|
Defining and measuring poverty were
the main tasks of the AERC collaborative research project,
Poverty, Income Distribution and Labour Markets in Sub-Saharan
Africa, launched in 1997. The project embraced a set of
collaborative research activities carried out by joint and
interacting teams of researchers and resource persons drawn from
various networks of African and other institutions. The first
phase of this project (1997–2001) included two distinct but
closely interrelated components. The first component consisted of
research that was broadly analytical and empirical in nature but
region-wide in scope and focus. Twelve papers written by
distinguished scholars in this field provided the broad conceptual
framework and a state-of-the-art assessment of the various
dimensions of poverty in Africa today. The second research
component consisted of country case studies. Each study was
conducted by a team of African researchers, assisted in some cases by non-African researchers
from the collaborating institutions. These studies focused on
poverty assessment, profiles, and analyses of macroeconomic and sectoral policies to alleviate poverty within the context of
economic growth – emphasizing the money-metric dimensions of
poverty. Given the data intensive nature of the case studies, a
two-year time horizon was envisaged for their completion. Twelve
country case studies were commissioned and completed by 2001:
Benin, Cameroon, Ethiopia, Ghana, Kenya, Madagascar, Mozambique,
Nigeria, South Africa, Tanzania, Uganda and Zimbabwe.
Capacity Building
The Poverty Project incorporated several capacity building
mechanisms and activities. One of these is twinning visits by
individual country teams to one of the collaborating universities
(Cornell, Laval, Copenhagen and Gothenburg). Twinning visits gave
the opportunity to African teams to improve their skills during
the implementation of their research through closer contacts with
renowned scholars and institutions in the field of poverty
analysis. Furthermore, these institutions organized and conducted
training and research workshops and continue to be a mainspring of
Phase 2 of the Poverty Project, which is presently under way.
|
A major contribution of
Phase 1 of the Poverty Project to policy formulation was the
direct involvement of African researchers in the preparation
of poverty reduction strategy papers for their respective
governments and the World Bank.
|
An additional important capacity building activity at the
outset of Phase 1 consisted of training workshops in which
prominent economists presented the state of the art relating to
poverty analysis (including methodologies to explore income
distribution, and the operations of labour markets) for
participants from many African countries. The training workshops
covered conceptual and analytical issues as well as hands-on
computer work using data from African household surveys. Three
training workshops were held in Kampala, Capetown and Abidjan,
respectively, with the total participation of about 150 African
researchers and 30 resource persons.
Poverty Incidence and Measurement
The determination of the poverty line is the most critical
first step in poverty analysis. In the consumption based,
money-metric approach to poverty there are two methods to derive
the poverty line, the food energy intake (FEI) method and the cost
of basic needs (CBN) method. The CBN approach has the advantage of
ensuring consistency (treating individuals with the same
living standards equally), while the FEI approach has the
advantage of specificity by reflecting better the actual
food consumption behaviour of individuals around the caloric
threshold given their preferences and the relative prices of food.
If one national poverty line based
on one national reference basket is used, it may be inconsistent
with the actual, observed food bundle consumed by the poor in
different regions. Differences in tastes, diets and relative
prices lead often to significant variability in the food basket
consumed across regions. While imposing one national bundle
guarantees welfare consistency, it is likely to be inconsistent
with the diverse actual (regional) consumption patterns of the
poor. The analysis of the inherent conflict between the
consistency and specificity criteria conflict was at the heart of
the Mozambique Report, which analysed poverty on the basis of
regional poverty lines.
The Ethiopian and Kenyan reports, after having used both the
FEI and CBN methods in their analyses of poverty, argued that
these two approaches are not mutually exclusive and can be used in
a complementary way – the former to set the food poverty line and
the latter to arrive at the total poverty line.
Another issue here relates to the setting of the poverty line
to make comparisons over an extended period of time or even across
countries. The cost of a basket of goods satisfying food
requirements grows with GDP per capita for several reasons.
