POVERTY AND INEQUALITY:

ECONOMIC GROWTH IS BETTER THAN ITS REPUTATION

By Arild Angelsen ♣,♠ and Sven Wunder ♠

Chapter in Dan Banik (ed.): Poverty, Politics and Development: Interdisciplinary Perspectives.
Fagbokforlaget, Bergen, 2006.

INTRODUCTION 1
The debate on the relationship between poverty, inequality and economic growth is characterized
by confusion and strong, polarized positions. Some consider economic growth to be the key for
the reduction of poverty, while others argue that it tends to lead to marginalization and greater
inequality and poverty. These positions reflect two major historical stands in the discussion
about the causes of poverty. First, the ‘developmentalist position’, which explains poverty in
terms of lack of economic advancement, normally equated with insufficient economic growth.
Second, ‘class-based’ (and Marxist inspired) theories, which view poverty as a result of uneven
development and exploitation, resulting in skewed asset and income distribution. According to
the first view, the income poverty problem is solved by making the ‘cake’ (total income or Gross
Domestic Product – GDP) bigger, while the second argues the problem should be addressed by
giving the poor a bigger share of the cake.

These two views produce fundamentally different predictions on whether and how economic
growth and structural change can help reduce poverty. For example, the typical Marxist view
would hold that: ‘In countries at low levels of development, any kind of structural change such
as industrialization or expanded commercialisation tends to increase poverty among the poorest
members of the population’ (Adelman and Morris 1978: 256). However, much empirical work
over the last decade, facilitated to a great extent by better income and consumption data at
different levels of aggregation, has enlightened this debate. At present, there is a growing
consensus (e.g. Fields 2001; Ravallion 2001) on two sets of issues. First, macroeconomic
growth, in most cases, raises the income of the poor and reduces the number of people below the

♣
Department of Economics and Resource Management, Norwegian University of Life Sciences (UMB),
Ås, Norway.
♠
Center for International Forestry Research (CIFOR), Bogor, Indonesia.
Email: [email protected] & [email protected]
1
This chapter draws heavily on Angelsen and Wunder (2003), in particular the discussion on poverty
definitions in sections 2, on inequality and growth in section 3, and on pro-poor growth in section 4.

1
poverty line. Thus, growth ‘trickles down’ – at least in the medium and long term and at an
aggregated (national) scale. Second, in a minority of deviating cases, little or no poverty
reduction is achieved through growth because of a skewed initial asset distribution and/or ‘bad-
quality’ economic growth. This development pattern is characterized by low labour intensity,
low human-capital accumulation, rural neglect, and high levels of corruption.

This chapter first clarifies the definitions of the three key words in its title – poverty, inequality
and economic growth. In particular, the evolution in the concept of poverty is discussed followed
by a section on the mechanical and empirical links between income poverty, inequality and
growth. The concept of ‘pro-poor growth’ is introduced thereafter and defined as growth which
raises the income relatively more for the poor than the better-off groups, i.e., growth with
improvements in the income distribution. We argue that the prospects for pro-poor growth
depend both on the initial conditions as well as on the type of growth, e.g. in which sectors the
growth occurs and its labour intensity.

DEFINITIONS AND THE ELIMINATION OF UNNECESSARY DISAGREEMENTS

Economic growth

Economic growth is generally defined as an increase in Gross Domestic Product (GDP), either in
total GDP or in GDP per capita. With high annual population growth rates of 2-3 % a country
may easily be in a situation with growth in total GDP but not in per capita GDP, and we shall
therefore refer to economic growth as an increase in GDP per capita. The GDP is nothing but the
value of total production or total income for a country. Thus, economic growth is the same as an
increase in average income. The growth definition is the first potential source of disagreement in
the growth-poverty debate. To an economist it is a technical concept, as defined above. To others
a statement like: ‘we have to promote economic growth’ is understood as a suggestion to
promote neo-liberal policies like the ones suggested by the IMF and the World Bank. Thus the
term is often interpreted beyond its technical meaning. This chapter will use the term economic
growth in its pure technical meaning – as an increase in average income.

Inequality

A discussion of the philosophical meaning of term inequality (and inequity) is beyond the scope
of this chapter. Use in economic analyses of growth, inequality and poverty, the term refers to
end results on welfare, and not in material and intangible assets that may contribute to the

2
explanation of these end results. . The most common measure of income inequality is the Gini
coefficient or index (G), named after the Italian statistician Corrado Gini (1912). The Gini
coefficient has a value between 0 and 1, with 0 being perfect equality (all have the same income)
and 1 being perfect inequality (all income earned by one person). In most countries, it ranges
between 0.3 and 0.7. The Gini coefficient can thus be intuitively interpreted as the share of the
total income (GDP) that has to be redistributed to hypothetically obtain perfect income equality.
For example, a country with a relatively high inequality and G = 0.6 must take an equivalent of
60 % of its GDP from the rich and give to the poor to make all have the same income.

