Notes from Seminar I and II
of the Seminar Series:
The Determinants and Consequences
of Female Headed Households
Sponsored Jointly by the Population Council
and the International Center for Research on Women
The Population Council
The International Center for Research on Women
DRAFT NOTES FOR SEMINAR 1
Classifications of Female Headed Households: Implications and
Applications for National Statistics
December 12 and 13, 1988
The Population Council, New York, New York
On December 12 and 13, 1988, the Population Council and the International Center for Research
on Women held the first seminar of their joint series on "The Determinants and Consequences of
Female Headed Households". The seminar series, funded by the United Nations Fund for Population
Activities, is bringing together development professionals, interested bilateral and multilateral donor
agencies, and scholars from developing and industrialized countries to discuss and debate the issues
surrounding female headship and maintenance of families. It is expected that dialogue generated
from the seminars will frame the debate, providing direction in future research on female headed
or maintained households and defining policy directions.
The first seminar was developed with the collaboration of the United Nations Statistical Office.
This two day session included discussions of diverse definitions of headship, the usefulness of these
definitions for policy purposes, and included reflections on what is gained and lost by applying
various concepts of headship. Participants/presenters included representatives from the United
Nations Statistical Office; national census bureaus of Zimbabwe, Mexico, Ghana, the English
Speaking Caribbean, and Sri Lanka; and several U.S. universities.
Dr. William Seltzer, Director of the United Nations Statistical Office, provided opening comments
that set the stage for the presentations and discussions that followed. Dr. Seltzer outlined three
conflicting forces at work that cause many difficulties as we try to identify and utilize concepts
and classifications of female headed households for national statistics. These forces are:
1. diverse analytical and policy concerns;
2. complex definitional distinctions become lost in large scale data collection efforts;
3. the mechanics of collection and analysis impose a certain order not necessarily representative of
Given these difficulties, it is helpful to see problem resolution as encompassing the following three
1. the need to make better use of existing data;
2. the need to influence on-going data collection;
3. and the need to develop new concepts and methods, and to experiment with new classifications
of households and families.
George Zeidenstein, President of the Population Council, welcomed guests on behalf of the
Population Council. He urged attention to the neglected area of women's roles in development.
He cited ten years of valuable research on which too little has been built. He suggested that the
incidence of female headship or maintenance of families is likely growing because of unabated
adolescent fertility in some parts of the world, continuing differences in age at marriage in other
parts, and increasing rates of marital disruption, the nuclearization of the family, and the
intensifying internal and external migration on the part of both men and women.
He urged efforts be made to quantify and qualify women's experience in maintaining households
to encourage policy support for women in their search for livelihoods.
Dr. Y.C. Yu, Chief of the Demographic and Social Statistics Branch of the United Nations Statistical
Office spoke on the prevailing concepts of the household and the family recommended by the U.N.
for population census and household survey purposes.
He indicated the definition of household varies by country, as does adherence to U.N.
recommendations. For example, some countries use "family" instead of "household". Household,
itself, is subject to some seventy-seven different definitions.
The U.N. recommended that to identify members of a household or of a family, countries should
collect information on relationship to the head of the family or household. It was recognized,
however, that in applying the concept of household head, with its implications of authority and
economic responsibility, the true picture of household relationships may be obscured, particularly
with regard to female heads of households. As a response to this situation, the U.N. now
recommends that countries use either "head of household" or "reference person" to determine
relationships within the household.
Because there are no rules for determining who is the reference person, this term is not accepted
by many statistical offices and causes some confusion. Some countries simply automatically define
the reference person as the oldest person in the household. No uniform rule has been established.
The term "householder" also has been suggested, but there remains the problem of translation into
Because the incidence of female economic maintenance of families rises as age of females increases,
except at the very old ages, the attempts to make both the concept and the classification of female
heads more precise and illustrative is extremely important. A crude comparison of census data
suggests that the overall female headship rates are higher in European regions than in Southeastern
Asia although the rates are lower in southern Europe than those in northern Europe. The Caribbean
has very high rates of female headship, as does Norway. In Japan, the rate is very low. The female
headship phenomenon (as officially defined in censuses) is socially diverse both within and between
countries. The determinants of headship and its meaningfulness will also vary; the position of such
households may vary substantially between countries with similar rates of female headship and vice-
Points of Discussion
1. Some questions to keep in mind when considering these issues are: is the concept of a household
"head" useful? If so, how should it be defined--by who supports the household, or lives there, or
makes decisions? It is possible that there is no one such person. When data is collected using the
"head" concept, what is the best way to use the data in analysis?
