An often cited study from 2000 by the World Bank2 – ‘Growth is good for the poor’ – argues that, in general, the answer to the first question is yes. Using a sample of ninety-two countries from 1960 to 2000, the authors find that when average income rises, the average incomes of the poorest fifth of society rise proportionately. Why?
Empirically, the share of income accruing to the poorest fifth (bottom quintile) does not vary systematically with average income. The authors find that this holds across all regions, periods, income levels and growth rates.
More recently, one of the authors of this report, Kraay (2004), estimates that, in the short-run, 70% of variations in changes in poverty levels between countries can be explained by general income growth, and that this figure rises to 97% over the medium to long term. These are certainly impressive figures.
It is interesting to note how the first study was eagerly seized upon by a certain section of opinion in the international financial institutions (IFIs), particularly in the World Bank, but also within the International Monetary Fund (IMF). For some, this research, and its subsequent extensions, lends support to the view that efforts to proactively target the poor – so that growth would be ‘pro-poor’ – were often counterproductive, and could even adversely affect general growth rates, thus ultimately having a negative effect on poverty reduction.
However, as with many of the issues that we shall cover in this book, it is not quite as simple as that. It is perhaps human nature that we seize upon evidence that seems to support our pre-existing perspectives. However, when influential global development agencies have a dominant ‘institutional per- spective’, there is a danger that this can skew policy recommendations, not least because the type of research undertaken or commissioned by such agencies is also influenced by these institutional perspectives.
Here we shall take as balanced a view as possible of the evidence, but it would be nạve to think that this author is immune to these influences. As with the IFI policy community, I too have pre-existing perspectives, which the reader may discern without too much difficulty. Let us be open about this, however, whilst striving for as much balance as can realistically be achieved.
To return to the issue at hand, before we can assess the evidence on general growth vs. ‘pro-poor growth’, we must first define our terms. The first of these is straightforward enough and can be measured as changes in per capita GDP. What of the second though?
2.2.1 What is pro-poor growth?
On one level, growth that is pro-poor can be defined as that which results in a significant increase in the incomes of the poor. Indeed, this is broadly the official definition adopted by the United Nations, for example. However, while no one would disagree with this definition, it does not take us very far.
Specifically, the debate turns on what we mean by ‘significant’.
Lopez (2005) identifies two strands of the literature, which answer this question in very different ways. In simple terms, the first of these strands focuses on the impact of growth on poverty with regard to the outcomes for inequality within societies, whilst the second focuses on changes in ‘head- count’ poverty rates, regardless of any changes in inequality.
(a) Poverty and inequality
Within this strand of the literature, growth is seen as pro-poor if it at least proportionately benefits the poor. That is, if average incomes rise by 5%, but the incomes of the poor rise by 3%, this is not seen as pro-poor growth.
White and Anderson (2000) propose two categorisations of pro-poor growth within this framework. The first suggests that growth is pro-poor if the incomes of the poor rise by more than the average growth in incomes – if average incomes rise by 3%, but the incomes of the poor rise by 5%, we have pro-poor growth. This definition therefore sees pro-poor growth as reducing the relative inequality within societies.
The second, more radical, definition defines pro-poor growth as that which reduces absolute inequality. Here, for growth to be pro-poor, the share of total income growth which goes to the poor must be equal to or greater than their share of the population. Thus, if the poor constitute 60% of the popula- tion, then they must receive at least 60% of the proceeds of economic growth for us to be able to describe this as pro-poor growth.
(b) ‘Headcount’ poverty
The second approach takes a far less exacting approach to defining what is
‘significant’ in terms of the growth of the incomes of the poor relative to
general income growth. Ravallion and Chen (2004), for example, simply argue that growth is pro-poor if it raises the incomes of the poor, regardless of any effects on inequality.
(c) Pros and cons
Lopez (2005) describes what he sees as the advantages and disadvantages of both approaches. For those focusing on inequality, growth defined as pro- poor could actually result in less poverty reduction than that not defined in this way. For example, if growth in general incomes is 2%, but the incomes of the poor rise by 3%, we have pro-poor growth. However, if general income growth is at 6%, but the incomes of the poor rise by only 4% this is not pro- poor growth, despite the fact that poverty has been reduced in the second example more than in the first.