Another commonly used measure of inequality is the Kuznets ratio. This gives the ratio between
the average income of the richest and the average income of the poorest – typically undertaken
by focusing on the averages of the top and bottom quintiles, i.e., the richest 20 % and the poorest
20 %. This has a clear intuitive meaning: how many times richer are the rich compared with the
poor? The Kuznets ratio typically varies from about 5 for egalitarian European countries to more
than 30 in some Latin American countries. Although popular, this measure is from a scientific
viewpoint less satisfactory as compared to the Gini coefficient, because income changes in the
middle range are ignored (e.g. a transfer of income within the 60 % in the middle would not
affect the Kuznets ratio, but would change the Gini index).

The changing concept of poverty

By far the most complex concept is that of poverty. Concepts at the centre of international
debates often face pressures for a broadened interpretation. While adjustment of concepts over
time may be desirable, they also generally tend to become increasingly all-embracing and
inclusive and therefore face a real risk of overload and ‘concept degradation’. Certainly,
’poverty’ is no exception.

Traditional definitions of poverty have focused on income and wealth, or the lack of money or
material possessions. The definition of poverty as lack of income, inherited from classical
economists like Adam Smith and David Ricardo, was completely dominant until the 1960s,
when the focus of development policy was more or less exclusively on expanding monetary
income. In recent decades, however, it has become increasingly popular to extend the definition
of poverty to other, non-material aspects of human well-being. Indeed, the evolution of the
concept of poverty reflects changes in development theories and practices in general, and the

3
analysis of the causes of poverty in particular. Consequently, attempts at measurement,
description and analysis of poverty have expanded accordingly. We discuss four such attempts.

The first relates to the inclusion of non-monetary income and consumption – the ‘hidden harvest’
of subsistence goods that do not enter the formal cash-based economy. These include forest
products and subsistence production in agriculture. For many (poor) rural households these may
be more important than cash income. This expansion does not represent a fundamental shift in
the definition of poverty but points to the critical importance of including goods that do not enter
the marketplace. Indeed, the standard economic definition of income includes both subsistence
and cash income.

Second, during the 1970s, the emphasis in the development debate gradually shifted from
hardcore economics to that of the ‘basic needs’ of the poor. This accompanied changes in the
measurement of welfare. The poverty concept was thus subjected to what one might call a
‘human development extension’, implying that increasing attention was given to indicators
related to health, education and nutrition. The most popular indicator has been the Human
Development Index (HDI) published in UNDP’s annual Human Development Report. HDI
includes per capita income but with decreasing welfare ‘returns’ over growing income levels –
thus, an extra dollar in a poor country has a greater HDI effect than in a rich one. In addition,
health (with life expectancy as the indicator) and education (with literacy and school enrolment
indicators) are incorporated into the index. Thus HDI should capture the access to public goods
and services, such as health and education, and not just income in the definition of poverty. At
the individual level access to such services improve welfare, and interpersonal comparisons
disregarding unequal access to public services will be misleading. At the country level, HDI
‘punishes’ those countries that have unequal income distribution and poor public service
provision. Nonetheless, one can still argue that the choice of specific HDI indicators as well as
the weighting among the three indicators (income, education and health) remains arbitrary, as
highlighted by Ravallion (1997).

Third, opposition against the belief that economic growth automatically takes care of all human
needs also came from the environmentalist corner, notably with works such as ‘Limits to
Growth’ (Meadows et al., 1972). Again, this had implications for the poverty concept and during
the 1970s and 1980s, the emphasis gradually shifted from physical, man-made capital to human
and natural capital as the foundation for welfare improvements. This was, in the late 1990s,

4
conceptualized further through the Sustainable Livelihoods Approach (SLA) and the ‘Five-
Capital Approach’, suggested by Carney (1998), Scoones (1998), Bebbington (1999) and Ellis
(2000). For example, the livelihood approach – comprising of ‘the capabilities, assets and
activities required for a means of living’ (Warner 2000) – was made the cornerstone of the
poverty reduction strategy of the Department for International Development (DFID) in the
United Kingdom. In the ‘Five-Capitals Approach’, natural, human, social, physical and financial
capital represent the main asset categories. Poverty is defined as a lack of the assets needed to
generate satisfactory livelihood outcomes (e.g., income, well-being, vulnerability, food security
and sustainable use of natural resources).

Fourth, since the mid-1990s, we have witnessed what one could call an ‘empowerment’ and
‘institutional’ extension of the poverty concept. One of the reasons for this change was a
growing recognition that these power-related factors are welfare creating in their own right (e.g.,
it is more satisfactory for people to live in a society with democratic rights). Additionally, they
were assumed to have a positive impact on the creation of material benefits, as suggested in the
debate on whether democratic institutions enhance economic growth (Rodrik 2000).
Empowerment is thus both a means and an end to higher welfare. In recent years, there has been
an emphasis on ‘institutions’ as an independent factor in the SLA concept, although there is a
clear overlap here with the concept of ‘social capital’. Amartya Sen has also greatly influenced
the empowerment and institutional extension through his work which includes Development as
Freedom (Sen 1999).