2. What are the purposes/functions of the concept of "head of household"? If we see the household
as a decision making unit, then the implication is that this unit is particularly influenced by the
attitude of the head. This may or may not be true; declared headship is not definitive evidence of
authority in decision making.
3. The concept of headship may pre-empt our fuller gathering of information about other kinds
of decisions, because it stops us from asking the more specific questions of who makes the money,
and how much? Who makes the decisions about expenditure? Who makes what purchases?
4. On the other hand, it is unrealistic to expect that population and housing censuses, which typically
gather information from interviews lasting only ten to fifteen minutes, to gather detailed information
on decision making structures, income questions, etc. Such information may be explored in
household surveys, however.
5. The household model currently under fire may have fit once, and may still fit in some places
today. The model itself grew from the need to associate family characteristics to one person.
6. Countries may or may not follow the recommendations of the U.N., depending on whether they
have a preferred term. It is important to remember that censuses serve national needs first, and
therefore do not necessarily adapt to facilitate, for example, emerging social concepts or a desire for
the international comparability of national statistics.
7. The advantages of utilizing a term such as reference person, in the absence of other more suitable
terms, is that it backs off from implying more than we actually know about family or household
dynamics. In some Nordic countries, the reference person is always the oldest person, a procedure
that is automatic and objective.
Session 1: The Successful Effort to Eliminate the Concept of "Household Head" in the U.S. Census:
History and Rationale
Dr. Harriet Presser, Professor of Sociology and Director of the Center on Population, Gender, and
Social Inequality, University of Maryland, spoke about the successful effort to eliminate the concept
of household head in the U.S. Census.
In the 1970s, various groups of people in the U.S. were seriously questioning the use of the concept
of "household head". Social Scientists in Population Research was among the groups. At the time,
the "head" was defined by the Census as "the person who is regarded as the head by members of the
household." There was no further specification. It relied on people subjectively and intuitively
sensing the concept which may vary considerably among individuals. Thus, household head was the
most ambiguous variable in the census data collection; no other variable was left so open to the
whims of the respondents. In the best of circumstances this allowed census people to know who is
the self-defined head of a family, but it did not allow them to know what is meant by head, what
criteria is used to choose the head, or what the comparability might be among different censuses.
If a woman living with her husband was reported as a head, the husband was considered the head
for tabulation purposes. When asked to explain the phenomenon that headship seemed to identify
and why it was so important to retain, a census representative responded that it was a means of
identifying the authority structure of the family. The question then arose as to whether the census
should be in the business of assessing authority structures in the first place. (In a small census survey
of 940 married couples, 49 percent identified the men as the head, 11 percent refused to answer.
Moreover, there was also a problem of lack of accord between men and women. When both were
queried, only a 38 percent of husbands and wives both classified the man as the "head".)
Given the ambiguity of the headship concept, and disagreement over it even when defined in terms
of 'authority', the question was raised of what would be lost by dropping the concept? Could the
census still get at relationships within the household without determining a "head"? The answer was
yes. In fact, Canada had already dropped the headship concept from its census, without problems.
Comparability was not an issue because the concept was so fuzzy it did not provide the basis for
comparisons between different household types. It was pointed out that if the census wanted to
identify single parents or female supported households or one-wage families, then it should use
specific terminologies to describe these household types, rather than depend upon unclear "household
head" to convey nuanced definitional characteristics.
In addition to being an issue of statistical accuracy, headshipp" became a politically objectionable
concept. In a 1974 letter to the Census Director, Arthur Fleming, the Chair of the Commission of
Civil Rights, asked the census to eliminate the notion of household head, because the concept,
among other things: assumed a hierarchical structure among adults that doesn't allow for change;
carried a sex-discriminatory bias; implied that information about the household head was more
important than information about the other members; and prescribed, rather than described, social
relationships. The National Organization for Women threatened a lawsuit and the possibility of
urging mass non-compliance with the Census.
These efforts found support in the Congressional Sub-Committee on Population and the Census.
Thus, the United States Census officially moved away from the term household head in the 1980
Census. Though there were fears that a neutral concept also risked negative political feedback, no
repercussions were experienced after the change. Indeed in Presser's view, the use of reference
person has improved the quality of census information.