Also, for Lopez, from a policy perspective, holding to the more stringent versions of this definition in particular could lead to a disproportionate focus on reducing inequality, perhaps at the expense of growth. However, to remember our earlier caveats, this is to risk confusing ends and means, if taken to the extreme. That is, the purpose of growth is to reduce poverty, it is not an end in itself. If poverty can be reduced, regardless of the impact on growth, this achieves the same ends.
Where this runs into difficulties, however, is that one can divide up national income on a more equal basis and therefore reduce poverty, but the ultimate impact will obviously depend on the size of the ‘cake’ – i.e. national income – that is to be divided. In many developing countries, particularly the least developed countries (LDCs), the ‘cake’ is simply not big enough, regardless of how it is divided. Consequently, while levels of inequality are obviously extremely important, so is general growth of national income.
However, a balance must clearly be struck, since a strict adherence to the definition proposed by Ravallion and Chen (op. cit.) can lead to some per- verse examples of growth that is pro-poor. For example, as pointed out by Lopez (op. cit.) such an approach would define a situation where average incomes rise by 6%, but the incomes of the poor rise by 0.1% as pro-poor.
As is often the case therefore, there needs to be a sensible balance struck between growth in average incomes relative to growth in the incomes of the poor. Below we consider the evidence on where this balance should be struck.
2.2.2 Inequality, growth and poverty reduction
At the beginning of section 2.2, we saw that Kraay (2004) estimates that the variation in changes in poverty levels between different countries, over the short term, can be largely (70%) explained by differences in general rates of income growth. Furthermore, the same study estimates that the explana- tory power of general income growth rises to 97% over the medium to long term.
For some, this is the end of the matter. However, as we shall see, the answers one gets depend very much on the question that is asked, and the methodologies employed to answer it, particularly the level of aggregation.
For example, rather than examining variations in changes in poverty levels between countries, Ravallion (2004) focuses on the ‘growth elasticity of pov- erty’, which can be defined as the percentage change in poverty associated with a 1% growth rate (Lopez, 2005).
Where Kraay (2004) essentially says that poverty will be best addressed by maximising general income growth, Ravallion (op. cit.) demonstrates that this is a specific not a general case. In particular, the outcome in terms of poverty reduction of increasing aggregate income will be strongly determined by the initial level of inequality in society. Specifically, the lower the level of inequality to begin with, the greater the impact of general growth on poverty levels – i.e. the higher the growth elasticity of poverty. In this regard, Raval- lion estimates that a 1% general growth rate will produce 4.3% growth in the incomes of the poor in very equal countries, but for countries with very high levels of inequality the same level of growth results in the incomes of the poor rising by just 0.6%.
Clearly, therefore, the particular circumstances of individual countries will have a strong impact upon policy outcomes. As stressed in the introduction, policy X (i.e. maximising general growth) will not always result in outcome Y (i.e. maximum poverty reduction).
Lopez and Servén (2004) employ a different methodology, but arrive at a similar conclusion to Ravallion (op. cit.); namely, that high levels of inequal- ity are an obstacle to poverty reduction. However, they also find that poverty itself represents a similar obstacle to poverty reduction.
Lopez (2005: 7) summarises the policy implications of these findings as follows:
. . . these findings would justify poverty reduction strategies with a pro- growth bias in low income and low inequality countries and policy pack- ages that adequately balance growth and inequality objectives in richer and more unequal countries.
We have explored some recent research into the impact of growth on poverty, and seen how levels of inequality in a society within a society can influence this relationship. In Box 2.1 below we consider how these issues relate to a fast-growing economy with high levels of poverty: India.
Box 2.1 Growth and poverty in India
As we can see from the figure below, in recent times India has exhibited high and sustained rates of economic growth, averaging almost 7% per year from the early 1990s to 2004. However, as we can also see, absolute
poverty remains endemic: in 1993 85% of the population lived on less than US$2 per day; by 2004 after a decade of high rates of growth, this had fallen a little, but remained very high at more than 80% of the population.
Furthermore, even this relatively small reduction in poverty has been very unevenly distributed throughout the different regions of the coun- try, with much of the difference appearing to relate to ‘initial conditions’
in terms of land ownership and human capital:
1 States that have had more land reform tended to have had more rapid poverty reduction even though they have (if anything) had slower agricultural growth (Besley and Burgess, 2000).
2 States with more labour regulation have tended to have slower growth in manufacturing (Besley and Burgess, 2000).
3 States with more rapid bank branch expansion into unbanked areas have experienced greater poverty reduction (Burgess and Pande, 2005).