The drive for broader poverty definitions has not only been expressed in the publications and
work of NGOs and bilateral donors, but also by the leading international donor – the World
Bank. In its World Development Report of 2000/2001, the World Bank (2000) introduced a
three-dimensional understanding of poverty as consisting of: (a) ‘opportunity’ (e.g. income,
education and health, i.e., similar to UNDP’s HDI); (b) ‘security’ (e.g. vulnerability understood
as the likelihood and magnitude of shocks and the ability to deal with them); (c) ‘empowerment’
(e.g. access and control over local resources, public services, influence in local decision making,
etc.). Difficulties in measuring and comparing the ‘security’ and ‘empowerment’ dimensions are
also obvious from the World Bank’s own use of the poverty term. When referring to poverty
reduction in other publications, it almost exclusively refers to a reduction in income poverty (e.g.
the notion of ’a dollar a day’). In short, this development over the past four decades is being
characterized by a move:

5
• from a reductionist, one-dimensional, index to a vector with multiple indicators;
• from materialistic to ‘soft’ assets (capitals) with decreasing tangibility, measurability and
comparability;
• from (intended) objective measures to (consciously) subjective ones;
• from an economistic ‘top-down’ approach to a holistic ‘participatory’ one, and
• from a pure outcome measure to a causality-inclusive indicator.

What is a useful definition of poverty?

Choosing a definition is not a question of a right or a wrong, but rather of how useful it is for a
particular purpose and context. Usefulness can only be defined in terms of its purpose, and a
poverty definition has at least three important purposes and uses: First, it should be useful in
policy debates and formulations, e.g., to define the scope of poverty reduction strategies (PRS).
Second, it should help in targeting and measuring the impact of specific poverty alleviation
programmes and policies. Third, it should be useful as an analytical concept to understand and
analyze poverty, and also measure changes. It may well be that different purposes call for
different concepts. The trend over time to use more inclusive measures is a reflection of new
research on the causes of poverty, including proposed solutions, political trends and pressure
from NGOs and others to broaden previously narrow definitions. With poverty back on the
international agenda, donors may also use poverty alleviation as an umbrella concept for several
development objectives (e.g., health, education, equality, equity, economic growth and
empowerment) in much the same diffuse way as ‘sustainable development’ was/is used. Thus,
there is a risk that ‘poverty’ might become all too encompassing and therefore eventually too
vague to be useful for analytical and practical purposes.

There is a need to distinguish between the conceptual analysis and the measurement of poverty.
Since poverty contains an important quantitative dimension, a key criterion for a poverty
measurement indicator is to allow for a consistent distinction between the poor, the not-so-poor
and the non-poor. Income (monetary and non-monetary) and consumption are still key concepts
in this respect, while most other indicators are poorly suited for the job, especially for
comparative purposes. In contrast, the sustainable livelihoods and five-capital approaches can
help us better understand the causes of poverty-related processes, especially in specific local
contexts.

6
It would be natural in a philosophical discussion of the poverty concept to include both its multi-
dimensionality and its subjective character. In practical work – and in poverty measurements and
comparisons – one must, however, make several necessary and simplifying assumptions. In this
context, Ravallion (1997) notes the great divergence between the comprehensive definitions of
poverty found in the recent literature and what is possible to operationalize and measure on the
ground – both in terms of methods and data availability. Hence, for research purposes, it may be
convenient to think about poverty in terms of the livelihoods approach or the three-dimensional
method outlined by the World Bank, while at the same time measuring the material indicators
that are closer to the original meaning of the term. This means that measurement should be done
in terms of income, while the value of consumption or other multi-factor indices may take into
account dimensions of human development. While a broad analysis of poverty is important, a
definition close to its original meaning is more suitable both for analytical purposes and as well
as in practical policies for targeting and measuring impact and effectiveness. For the sake of
clarity, we will in the following refer to this as income poverty. This would at the same time
acknowledge that reducing income poverty is not the only aim of development policies. An
additional argument for using an apparently narrow income or consumption-based definition of
poverty is that, in the long run and at an aggregated scale, economic growth and reduction in
poverty broadly defined (income, health, life expectancy, education, vulnerability and political
influence) often go hand in hand, cf. Ray (1998) and World Bank (2000).

Poverty measures

Having decided on an income or consumption-based definition of poverty for quantification

purposes, the next challenge relates to defining the poverty line. The most widely used income
poverty line internationally is the ‘one dollar a day’ measure, or more precisely 370 US dollars
per year at their 1985 value (World Bank 1990). 2 In addition, national governments have their
own poverty lines, based on what is considered a minimum income to meet basic needs in the
country. The need for an international standard has lead to widespread adoption of this measure,
although one certainly could question the empirical basis for this. Other poverty lines have also
been used, including extreme poverty (0.6 dollar) and moderate poverty (2 dollars a day).

2
This poverty line is applied in the individual countries controlling for differences in Purchasing Power
Parities (PPP), that is, the differences in the costs of living across countries.

7
Organisations like the OECD and the EU apply relative measures of poverty for their member
countries, i.e., the poverty line is set as a certain percentage (e.g., 50 or 60%) of the median
income. In poor countries, however, the absolute measures of poverty continue to dominate. The
choice between absolute or relative poverty lines is particularly important in the discussion on
the link between poverty and growth. If one applies a relative poverty line, economic growth will
increase that poverty line and only reduce poverty to the extent it is accompanied by a more
equal income distribution. Poverty defined in terms of a fixed poverty line (e.g. a dollar a day)
can be achieved with an upward shift in average income without changes in relative incomes. It
may, however, be argued that absolute poverty measures capture more immediate and pressing
concerns than relative deprivation, in that per capita incomes below one or two dollars a day will
typically be insufficient to satisfy even the most basic needs for healthy food, education and
health care. The absolute poor will of course also be suffering relative deprivation. Further, the
meaning of ‘relative poverty’ is also captured by the term ‘inequality’, which we believe is a
more appropriate term instead of overloading the concept of poverty.