Points of Discussion
1. A wide variety of information must be collected by 150 or so questions in a census survey.
There are a range of government and policy constituents for this information. Thus, the information
to be collected is broadly negotiated. It is arguable that the analytic strategy that is the designation
of tables to be generated is more important than changing the basic concepts used to collect the
data, provided the census does not actually obscure any information. Thus, in recent years the
movement for change has been directed at expanding and refining the regular tabulations rather than
changing the fundamental concepts of the census.
2. Census surveys can by nature yield up small amounts of discrete information. In order to define
the sociology of the family or it economic support base, one must use smaller, more indepth
3. "Headship" is a social concept, and is valid in many societies (even if less relevant or objectionable
in the U.S.). Thus, the concept should be retained for social research but dropped from censuses.
4. Issues around headship have been discussed for decades now. While the census professionals
may understand the limitations of the head of household concept, policy makers almost invariably
impute authority, economic responsibility, or residential characteristics from the 'head of household'
terminology. The cautions suggested by researchers and census bureaus are often overlooked and
thus invalid implications may be drawn by those who do not look beyond basic tabulations.
5. We need to split the demographic, social, and economic concepts (now amalgamated into the
overburdened, ambiguous household headship concept) and to ask direct questions about these
aspects of people's experience or household operations. In this way, we know what we are measuring
and relationships to other variables.
Session 3: Alternative Approach to Using Census Data to Highlight Women's Household
Dr. M.V.S. Rao, the Technical Advisor to the NHSCP Statistical Office, presented an approach to
using census data to highlight women's household maintenance roles. This approach, though
admitting the biases present in census surveys and that these do not adequately describe women's
roles, is not designed to define or do away with headship, Rather, it is an experiment designed to
make better use of information already available in large scale survey data.
When lacking specific data on the economic contribution of various members to the household, one
method of obtaining information from national statistics on the extent to which economic support
is provided by female members is to classify households on the basis of the number of economically
active male and female members. Thus, households with more females economically active than
males may be assumed to be more dependent on females, and vice-versa. Households, where there
are an equal number of economically active males and females, may be considered to be jointly
There are several advantages to this approach. In the first instance, it identifies economic activity
as its central concern rather than declared headship. Secondly, it acknowledges multiple earners
within the household. Within Rao's approach, male and female headship are parallel concepts, and
it allows for the possibility of shared economic responsibility by including a "joint" headship
category. On the other hand, the fact that one member is more active than another is not accounted
for; nor is the likely possibility that female earnings are lower than the earnings of males.
Nevertheless, this approach to analyzing available data does provide a starting point for looking
further into the questions surrounding household maintenance.
Points of Discussion
1. Because of the importance of children as unpaid family workers and as actual income contributors
to the household, they should be integrated into the above classification system. Domestic servants
as well might be included, as they affect the economic concept of the household.
Session 4: Comparative Data on Some Socio-Economic and Demographic Characteristics of
Households Mainly Dependent on Females as Against Those Mainly Dependent on Males
Using Dr. Rao's concept, Dr. Robert Tendere, the Deputy Director of the Central Statistical Office
of Zimbabwe, presented comparative data on some socio-economic and demographic characteristics
of households mainly dependent on females as against those mainly dependent on males. Using the
Zimbabwe Labour Force Survey of 1986/87, a series of tables were produced incorporating
information on persons aged 15 +. Mr. Tendere noted that because of information processing
constraints, it was necessary to limit the sample size to those over 15 years of age, which inhibited
the meaningful determination of dependency ratios in the groups to be studied. It was also noted
that since the basis of separation for households, i.e. economic activity, correlates positively with
many socio-economic characteristics, most of the results can be said to be predetermined by the
criterion for selection. For example, there is a high probability that the female dependent group
will have more females than males.
