutual Reinforcement
of Speculation and. Research." gnznanta_Annn;onlpgy,
28(1):25-43.

58

59

Courier. 1982. "Rwanda: The Shadow of Malthus Over a Genuine
Hope of Economic Progress. " W, 72(March-April) .

d'Hertefelt, Marcel (ed.) 1965. Winn. Holt,

Rinehart and Winston, Inc.

DeVos, Susan. 1985. "An Old-Age Security Incentive for
Children in the Philippines and Taiwan." Epgnpmig

WW 33(4): 793- 814-
Dixon, John (ed.) 1987. W. London:

Croom Helm.

Entwisle, Barbara and C. R. Winegarden. 1984. "Fertility and
Pension Programs in LDC's: A Model of Mutual

Reinforcement . " WWW].
Qhanga, 32(2):331-354.

Fapohunda, Eleanor R. and Michael P. Todaro. 1988. "Family
Structure, Implicit Contracts, and the Demand for

Children in Southern Nigeria. " WW1;
nggigy, 14(4):571-594.

Gethaiga, Wacira We and Williams, Lorece P. 1987. "Kenya."
99- 100- 120. MW. John Dixon.

editor, London: Croom Helm.

Goldthorpe. J- E- 1984. WWW
Cambridge: Cambridge University Press.

Greenhalgh, Susan. 1985. "Is Inequality Demographically
Induced? The Family Life Cycle and the Distribution of

Income in.Taiwan." AmarigannAnnthpplggiat, 87, 571-594.

---------- . 1988. "Fertility as Mobility: Sinic Transitions."
WW. 14(4)=629-674-

Handwerker, W. Penn. 1986a. "The Modern Demographic
Transition: An Analysis of Subsistence Choices and

Reproductive Consequences." American__Anthronologist.
88. 400- -417.

-------------- (ed) 1986b. WWW

 

Thapry. Boulder, Colorado: Westview Press.

Harrison, Paul. 1984. WE- Middlesex:

Penguin Books.

Hohm, Charles F. 1975. "Social Security and Fertility: An
International Perspective." Dampgraphy, 12(4):629-644.

60

Johnson, G. E. and W. E. Whitelaw. 1974. "Urban-Rural Income
Transfers in Kenya: An Estimated Remittances Function."

W. 22: 473-479.

Johnson, Nan E. 1984. "Rural Development and the Value of
Children: Implications for Human Development." p. 172-
194. in 811W- Schutjer.
Wayne.A. and.C. Shannon Stokes (ed.). New'York.: Macmillan
Publishing Co.

Knowles, James C. and Richard Anker. 1981. "An Analysis of
Income Transfers in a Developing Country: The Case of

Kenya." W. 8:205-226.

KuntZ. Patricia S. 1985. WWW
Afziga1_Nnmpaz_§_Buanaa. University of Florida, Center

for African Studies.

Lucas, Robert E. B. and Oded Stark. 1985. "Motivations to
Remit: Evidence from Botswana." W1
Egpngmy, 93(5):901-918.

Mouton, Pierre. 1975. W. Geneva:

International Labour Office.

Nugent, Jeffery B. 1985. "The Old-Age Security Motive for

FertilitY-" WW. 11(1): 75-
97.

Nwafor, James C. 1977. "Constraints on Agricultural Planning
and Development in Rwanda: an Overview."Afn1gan

Enrironment, 2/3(4/1): 87- -96.

---------------- 1979. "Agricultural Land Use and Associated

Problems in Rwanda." W.
48: 58- -65.

Parsons, Donald 0. 1984. "On the Economics of

Intergenerational Control." EQDulation_and_nexelonment
Raging, 10(1):41-54.

Sanda, A. O. 1987. "Nigeria." pp. 164-183, in_Sgaia1_Walfiaza
1n_Afriga, John Dixon, editor, London: Croom Helm.

Schenk, Quentin F. and Emmy Lou Schenk. 1987. "Ethiopia." pp.
1- 21 in_Social_flalfare_in_Africa. John Dixon. editor.

London: Croom Helm.

Schutjer, Wayne A. and C. Shannon _Stokes (ed. ) 1984. Banal
. New York: Macmillan
Publishing Co.

61

Services Des Enquetes et Des Statistiques Agricoles (SESA),
1985. '

Besultats_de_1_enguete_Natienal_Agrieele1_Yel_11
Np1_1. Kigali: Ministere De L'agriculture, De L'evevage
et des Forets.

Services Des Enquetes et Des Statist iques Agricoles (SESA),
1987. c s e

. Kigali:
Ministere De L'agriculture, De L’evevage et des Forets.

I

Stark, Oded. 1981. "The Asset Demand for Children During

Agricultural Modernization." Eooulatien_and_0erelenment
Bayian, 7(4): 671- 675.

Stark, Oded and R. E. B. Lucas. 1988. "Migration, Remittances,

and the FamilY-" WWW.
36(3): 465-481.

Stokes, C. Shannon and Wayne A. Schutjer. 1984. "Access to
Land and Fertility in Developing Countries." p. 195- 215,
in Rural_he1elenment_and_fluman_fiertilitx Wayne A-

Schutjer' and. C. Shannon. Stokes, editors. New' York:
Macmillan Publishing Co.

Stokes, C. Shannon, Wayne A. Schutjer, and Rodolfo A. Bulatao.
1986. "Is the Relationship between Landholding and

Fertility Spurious? A Response to Cain."EQp111_at19_n
snndiga, 40: 305- 311.

Treas, Judith and Barbara Logue. 1986. "Economic Development
and the Older Population." '

Review, 12(4):645-673.

Vlassoff, M. and Carol Vlassoff.1980. "Old Age "Security and

the Utility of Children in Rural India."£9,p§11at_19n
Studies, 34(4): 487- -499.

Wildasin. David E- 1980. WWW-
Bloomington: Indiana University.

Willis, Robert J. 1980. "The Old Age Security Hypothesis and
Population Growth " p. 43- -69. in Demographie_8eha2ier1

WWW - Thomas
K. Burch, editor. Boulder, Colorado: Westview Press.

----------------- 1982. "The Direction of Intergenerational
Transfers and Demographic Transition: The Caldwell
Hypothesis Re-examined." p. 207- 234, in Ingpma
Distributien_and_the_Eamilx Yoram Ben-Porath (ed )

World Bank. 1989. Hp;1a_naya19pment_3epgrt. New York: Oxford
University Press.