Given these poverty lines, the standard measure is headcount poverty, in both absolute (number
of people below the poverty line) and relative terms (percentage of population below the poverty
line, i.e. the headcount ratio). The difference between the two is important as according to some
measures, the absolute number of poor in the world has increased slightly during the 1990s,
while the share of the global population that is poor has decreased. Thus answering the question
as to whether poverty in the world is declining will depend on whether we use an absolute or a
relative definition. This headcount measure – also referred to as the incidence of poverty – is a
zero-one measure, and does not take into account ‘how poor are the poor’. Other measures have
been applied to capture this, e.g., the income gap ratio, which measures the average income
shortfall of the poor (e.g., a measure of 0.8 means that on average the income of the poor is 80 %
of the poverty line). 3

MECHANICAL AND EMPIRICAL RELATIONSHIPS

Poverty reduction = growth + distributional change

Given the above discussion on the definitions of income poverty, growth and inequality, there is
a simple mechanical relationship between changes in headcount poverty, economic growth

3
See Fields (2001) for a further discussion of different poverty measures.

8
(higher average income) and changes in the income distribution. 4 This is graphically illustrated
in Figure 1. All individuals are ranked according to income, as represented by the solid bold line.
The absolute headcount poverty is given by the number of individuals falling below a predefined
absolute poverty line. There are two ways of reducing poverty. First, through redistribution, the
income line will rotate around the average income (i.e., no economic growth, so average income
is constant). This is shown by the stacked line. Second, the average income can increase (=
economic growth) in a way that does not change the income distribution, i.e., all individual get
the same relative (percentage) increase in income. This is shown by the dotted line in the figure,
and is termed ‘neutral growth’. Note that since the starting income of the poor is lower, the
absolute increase will be higher for richer individuals, and absolute differences will therefore
increase. Inequality measures are, however, neutral with respect to income levels, and only
relative differences matter.

Figure 1: Changes in poverty due to economic growth and/or changes in the income

distribution

Neutral growth

Redistribution
Average
income

Poverty
line

Headcount Total
poverty population

4
See Ravallion and Chen (2002) for the technical details on how this decomposition can be done.

9
Two important conclusions emerge from the discussion so far. First, both economic growth and
redistribution can be used to reduce poverty, and should therefore be considered as policy
alternatives at the outset. Second, economic growth will reduce income poverty, unless it is
accompanied by a significant worsening of the income distribution. Neutral growth (no
distributional changes), or growth which only worsens the income distribution slightly, will
increase the income of the poor and thereby reduce headcount poverty. Thus, ‘a rising tide lifts
all the boats’. Two questions then arise. First, what are the empirical links between growth and
changes in income distribution? And, can we achieve pro-poor growth, that is, growth where we
get a double effect on poverty reduction by both higher average income and a more equal income
distribution? These are addressed in the following sections of this chapter.

Growth and inequality

Economic growth usually changes the income distribution in a country – the extra pieces of the
growing cake are not distributed to all members of society in shares equal to their initial shares
of the cake. Economic growth is often due to, and/or accompanied by, new market opportunities.
At the beginning of an economic development process, higher market contact often increases
inequality as individual actors respond variably to favourable prices or other newly emerging
opportunities, because of specific skills and asset they possess, but also because of individual
differences in risk aversion, entrepreneurial spirit, and luck. These variable outcomes may also
impact the social coherence of communities and ultimately have certain negative welfare effects.
Obviously, this is what many observe at the micro-level, justifying their scepticism towards
economic growth.

In contrast, the following income equalising factors may start to work over time: more people
acquire the necessary skills or assets; new technologies spread to more producers; more efficient
markets eliminate price differentials across locations; demand for unskilled labour increases and
the higher income has local multiplier effects (e.g., increased demand for locally produced
commodities), insurance mechanisms are put in place to better distribute risks. These are some of
the micro-level mechanisms that have justified the hypothesis that inequality follows an inverse
U-curve over time – inequality rises in the initial development phase but then declines.

Similarly, there are macro-level processes that can ‘drive’ such a development in inequality over
time. Imagine a country with a large, low wage traditional sector employing 99 % of the labour
force and a tiny, high wage modern sector ‘island’, accounting for 1 % of employment. Imagine
10
also, that the wages in each sector remain fixed, so that economic growth in this dualistic
economy can only be achieved by modern sector enlargement, and 3 % of the total labour force
is transferred each year from the traditional to the modern sector. It can be shown that national
income inequality (e.g., measured by the Gini coefficient) in this trend scenario will follow an
inverted U-curve, i.e. the curve will rise initially and then start to fall (Williamson 1985).This
dualistic process of modern sector enlargement has high relevance for most developing
countries.

The inverse U-curve can also be supported by other societal trends during the process of
economic development. In the early stages of development, skills and higher education are
limited to a small group of people, who benefit the most from economic growth. Over time, the
greater spread of skills and secondary and tertiary education will have an income equalising
effect. Similarly, the development of insurance markets tends to improve risk management.
Sharing risk between economic agents becomes another equalising force.