An analysis of the total number of persons aged 15+ by household type shows that 9.3 percent of
people enumerated in the survey live in households mainly dependent on females. The total number
of people in female dependent households may be underestimated, however, because of those persons
14 and under that in fact live with females but were not counted in this analysis. The female
dependent group has a sex composition of 74 percent females and 26 percent males. The percentage
of persons aged 15+ living in male dependent households is about 35 percent; the sex composition
of this group is 37 percent females and 63 percent males. It is interesting to note that 55.8 percent
live in jointly maintained households; this percentage might be different if 0-14 year olds were
Some common characteristics of women in female dependent households were: high rates of
employment for women in female dependent households (59 percent against 7 percent of women in
male dependent households); most married women in female dependent households are married to
men who do not reside in those households; female dependent households tend to have relatively
more women who marry late (possibly career women); 31 percent of males, and 13 percent of females
in female dependent households attend school, whereas 14 percent of males and 15 percent of women
in male dependent households attend school; there is a higher percentage of women in male
dependent households who have never been to school and remain illiterate (16 percent females as
against 7 percent males), possibly indicating that male dominated women are less likely to attend
adult literacy classes than those in female dependent households; there is a relatively high percentage
of females in professional and technical occupations within the female dependent households, which
is due to the inclusion of primary school teachers who are predominantly female (it should be borne
in mind that their income is much less than other professionals e.g. engineers and medical doctors);
women in female dependent households have higher employment rates in the service sector, trade,
and manufacturing than do women in male dependent households, whose economic activity is highly
concentrated in agriculture; an analysis of rural area females shows that while 51 percent of the
female dependent females are employed, only 4 percent of the male dependent females are employed;
there is a higher percentage of males "not in the labour force" in female dependent households than
in male dependent households (37 percent compared to 13 percent); in the urban sector, 61 percent
of females in female dependent homes are employed against only 8 percent in male dependent homes.
In the above analysis, the number of economically active women may be undercounted because
their activity is often in the informal sector and therefore not captured by labour force surveys.
On the other hand, transfers and remittances from urban men may mean that female headship is
overstated. Misclassification of economically active persons may also result from seasonal labor
patterns. There is little difference in the incidence of female headed households occurring in urban
versus rural areas.
Points of Discussion
1. If information is not collected to reveal a certain phenomenon, one cannot squeeze it out of the
data regardless of the recasting.
2. If we are interested in the burden of economic provision carried by the woman, it would be
useful to compare the welfare of households where there is not an economically active male with
those that are female maintained but where there is one or more male earner. To what extent does
female dependent mean households with no access to male labor?
3. The majority of households in the national sample were defined as being jointly dependent on
male and female earnings. Thus 56 percent of the households do not fall into "normal" headship
classifications when the possibility of a joint headship was raised.
4. Given the high levels of unemployment found in the Zimbabwe data, what do multiple earners
mean? Are these "earners" or "would-be earners"? These findings merit a closer look.
5. When adjusting the data to reflect economic activity and social realities not apparent in the
labour force survey findings, one must take into account resource and remittance flows from rural
to urban areas (not just from urban to rural), and the significant proportion of cases where a man
has two (or more) wives.
6. To get a picture of who is sharing income, we may have to relinquish the term household to the
extent that this suggests some high degree of co-residence. We also need to review the concept of
economically active, which excludes non-remunerated contributions such as child care, housework,
gathering fuel and water, etc. without which a household could not survive.
Session 5: Alternative Ways of Detecting Female Headship and Maintenance Roles in Mexico
Dr. Maria de la Paz Lopez, Coordinator of the Mexican Census, presented data on women headed
households drawn from Mexican Census data.
The Mexican Census found that women invariably named a man as the head of the household, unless
there was no man present. The solution proposed to this problem was to better instruct the
enumerators in gathering accurate information through a series of questions that would give a more
profound and accurate picture of who had responsibility of maintaining the household than is
possible from the simple declaration of a household head. Specifically, enumerators were instructed
to answer all questions, and not leave blanks in the survey questionnaires (a problem in past
censuses). The effort to improve the quality of information on the household head was part of an
effort to improve the census overall. One of the changes brought about to accomplish this goal is
the shift from using one and a half million volunteers in 1980 to using half a million trained and paid
enumerators in 1990.
Despite the biases outlined above against naming women as heads of households, there is still a
significant percentage (14 precent) of households headed by women. Some of the characteristics of
these household are as follows:
o eighty-nine percent of women headed households in Mexico had no spouse habitually
residing in the household;
o the average size of households headed by women is smaller than that of households headed
by men in both urban and rural areas;
o A higher proportion of households whose head is 50 years old or more is headed by women
than by men. Male headship rates increase rapidly after age 25 reaching a "peak" at ages
45-54, while for women an abrupt increase is observed after age 35;
o women heads of households have significantly higher labor force participation rates than
other women, at all ages.
Points of Discussion
1. There seems to be a strong correlation of poverty with declared female headship.
2. "Incomplete" households--households where no men are present--are more often in urban areas.