Examples include changes in the range of goods consumed as income
increases, rising prices of basic foodstuffs compared with prices
of other goods, increasing proportions of the population in urban
areas where foodstuffs may be more expensive than in rural areas,
and the gradual disappearance of subsistence farming. It can be
readily observed that basic needs expand with development –
particularly at an early stage of development. For all these
reasons it may be reasonable, over an extended time horizon, to
update and re-compute the basic needs basket and
|
raise the poverty line accordingly. A method to adjust the poverty
line as a function of mean income was developed within the scope
of this project.
There are essentially two very distinct ways of apprehending
and measuring poverty: a quantitative consumption-based money
metric approach and a qualitative approach, based on subjective
perceptions of individuals. Every country team relied on the
quantitative approach and one team (Ethiopia) also utilized the
subjective approach to estimate poverty, allowing thereby a
comparison to be made between the results of these two methods, as
will be seen shortly. Each team used the Foster–Greer–Thorbecke (FGT)
class of additively decomposable poverty measures and reported the
head-count ratio and the depth and severity of poverty. Estimates
of poverty incidence in the 12 country studies ranged from
one-third to two-thirds of their respective populations.
The Ethiopian research team followed a subjective approach to
poverty measurement to complement the money-metric approach. The
definition of subjective poverty is based on the perception of
individuals about their wellbeing and attempts to relate the
self-reported subjective welfare levels with observed income.
Surprisingly, the poverty incidence estimates obtained under the
quantitative, money-metric approach and under the subjective
approach were strikingly close.
A number of teams used stochastic
dominance and poverty ordering techniques to explore whether
poverty incidence had changed
over time and to determine whether certain sub-groups were
unambiguously poorer than others.
Poverty Profiles,
Determinants and Correlates
A poverty profile identifies the major household and
environmental characteristics that are associated with poverty.
The major correlates of poverty appearing in virtually all the
country profiles are: labour force and occupational status, race,
location (e.g., urban/rural, provincial or regional breakdown),
education and health, size of household, gender (e.g., male- vs.
female-headed household), age, dependency ratio, market
orientation, ownership of land and assets, and marital status.
One of the most important contributions of the South African
study was to show convincingly the close link between unemployment
and poverty. Employment of adult labour market participants is
shown to be the largest single contributor to household per capita
income and household poverty avoidance.
All the studies found that the incidence of poverty was
relatively greater in the rural areas than in urban areas. Indeed,
because most of the African population is rural, African poverty
is a rural phenomenon, despite the recent emergence of high urban
poverty incidence in some countries.
In general, the country reports find a strong negative
correlation between education and poverty. But there were some
unexpected results. In Ethiopia, returns to primary education –
particularly in the rural areas – are close to zero but high at
the secondary and tertiary levels. In some other African
countries, notably South Africa and Kenya, rates of return to
primary education have been estimated to be much lower than the
returns to secondary and tertiary education.
A universal finding was that
poverty per household member or per adult-equivalent increases
with the
size of the household. A key issue in this context is the
extent to which economies of scale within a family play a role.
Most studies explored the impact of the gender of the head of
the household on poverty incidence. Although male-headed
households were found to suffer from less poverty than
female-headed households, some countries (Cameroon, Mozambique,
Ghana) reported less poverty among female-headed households.
Generally, age of household head and dependency ratio are found to
be positively correlated with poverty.
Inequality and Income Distribution Issues
As befits one of the most unequal economies in the world, the
South African study undertook a thorough analysis of the extent of
inequality. With a Gini coefficient of around .60, South Africa is
only surpassed by Brazil (.63) in terms of income inequality.
Benin, Ethiopia and Uganda display high income inequality in urban
areas. One important lesson from the project is that measurement
errors are a constant source of concern in the analysis of poverty
and inequality. Researchers everywhere are at the mercy of
government statistics offices and other groups designing and
running the surveys and are often faced with a black box and the
need to be forewarned of the absolute necessity of scrutinizing
data used in research very critically.