Is inequality in the world actually following an inverted U-curve or is it just a theorist’s fata
morgana? Do things have to get worse (more unequal) – as an initial result of growth and
structural change – before they eventually can get better (more equal)? In his pioneering article
in the 1950s, Simon Kuznets – the ‘father’ of the inverse U-curve or Kuznets curve – observed
this empirical pattern over time in three developed economies (the United States, Germany and
the United Kingdom). He also noted that inequality levels were somewhat higher for some low-
income countries (e.g. Ceylon, Puerto Rico and India) and considerably higher for others (e.g.
Kenya and Rhodesia) which, at that time, had relatively rapid growth rates (Kuznets 1955).
Later, Kuznets backed up his conclusions with time series data for 16 countries, 9 of which were
developing countries (Kuznets 1963). In the long term, inequality seemed in most cases, to be
reduced well below the ‘initial’ level. As (Kravis 1960: 409) argued: ‘The distribution of income
tends to be more equal the longer and more thoroughly the country has been exposed to the
processes of economic and social change associated with the idea of industrialization’.

While the Kuznets curve has been seen by many as one of the empirical regularities in economic
development, since the mid-1990s, many scholars including Deininger and Squire (1998),
Ravallion (2001) and Fields (2001) have questioned the existence of such a pattern. Several
researchers have thus tested the Kuznets hypothesis based on cross-country data. Yet, the
problem is that the inter-country differences normally are much larger than the variation

11
determined by the development phase, and hence the cross-country pattern cannot be given a
time series interpretation. Specifically, some middle-income countries in Latin America and the
Middle East are very unequal and come to dominate the cross-country pattern. Hence, the U-
curve critics argue that income distribution tends to be very different across countries but fairly
stable over time. Thus the structural differences across countries can create the illusion of an
inter-temporal Kuznets curve. Consequently, the limited long-term time series data available
give only limited support to the existence of a general pattern as hypothesised by Kuznets.

Nevertheless, some analyses have supported the Kuznets hypothesis when applied to specific
countries (Williamson 1997). For example, China provides an affirmative case of pronounced
dualism that may generate a Kuznets curve. After 1984, income inequality rose as a result of a
widening gap between the high growth urban/coastal areas and the rural hinterlands. Inequality
increased in China and in Indonesia and Russia, ‘by differential access to the benefits of the new
economy, not by widening gaps among those who participate in it, or among those who do not’
(Lindert and Williamson 2001: 36). Although inequality in China has increased during the recent
high growth periods, trickle down effects have still made the poor better off in absolute terms.
The number of Chinese who are poor by the ‘one dollar a day’ standard has decreased by about
186 million over the last 30 years (Sala-i-Martin 2002).

Generally speaking, available income distribution time series data is too poor to fully resolve the
Kuznets controversy. In some cases, inequality rises, in others it decreases (Fields 2001). Why
do the micro and macro level processes from above not consistently produce Kuznets curves at
the national level? One possibility is that aggregated national data ‘conceal’ micro processes by
adding data for regions that are at different stages in their regional development process. Another
reason is that the preconditions and types of growth eventually matter more than income levels in
regard to how economic growth changes the income distribution. This is discussed in the final
section of this chapter under the heading of ‘pro-poor growth’.

Research over the past decade has also looked into the reverse link – how inequality affects the
growth rate. Traditionally, two arguments have been forwarded in support for the view that
‘inequality is good for growth’. First, that savings rates often increase with income, thus
redistributing income from the poor to the rich should increase aggregate savings. Early
economic growth models stressed the role of savings, investments and capital accumulation for

12
economic growth. Second, a certain degree of inequality reflects that people are paid according
to merit, and this creates incentives for hard and smart work.

More recent work has stressed that inequality is harmful for growth, for at least three reasons.
First, high inequality increases the demand for redistribution and may lead to higher taxes and
other measures which have a negative impact on the growth rate. Second, high inequality leads
to more social conflicts and instability, which also are harmful for the growth prospects. Third,
and in our view the most convincing argument, high levels of inequality exclude the poor from
‘joining the party’. In other words, limited access to land, education, credit, etc. make it difficult
for the poor to exploit new opportunities created. For instance, highly talented and
entrepreneurial people from the lower income strata will find it basically impossible to advance
their skills, creating limited social mobility. These factors of restricted participation and tend to
lower the growth rate. We return to this in the final section of the chapter. Fourth, countries with
high inequality develop institutions and policies that favour the rich and may invest too little in
public goods that both favour the poor and are growth enhancing, such as universal basic
education and rural infrastructure. Inequality is entrenched in institutions that are detrimental to
growth (Engermann and Sokoloff 2005).

There is some empirical evidence supporting the view that ‘inequality is harmful for growth’,
e.g., Alesina and Rodrik (1994). However, the lack of good data, and methodological challenges,
make it difficult to firmly conclude on this issue, and we need to be wary of simply adopting
politically correct solutions.