3. Once again, terminology obscures the household structure. For example, if we want to identify
households where there are no men, we should say "manless", rather than using a term such as
"incomplete", which seems denigrating.
Session 6: The Relevance of Diverse Headship Concepts in the EnElish Speaking Caribbean
Mr. Jack Harewood, the Regional Census Coordinator for 1990 Population Census of the English
Speaking Caribbean, spoke on the relevance of diverse headship concepts in that region.
In the Caribbean, a new approach to classifying households divides households into six different
categories: 1) women alone; 2) single family household with women, a partner, and children; 3)
multi-family, vertically extended; 4) multi-family, laterally extended; 5) multi-family household
with a mixture of vertical and lateral extension; and 6) other. A proposal that the census should
use these classifications resulted from a study in Barbados where 4 percent of households were
women living alone, and 11 percent were women living with children.
Mr. Harewood raised the question of whether the concept of households should be geared to
dependance statistics. A number of criteria could be used to determine the head of a household,
which might result in different people being named the "head". For example, the definition of a
head of household as a reference person would likely throw up a different person than a head
defined as the principle economic provider or the economic decision maker.
The very use of the concept of headshipp" is misleading, in that it provides a stagnant picture of a
what is, in reality, a fluid situation. Perhaps what we should be asking is: to whom does the
household look for protection, both in normal times and in a crisis? Who dominates within the
household? Who is listened to? These types of questions, however, and the answers they may
provide, have no place in census data; rather, the place to collect such data would be household
Studies examining fertility patterns in the Caribbean categorize women 15 years and older by their
union status: married, common law, or visiting. The census, however, cannot do such in-depth
evaluation or analysis of women's union status.
More time and attention is needed to define some of the issues surrounding headship than is available
before the 1990 Census. The term "reference person" will be recommended for the 1990 Census, but
in the meanwhile users will have to try to make better sense of the census.
Mr. Harewood also commented upon Mr. Rao's concept for classifying households by number of
economically active men and women. He stated that it was misleading to assume that all household
members classified as "economically active" will make an equal contribution to the household, because
of the problems of comparison between wages, number of hours worked, etc. Related is a question
about what types of activity are counted as "economic". Hours worked might be a more accurate
proxy, although there one runs into the problem that the poorest work the longest hours. A better
indication would be measuring individual income, which in itself is only a start because earning more
does not necessarily mean that one contributes more to the household. It is also necessary to factor
in lodgers who are often the members who contribute the most money to the household. Mr.
Harewood noted that although he felt Mr. Rao's classification system won't measure the extent to
which household's depend on the economic support of women in the Caribbean, he finds it useful
to talk of household levels of employment as opposed to individual employment or unemployment,
and to have a new approach to data gathering.
Session 7: Comparative Data on Some Socio-Economic and Demographic Characteristics of
Households Mainly Dependent on Females as Against Those Mainly Dependent on Males
Dr. R.B.M. Korale, Director of the Department of Census and Statistics in Sri Lanka, presented
comparative data on certain socio-economic characteristics of households mainly dependent on
females as against those mainly dependent on males.
The concept of the household and head of household used in Sri Lanka censuses and surveys have
evolved over time. The definitions used have been guided by the formulation of the concept by
the U.N. Statistical Office subject to national requirements. The changes in the concept have
affected enumeration but they have not been significant enough to result in a loss of comparability
Respondents' declarations of household headship are not necessarily based on the economic and
social responsibilities of the person identified. They could be based on seniority in age and legal
ownership of assets, irrespective of the supervisory and work responsibilities for their management
Although information on many of the characteristics required for the identification of female
headship of households have been collected in recent censuses and surveys, the data disaggregated
by sex and relevant variables required have not been processed and published.
It is necessary to obtain data on incomes and earnings from property and employment accruing to
females apart from information on their labour force status either as employed or operators of
agricultural or industrial assets. Data relating to incomes and earnings are not available from
censuses and this naturally limits this source for a complete identification of the nature of economic
contribution of the economically active females to their households.
The analysis of population data with the design suggested by Dr. Rao shows that approximately 10
percent of the population were from female supported households and as much as 42 percent of the
employed female population were supporting this 10 percent of the population.
Eleven-and-a-half percent of the total number of workers are located in female supported
households, but their occupational and industrial attachment suggests that their income earning
capacity would be on the average lower than that of the total population. Thus, in spite of the fact
that these households contain more economically active persons, their economic status is on average
below national levels.