Operation and Functioning of Labour Markets
Both the South African and Ethiopian studies contained an
extensive analysis of the labour markets. On the basis of a
combination of techniques, including estimated earnings and labour
force participation functions, the South African team reached the
following conclusions:
• Domestic workers and farm workers are the most vulnerable
groups in the labour market. Given the total dominance of Africans
(Blacks) and women within these groups, this labour market
vulnerability translates into a poverty vulnerability.
|
• Labour markets are strongly segmented, with large informal
markets affecting women and men differently.
• Spatial rigidities prevail, with urban work-seekers having
access to rural jobs on average, if they so desired, while rural
work-seekers typically do not possess the characteristics
necessary to compete in the urban labour market.
• Labour markets appear to be
stacked against new entrants, with the younger cohorts facing a
much lower probability of finding jobs than older cohorts.
Policy Implications
One of the few and most important instruments available to a
government of a poor country in fighting poverty consists of
public spending on social services. For example, in Mozambique
about one-fourth of the central government’s total spending was on
education and health services and about one-third on all social
services combined including education and health. Given the
relative magnitude of such expenditures, it is essential to
determine how well targeted those services are towards the poor
and how far they go in alleviating their plight.
For lower primary education the
distribution of benefits is almost equal – the poor share equally
in the benefits of lower primary education as the non-poor. On the
other hand, secondary education benefits are distributed even more
unequally than consumption. With regard to health services, the
team found that the participation of infants and pregnant women in
basic preventive health care is progressive (in the sense that the
poor receive a larger proportion of the benefits than their share
of total national income or consumption) and appears particularly
progressive for infant vaccination. Hence access to preventive
care helps to equalize the income distribution. Essentially, the
same finding applies to visits to curative health services.The
Madagascar report also analysed the benefit incidence of public
expenditures on education and health. Their findings were fairly
similar to those from Mozambique. The distribution of primary
education services was absolutely (per capita) progressive, while
the secondary education benefit appears slightly regressive (in
the sense that the poor receive a smaller proportion of the
benefits than their share of total national income or
consumption). Health expenditures on main and secondary hospitals
benefit the rich slightly more than the poor, while public
expenditures on other heath care centres are relatively
progressive.
Recommendations
Five broad elements of a pro-poor growth and development
strategy may be recommended for sub-Saharan African countries on
the basis of the research emanating from this collaborative
project:
• Eliminate the discrimination against rural and agricultural
development. Most of Africa is still in an early phase of
development – what can be referred to as the pre-takeoff phase.
A lesson learned from those countries that were most successful
in achieving both growth and equity throughout their development
history (e.g., East Asia) is that a continuous flow of resources
should be provided to agriculture in the form of such elements
as rural infrastructure, inputs, research and credit combined
with appropriate institutions to increase this sector’s
productivity and potential capacity for contributing an even
larger reverse flow to the rest of the economy. Instead, the
typical pattern in sub-Saharan Africa has been to exploit
agriculture and tax and starve it early on, forcing it into
stagnation and thereby "killing the goose before it had a chance
to lay a golden egg". In general, a development strategy that
promotes equitable growth should be designed and implemented as
such, and should not be expected to emerge on its own through
trickle-down effects.
• Invest in human capital (education and health, including
nutrition and sanitation) so as to improve human welfare
directly. Making these services broadly available should be a
key objective of development strategy.
• Improve intra-household allocation of consumption. Some
household members, particularly children, widows and the aged,
may be deprived of basic needs even in non-poor households.
Public policies, including safety nets, need to be designed to
reduce the vulnerability of these groups.
• Make markets and public infrastructure accessible. Access
to all markets (labour, business assets, land, social capital,
food, etc.) is associated with poverty reduction. Public
programmes to improve access to markets and to social services,
including security of property and persons, are important
components of a pro-poor development strategy.
Finally, a number of reports emphasized the critical need to
facilitate the creation of new jobs. Given the seriousness of
overt unemployment and under-employment in Africa, any growth
strategy should be centred on employment creation.
Erik Thorbecke is Graduate School Professor and
Professor of Economics Emeritus at Cornell University, USA.
Germano Mwabu is Professor of Economics at the University
of Nairobi. They are co-coordinators of the Poverty Project.
|