Growth and poverty

If there is no macro-level evidence for systematic changes in income distribution during

economic growth, we can expect growth to facilitate poverty reduction. Fields (2001: 99)
concludes the following from his literature review: ‘It is overwhelmingly the case that growth
reduces poverty and recession increases it, though in about 10 percent of the cases, poverty did
not appear to fall when growth took place.’ Likewise, Ravallion (1997: 637) notes on his
discussion of the Human Development Reports, that ‘arguably the biggest problem facing the
world’s poor today is not “low quality growth” – in HDR terms – but too little growth of even
quite normal quality!’

13
Another comprehensive study by Dollar and Kraay (2001) is appropriately called Growth is
Good for the Poor. The authors studied a series of 236 cases of economic growth episodes (i.e.
five year periods) in 80 countries, and tested how the income of the poorest 20 % change during
the economic growth process. They reached several conclusions which are relevant for our
discussion:
The elasticity of income to the poor with respect to average income is slightly above one, that is,
a 1 % increase in average income increases the income of the poorest quintile by 1.07 %. Thus,
the poor’s share of the cake is lightly increasing during growth.
About 80 % of the income level of the poor can be explained by the average income. In other
words, people are poor because they live in poor countries, rather than in countries with a
skewed income distribution.
However, looking at changes in the income, a smaller share – about 50 % – is explained by
changes in average income. Thus changes in income distribution matters too for the income of
the poor. There is no significant difference in these results between rich and poor countries, cases
with positive and negative growth, and experiences before and after 1980.

Other studies have looked at the growth elasticity of poverty, i.e., the reduction in headcount
poverty due to economic growth. The average elasticity is in the range of minus 2-3, which is to
say that a one percent increase in average income reduces the headcount poverty by about 2-3 %
(Ravallion 2004). But such elasticities vary greatly, depending on the initial poverty incidence,
the initial income distribution and – partly correlated with that – the depth of poverty.

Thus, any generalised statement that ‘economic growth does not generally benefit the poor’ is
wrong. Generally speaking, the poor benefit about as much in relative terms as the rich.
Economic growth is thus better than its reputation. Macroeconomic growth in most cases does
actually trickle down to raise the absolute incomes of the poor, at least over time and at
aggregate scales. The term ‘trickle down’ – commonly used to describe this effect – can
nevertheless be misleading since it gives the impression of the cake being baked by the non-poor
and then shared with the poor. This is not the case. GDP is the sum of the total production or
income within a country. The lion’s share of the income of the poor is what they have earned
themselves, not transfers from the state, development agencies or individuals. Thus the income
of the poor increases because they produce more in agriculture, get better prices for their crops,
and higher off-farm wages. That is their contribution to economic growth, and not money that
trickles down from above.

14
PRO-POOR GROWTH

Growth is good for the poor, but some types of growth are better than others. ‘Pro-poor growth’
can either be defined as (1) growth which increase the income of the poor (and thereby reduce
the incidence of poverty), or (2) growth which increase the income of the poor by more than the
growth rate (Ravallion 2004). In the remainder of the chapter, we will use the latter and stronger
definition, i.e., growth accompanied by a more equal income distribution. This is also discussed
under the term the ‘quality of growth’ (Thomas et al. 2000). The prospect of achieving pro-poor
growth depends on two sets of factors – the preconditions that exist and the type of growth the
country experiences. These are discussed below.

Preconditions

There is strong evidence to suggest that an egalitarian asset distribution – in particular of land
and human capital – will enhance the poverty reducing effect of growth. The combination of
land reform, together with improved infrastructure, education and labour intensive urban growth,
was a key factor in the successful development of South Korea and Taiwan in the post WWII
period. In India, the variable degree of literacy among states is the prime factor explaining
different growth-poverty outcomes (Ravallion and Datt 2002). In terms of the reverse causality,
high asset inequality in poor countries can also obstruct growth. Conversely, a population with
widely distributed skills across regions and socioeconomic groups is an important growth
promoting asset, as mentioned above.

To investigate how the poverty-reducing effect of growth (‘growth elasticity of poverty’) is

conditioned by the income distribution, Ravallion (2004) analyzed data from 62 countries.
Although the experience obviously differs across countries, the following model gave the ‘best-
fit’ to the data:
Rate of poverty reduction = (– 9.33) * (1 – Gini index)3 * economic growth rate

Thus, the reduction in poverty is determined by two factors: the economic growth rate and the
Gini index (the measure of inequality discussed earlier). Note that the equality measure (1-G) is
raised to the power of 3, which dramatically magnifies the impact of inequality of growth on
poverty reduction.

15
Consider two countries, one egalitarian country E with a Gini = 0.3 and another country S
characterized by a skewed income distribution and a Gini = 0.6. Both enjoy a modest annual
economic growth of 2 % (per capita). These Gini coefficients are representative of the most
equitable and some of the more inequitable developing countries today. Given the model above,
country E will have its poverty incidence reduced by 6.4 % every year, while country S will have
its poverty incidence reduced by only 1.2 % per year. In other words, the inequitable country S is
expected to spend 60 years to cut its poverty rate in half, while the equitable E one will do the
job in just 11 years.