The data from the Agricultural Census confirmed that female agriculturalists were operating on
smaller land parcels than male agriculturalists. The Survey of Economic Activities disclosed that
females were in receipt of substantially lower earnings than males in informal sector employment.
These economic circumstances would have limiting effects on the income situation and economic
well-being of female supported households.
The Rao concept of classifying households by the number of economically active males and females
was designed based upon the hypotheses that female headed households have certain special
problems, and therefore needed to be identified so that these problems could be examined. In that
sense, the Sri Lanka experiment using the classification system on available national statistics was
a success. Nevertheless, in order to get a fuller picture of men's and women's contributions to
households, we should seek out a description of their contribution to household types.
Points of Discussion
1. The central question for Sri Lanka (and one Korale suggested for other censuses) was to learn how
households as a unit cope and deploy their human resources. In the case of the Mexican data
presented by Maria de la Paz Lopez, Mr. Korale suggested that households with no men brought in
more workers--in this case, these workers were female, but the central strategy issue is to increase
the number of workers irrespective of their gender.
2. Korale also commented upon the biases in the Rao proposal in which the relative number of
economically active females, for example, statistically predisposes these households to having higher
economic activity rates, less access to male labor, etc.
3. The phenomenon of female headed households--however defined, including the Rao definition-
-exhibits regular patterns across countries and cultures in comparison to other households. For
example, these households are generally less well off, are often smaller, and their heads have lower
levels of education, and so forth.
Nancy Folbre. Discussant of Sessions 1 and 2
Dr. Nancy Folbre, Professor of Economics at the University of Massachusetts at Amherst, provided
a wrap-up discussion on the first day of presentations.
Dr. Folbre noted that a Census is a social and political process that is changed as it is approached
by its various planners, implementors, and users, and that in turn changes us. There is a
demonstrable history of lags in censuses; society changes faster than both the methods for recording
various aspects of society and the underlying philosophy of those measurements. An example of this
is the "family" which gave way as a unit of production to newly emerging forms far ahead of the
change in census measurement of "family" and household.
Another example of the lag in census measurements is found in the history of the "head of
household" measure in the U.S. Census, which underestimated female economic independence and
financial responsibility in the early and middle 1900s. As a result, when in the 1950s available
housing increased substantially, women who had been financially but not residentially independent
began to move to their own housing. The census recorded this as a sudden increase in female headed
households, thus overstating the rate of change in women's roles in maintaining families. In this case
the census, constrained by its inadequate definition of headship, failed to identify a trend that had
begun decades before.
It is clear that definitions of key census concepts change with changes in the social, economic, and
political milieu. In England in the 1840's, for example, "housewife" was considered an "occupation";
this classification of women's work evolved over the years, however, to "inactive", "unemployed", and
then the socially suggestive "dependent".
The ILO has conducted important work on redefining and restructuring the prevailing concepts of
economically active versus economically inactive, but it has not gone far enough. The current
classifications of primary and secondary activities, and the distinctions (often gender-bound) between
own account and unpaid family labor must give way to concepts which recognize the value of all
work, whether or not it is or could be traded in the marketplace.
Folbre also pointed out that male presence does not always mean wealth, nor absence of the male
mean poverty. She commented that the male presence in the household has labor distribution
implications as well as income and expenditure implications. Women in male headed households
are not necessarily better off than women in female headed households despite the evidence that
female headed households tend to be poor. There is some evidence that children of female headed
or maintained households receive the benefit of a larger share of income than those in other
She concluded that given the biases we know to exist in the socio-economic contexts of most
censuses, headship, while a socially valid concept, should not be applied in census or large scale
Session 8: Identifying the Poor: Is Headship a Useful Concept?
Sandra Rosenhouse, Associate, Population Council, spoke on the usefulness of headship as a concept
to identify the poor. She presented her study on female headed households which used data collected
from a national Peruvian sample between 1985 and 1986 by the World Bank's Living Standards Unit
and the Peruvian Instituto Nacional de Estadistica.
Policy makers concerned with the amelioration of poverty have singled out female headed households
as one of the key target groups deserving intensified attention. Studies have found that households
headed by women are over represented among the poor. However, research conducted to examine
the characteristics and well being of these households has employed a definition of headship
commonly used in surveys, i.e., the person other household members recognize as the head of
household. This term was originally introduced in surveys to avoid double counting of household
members in household rosters, and in no way reflects any of the dimensions the concept of headship
assumes: regular presence in the household, overriding authority, and primary economic support.
This paper examines the characteristics of households reported to be headed by women to show the
limited value of reported headship to reliably identify the economic support base of the household.
Rosenhouse's paper used an experimental indicator of headship which reflected work effort in
support of the household. The working head definition (the person who works the greatest number
of market oriented hours) was found to discriminate more clearly women supporting households in
extreme circumstances than the reported head definition because it highlights the elements that
contribute to the overall disadvantage of female heads: the lower return on their market hours of
work and the overall work burden they carry to attain a given level of consumption for the members
of their household.
When using the working head definition, the percentage of female headed households in this sample
is greater than that of the reported head definition (17 percent compared to 29 percent).
Rosenhouse's analysis suggests that the current definition of household in Peru excludes the
significant portion of households primarily maintained by women. In addition, she found that other
aspects of the household composition, such as the number of workers, may be important in detecting
groups at risk of poverty. Fifty percent of the households in this national sample have three or more
workers. It is interesting to note that in 80 percent of all households surveyed, the person reported
as head was the eldest person in the household, and in an additional 18 percent was the second oldest.
Only 68 percent of the reported heads were also the person who worked the most market oriented
Rosenhouse concluded by suggesting that if policies designed to combat poverty are to succeed,
intrahousehold dynamics need to be better described, i.e. multiple earners, patterns of authority,
fluidity of households, and the roles of different generations in household maintenance.
Points of Discussion
1. The age distribution of women in the working and reported heads categories showed that working
heads are a much more youthful population in Peru than in, for example, Mexico, where women tend
to assume headship in older age groups. These differences also show the importance of family
composition and life cycle. Women become heads when no one else is responsible--this is correlated
to poverty levels, the problem becoming more serious when there are no other earners and young
children in the household.
2. The life cycle question brings up the "chicken or the egg" scenario. Are households poor because
of their structure, or do they have a certain structure because they are poor?
3. Hours worked including market oriented and household maintenance might be considered a
drudgery index--but there remains a lack of clarity about the boundary between the two types of
Session 9: Analysis of Ivory Coast Living Standards Measurement Survey: The Childbearing Family
in Sub-Saharan Africa
Dr. Odile Frank, a Social Scientist with the Special Programme of Research, Development, and
Research Training in Human Reproduction, World Health Organization, presented her paper entitled
"The Childbearing Family in Sub-Saharan Africa: Structure, Fertility, and the Future."
Dr. Frank's paper focused on the structure and characteristics of the childbearing family in Africa,
their implications for fertility, fertility regulation and demographic trends, and their relevance to
Africa's future. Sources of data included a broadly scattered social science literature on Africa that
is demographically relevant and a primary analysis of the Living Standards and Measurements Survey
of the Cote d'Ivoire. Dr. Frank explained that the dominant childbearing family in Africa is not
nuclear. Typically, the African childbearing family is segmented, consisting of a husband and father
who is head but not necessarily breadwinner, and an economically autonomous wife and mother, each
of whom is more strongly affiliated by lineage than by conjugal bond. As a result, there is a
cleavage of the "nucleus," and norms for breadwinning and childbearing are separately reinforced.
Just as the husband and father in Africa is the head, but is not necessarily the major breadwinner
or the principal family support, the member of the family most often responsible for subsistence is
the wife and mother. This generally means that she is not only responsible for her own livelihood,
but that she has to be economically active enough to produce or purchase the necessary sustenance
for her family -- her children and often her husband.
Generally speaking, analysis of the Living Standards Measurement Survey for C6te d'Ivoire revealed
that important characteristics of women's lives, such as their sources of support and their degree of
economic responsibility or self-sufficiency, need to be documented on the basis of information
gathered from, and regarding individual women, since coresidence and dependant status rendered
women's economic situation within the household otherwise invisible. In sum, good research in the
African context requires that women be considered individually, and not as members of pooled,
Analysis of the C6te d'Ivoire survey could not uncover the essential economic autonomy of the
mother-children unit within the African household, since internal household structure was opaque.