Hence, the initial income distribution matters a lot for the poverty reducing impact of economic
growth. First, for any given poverty rate, the depth of poverty is likely to be higher the more
unequal the income distribution is. Accordingly, a larger income increase is required to lift the
poor above the poverty line. Second, the growth process is likely to be more equalizing the more
equal the initial income and asset distributions are. If, for example, land distribution is relatively
even, also poor farmers may take advantage of new market opportunities associated with
economic growth. This is indeed what is observed by Ravallion (2001): growth tends to make
economies with an initially equal income distribution more equal, while conversely it can also
reinforce large pre-existing inequalities

Types of growth

While initial conditions are important, they do not tell the full story of pro-poor growth.
Fortunately, however, there is scope for governments to choose growth strategies which are pro-
poor. Generally speaking, it is beneficial for poverty reduction to have a growth path
emphasising any combination of the following five factors:

(1) More education: Education is a direct and embodied investment in poor people – once the
asset has been built, nobody can take it away. Education, and particularly improved and more
extensive primary education, represents a valuable pro-poor human capital. For instance, Sen
(1999) argues that one main difference between the recent growth paths of China and India has
been the larger education emphasis in the former, triggering greater poverty reduction. Similarly,
(Thomas et al., 2000) conclude that investment in education is the single most important factor in
simultaneously promoting economic growth, poverty reduction and a more equal income
distribution.

16
(2) Labour intensive technologies: While education is the prime pro-poor asset to gain,
unskilled labour is the most important asset they already possess. Growth paths that favour a
higher demand for unskilled labour will tend to raise employment and wages, which helps to
alleviate poverty. Such labour intensive sectors can be rural (e.g. small scale agriculture or wood
processing) or urban, with the latter including industries like textiles, electronics or construction
and services such as tourism or commerce. Conversely, growth strategies that are highly
intensive in the use of natural capital (e.g. mining or oil) or in man made capital/high skill
technologies, are not likely to benefit the poor much, especially in early development phases.

(3) Rural focus: Rural growth tends to reduce inequality in particular, especially in the early
development stages when the bulk of the poor are actually rural based (Mellor 1999).
Investments in rural infrastructure, technology and R&D can help support this strategy, as for
instance, under Indonesia’s New Order policies (Warr 2000). Conversely, countries that try to
jump-start modern sector development with strong urban biases tend to reduce poverty to a much
lesser degree.

(4) Technologies for pro-poor consumption goods: The poor not only benefit from growth in
their nominal incomes but also from technologies that make the goods they consume cheaper in
the marketplace, thus increasing their purchasing power. Staple crops are a primary example.
The evidence so far suggests that urban consumers, and not rural producers, capture the lion’s
share of the benefits from new agricultural technologies. Increased food supplies tend to lower
food prices because food markets in developing countries tend to be competitive and elastic
demand causes prices to decline. This is why the Green Revolution was good news for poor
urban consumers – their staple crops became considerably cheaper.

(5) Good governance: Clear legislation and transparent government policies that recognise (the
often informal) rights of poor people over land and resources are important in creating benefits
for the poor. In regard to the use of high value natural resources – such as timber – an important
negative example is the extraction of high rent products from rich natural forests, where valuable
economic rents (defined here, as high profits from exploiting natural resources) are often
captured by elites who manage to influence policies and rules in their favour. Specifically, it has
been shown that a high incidence of corruption has a strongly anti-poor effect (Thomas et al.
2000).

17
CONCLUDING REMARKS
We started this chapter by contrasting developmentalist (growth) and class-based (redistribution)
views on how to address the problem of poverty. The empirical evidence reviewed provides
significant support to the developmentalist view, and while it downplays the relevance of the
class-based approach, it does not totally discard it. A strong point in case is that it is hard,
empirically, to point to countries where any major poverty alleviation has been achieved without
economic growth. Places like Cuba, Sri Lanka or the Indian state of Kerala are exceptional cases
where sustained welfare gains have been made through redistribution focused strategies,
although more through gains in social indicators than in income. On the contrary, experiences in
East and Southeast Asia (Taiwan, South Korea, Thailand, Hong Kong, Singapore, Malaysia,
Indonesia) provide ample evidence of the poverty reducing impact of economic growth. The
political-economy obstacles of redistribution are much larger if one needs to take away a piece of
a stagnant sized cake from the rich, rather than have them accept that they will receive less of
whatever increment there is in the cake.

Economic growth remains the key vehicle for the reduction of income poverty in poor countries.
Most of the world’s poor live in South Asia and Sub-Saharan Africa. These countries are
characterized by generalized poverty, where the cake is simply too small to make redistribution –
even if it was politically feasible – have any major impact on the poverty rates. However, the
income (and asset) distribution aspect is of great importance for the magnitude of the poverty
reducing effect of economic growth. Thus in regions with high degrees of inequality, like Latin
America and the Middle East, the poverty reducing effect of economic growth is much smaller
and may, in a few cases, be negligible. The bad news is that inequality in a historical perspective
has been hard to reduce, with much larger variation between countries than over time within
individual countries. Again, the exception to this are some of the Southeast Asian “tigers”
(Taiwan, South Korea) that combined early land reforms with sustained labour-intensive growth,
thus significantly reducing initial inequality levels without an intermediate “Kuznets type” rise.
The good news is that policymakers, by promoting pro-poor types of growth, can still combine
growth with a more equal income distribution. In this task, education, rural focus and labour
intensive technologies are key issues.