Nevertheless, it served to confirm some micro level observations relevant to the economic and social
conditions of women particular to the region, such as:
oa very high participation of women in the labour force (especially in agriculture) that increases
rather than decreases with age;
o fairly regular attribution of household headship to males, very possibly on grounds
of social status and presence;
o relatively undisadvantaged self-reported women headed households that compare
favorably even with households headed by men;
o less land ownership in households headed by women than households headed by men
(controlling for the size of household) -- in association with greater participation in
nonagricultural sectors by women heads of households than by wives and women dependents
in men's households -- suggesting that women have greater access to land when living in a
Discussion: Beatrice Rogers
Dr. Beatrice Rogers, Associate Professor of Food Policy Economics at Tufts University School of
Nutrition, provided comments on the 2-day seminar.
Throughout the discussions, the "components" of headship--such as authority, economic
responsibility, number of hours worked, economic contribution, well-being contributions, etc.--
were used to talk about the phenomena. The uses of the term female headed household were also
discussed. The term has been used to identify poor households; to indicate the importance of
women's economic contribution; and as more specialized uses, such as the determinants of male and
female labor force participation, resource transfer changes as a result of economic changes, etc.
Thus, it seems that headship is only interesting if we want to know who a household recognizes as
head. She moved on to discuss how census tabulations can be used to overcome the obscuring effect
of the headship category.
What use can be made of census data to inform and describe the complex social and economic issues
surrounding family formation? Imaginative use of data on household composition and contributions
to the household, such as what are the labor resources available, and to what extent are they being
used to contribute financially to the household, could be used to answer important questions such
as how a household allocates is available resources in order to survive.
Large scale surveys are not likely to get good information on authority, decision-making, pooling
of income, etc. For instance, some surveys have shown that there is a 75 percent disagreement rate
between wives and husbands on who makes the major decisions of the household; clearly census data,
which questions one member of the family and records only that person's answers, is incapable of
discovering such a distinction. Small scale studies can be used, however, to supplement and
complement the information derived from censuses.
Finally, by forcing an inappropriate concept of headship, we obscure essential economic information,
lose valuable data, and produce misleading information.
Ms. Bruce, Senior Associate, Population Council, reviewed the conclusion of most discussants that
headship is a problematic concept and category for census use. However, if through inertia and
familiarity the term headship is maintained then, at the very least, the concept should be applied
in parallel fashion for men and women; i.e. if female headed means no male present then male
headed should mean no female present. We should expand headship categories to include the
possibility of jointly headed households extended laterally and vertically.
Using present techniques of data collection, headship rarely overstates the number of female headed
households -- though a number of widows supported by others may be included. By definition it
overstates the number of "male headed households" as these are usually defined to include both
households where there is an adult male with dependents and households where both an adult male
and female are present. Even where adult women partners of male "heads" contribute substantial
and even majority income and exercise considerable authority, the simple presence of an adult male
tends to mark a household as "male headed". Thus, for the moment female headship serves as a crude
indicator of the minimum number of households that may be economically dependent upon women
and absent a male earner.
Ms. Bruce then turned to discuss some of the other political and policy concerns which have
motivated the renewed discussion of female headship or maintenance of households.
She noted that most discussants had suggested that new cross tabulations and more sensitive social
analysis would yield a better picture of the relationship between family formation and poverty.
Yet, female headship has been a very useful label to attract policy attention. If gender disaggregated
analyses are not undertaken, yet female headship has become an obsolete term, we may lose a label
that while overbroad has served as a short identifier of the disadvantaged condition of a large,
possibly growing portion of individual women and the families they support.
A second issue of strategic importance to those concerned about women's maintenance of households
is how to apply different income concepts. While the 'full' income concept values women's domestic
and household production work, it may also overstate the wealth of such households. Poor women
who generate a respectable 'full income' often do so at a tremendous and unrecognized cost. Perhaps,
we need a full income concept which holds sleep and leisure constant to recognize (as Sandra
Rosenhouse's paper demonstrated) the punishing lengths women in such situations must go to to
ensure even a bare consumption level. As economies are modernizing, women without access to
wages will not be able to substitute -- through any amount of domestic labor -- certain vital
elements necessary for the household's health and wellbeing. Women's access to income and wages
is a measure of both their and their families' sustainable participation in modernizing economies.
Finally, she pointed out that the discussion of the differential contribution that women's income
makes to children returns us to an original source of concern about women (the effectiveness of
their mothering) but one we identify with a welfare instead of productivity approach. Ten or fifteen
years ago to talk about women in relation to their support of children may have suggested women's
economic passivity, marginality as economic actors, and so forth. In this new policy environment,
women's and children's welfare are closely tied. Yet, it is an important political decision to renew
attention to women as important economic actors explicitly because of their role in providing income
and other resources to children.