The chapter has not gone into the difficult question of how to build and strengthen institutions
that promote pro-poor policies. These are currently some of the most challenging research issues.
We still lack the magic formula of development, and each successful country has its own unique

18
story to tell. 5 We should therefore have modest hopes for finding such a formula. Perhaps the
current status of research can be summarized by the following quote by Dani Rodrik (2004: 1),
who observed in a discussion of development policies advocated by aid and multilateral
organizations: ‘The central economic paradox of our time is that “development” is working
while “development policy” is not.’

REFERENCES
Adelman, I., and C.T. Morris. 1978. Growth and impoverishment in the middle of the 19th
century. World Development 6 (3):245-273.
Alesina, A., and D. Rodrik. 1994. Redistributive Politics and Economic Growth. Quarterly
Journal of Economics 108:465-490.
Angelsen, Arild, and Sven Wunder. 2003. Exploring the forest - poverty link: key concepts,
issues and research implications. Bogor: CIFOR.
Bebbington, Anthony. 1999. Capitals and Capabilities: A Framework for Analyzing Peasant
Viability, Rural Livelihoods and Poverty. World Development 27 (12):2021-2044.
Carney, Diana. 1998. Implementing the Sustainable Rural Livelihoods Approach. In Sustainable
Rural Livelihoods. What Differences Can We Make?, edited by Department for
International Development. London.
Deininger, Klaus, and Lyn Squire. 1998. New ways of looking at old issues: inequality and
growth. Journal of Development Economics 57 (2):259-287.
Dollar, David, and Aart Kraay. 2001. Growth is Good for the Poor. Washington D.C.: World
Bank.
Ellis, Frank. 2000. Rural livelihoods and diversity in developing countries. Oxford: Oxford
University Press.
Engermann, S. L. , and K. L. Sokoloff. 2005. Colonialism, Inequality, and Long-Run Paths of
Development. In NBER Working Paper.
Fields, Gary S. 2001. Distribution and Development. A new look at the developing world. New
York, Cambridge & London: Russel Sage Foundation & MIT Press.
Kravis, I.B. 1960. International differences in the distribution of income. Review of Economics
and Statistics (November):408-416.
Kuznets, Simon. 1955. Economic growth and income inequality. American Economic Review
XLV (1):1-28.
———. 1963. Quantitative aspects of the economic growth of nations: distribution of income by
size. Economic Development and Cultural Change 11 (2):1-80.
Lindert, Peter H., and Jeffrey G. Williamson. 2001. Globalization and Inequality: A Long
History. Paper read at World Bank Annual Bank Conference on Development Economics
- Europe, at Barcelona.
Meadows, D.H. et al. 1972. The Limits to Growth. New York: Universe Books.
Mellor, John. 1999. The Relationship between Growth in Agriculture and Poverty Reduction:
Manuscript.

5
Some of them are told in a nicely edited volume by Rodrik (2003).

19
Ravallion, Martin. 1997. Good and Bad Growth: The Human Development Reports. World
Development 25 (5):631-638.
———. 2001. Growth, Inequality and Poverty: Looking Beyond Averages. World Development
29 (11):1803-1815.
———. 2004. Pro-Poor Growth: A Primer. In World Bank Policy Research Working Paper.
Washington DC: World Bank.
Ravallion, Martin, and Shaohua Chen. 2002. Measuring pro-poor growth. Economics Letters 78
(1):93-99.
Ravallion, Martin, and Gaurav Datt. 2002. Why has economic growth been more pro-poor in
some states of India than others? Journal of Development Economics 68 (2):381-400.
Ray, Debraj. 1998. Development Economics. Princeton, NJ: Princeton University Press.
Rodrik, Dani. 2000. Institutions for High-Quality Growth: What They Are and How to Acquire
Them. Studies of Comparative International Development 35 (3):3-31.
———. 2004. A Practical Approach to Formulating Growth Strategies. Harvard University.
———, ed. 2003. In Search for Prosperity: Analytical Narratives on Economic Growth.
Princeton & Oxford: Princeton University Press.
Sala-i-Martin, Xavier. 2002. The World Distribution of Income (Estimated from Individual
Country Distributions). Cambridge, MA: NBER.
Scoones, Ian 1998. Sustainable rural livelihoods: a framework for analysis. Brighton: Institute
for Development Studies.
Sen, Amartya K. 1999. Development as Freedom. New York: Anchor Books.
Thomas, V., M. Aailami, A. Dhareshwar, D. Kaufmann, N. Kishor, R. López, and Y. Wang.
2000. The quality of growth. Washington, D.C.: World Bank.
Warner, K. 2000. Forestry and sustainable livelihoods: What part can forests and forestry play in
reducing poverty? Unasylva 51 (202).
Warr, Peter G. 2000. Indonesia: Trade shocks and construction booms. In Trade shocks in
developing countries. Volume 2: Asia and Latin America, edited by P. Collier and W.
Gunning. Oxford: Oxford University Press.
Williamson, J.G. 1985. Did British capitalism breed inequality? London: Routledge.
———. 1997. Industrialization, inequality and economic growth. London: Edward Elgar.
World Bank. 1990. World Development Report: Poverty. Oxford; New York; Toronto and
Melbourne: Oxford University Press for the World Bank.
———. 2000. World Development Report 2000/2001: Attacking Poverty. Washington D.C:
World Bank.

20