4 States with systems of land revenue collection that remained in indigenous hands have tended to experience better public goods provision (Banerjee and Iyer, 2002).
5 States with more rural industrialisation have reduced poverty more rapidly (Foster and Rosenzweig, 2003).
6 States with more female literacy have done better in reducing poverty (Ravallion and Datt, 2002).
Thus while growth may be a necessary condition for poverty reduction, it is far from being sufficient. The IMF estimates that India will have to Growth and poverty in India, 1993–2004.
generate at least 100 million jobs over the next decade simply to stop unemployment from rising. If there is to be an equitable distribution of the proceeds of growth, however, it is clear that the sources of the current imbalance in this regard that are described above will also need to be seriously addressed.
However, the relationship between growth and inequality has itself long been a subject of research and (sometimes fierce) debate. In the next section some key issues in this regard are considered.
2.2.3 Growth and inequality: some history, theory and practice
(a) From growth to inequality
The linkages between growth and inequality have a long tradition, which in modern times can be traced to the work of Simon Kuznets. Kuznets received the Nobel Prize in 1971 for his groundbreaking, empirically based work on the process of economic development and growth. His key insights in this respect live on in the form of the famous ‘Kuznets curve’.
Box 2.2 Simon Kuznets
Simon Kuznets (1901–1985) was an American empirical economist who was a founding father of the field of econometrics. Today Kuznets is best known for his empirical work on the impact of economic growth on income distributions, which is described below. Although Kuznets received the Nobel Prize in 1971 for his work on the empirics of growth, he was also influential in the economics of public accounts, the empir- ical analysis of Keynes’ Absolute Income Hypothesis, which he found did not hold over the longer term, thus opening the door for Milton Friedman’s Permanent Income Hypothesis, which sought to explain the relationship between income and savings rates.
Kuznets’s empirical work used cross-sectional analysis of countries at differ- ing levels of development, where the relationship between per capita income and inequality was assessed. Kuznets (1955) found that in the early stages of growth income inequality tends to widen, before levelling off once income reaches a certain level and then falling as incomes continue to rise.
For Kuznets, this relationship could be explained by the process of devel- opment itself, notably the shift from agricultural to industrial production. He argued that in the early stages of growth, where much economic activity is based on agricultural production, the movement of people from the land to
the urban centres would lead to a widening of inequality levels. This process was driven by the fact that the potential for productivity increases in indus- trial production in urban areas was greater than in the agricultural sector, so that urban (industrial) workers’ wages grew much more rapidly than rural incomes. As urbanisation progresses inequality continues to increase, peaking when half of the workforce have moved into the industrial sector, and then declining as industrialisation continues and more and more workers are drawn into the higher-paid sectors.
Kuznets’s work was extremely influential, leading many to argue that grow- ing inequality in developing countries was a ‘natural’ part of the process of development, which would be inevitably reversed as incomes continued to rise. If this position is accepted, then there is no point in attempting to address high levels of inequality in developing countries. Indeed, if these efforts serve to slow the rate of growth, they may be seen as counterproduc- tive, as they merely postpone the point at which the high point of the Kuznets curve is reached and inequality begins to fall.
The universal applicability of the Kuznets curve is now widely questioned.
From a methodological point of view, Kuznets based his empirical work on cross-sectional analysis of different countries (at different levels of develop- ment) at the same point in time. Critics have argued that a time-series analysis of the development of individual countries would have been a more appropriate approach to take. In particular, at the time Kuznets undertook his work, many of the countries he used to produce the middle-income part of his curve were in Latin America, which is a region characterised – then as now – by high levels of inequality. Consequently, rather than being an ‘iron law’ of development, which is applicable to all countries, the ‘relationship’
described in the Kuznets curve may represent the unique circumstances of particular countries at one given point in time, and be skewed by the fact that, at the time, many middle-income countries happened to be in Latin America.
The status of the theory has also been adversely affected by the develop- ment experience of many countries in East and Southeast Asia, where growth has been associated with very low levels of inequality, the pattern of which in no way resembles that depicted in the Kuznets Curve.
However, the fact that the proposition has had such a powerful impact upon policy-makers historically again highlights the dangers of assuming that ‘iron laws’ of development exist, and that therefore the same policy approaches (or the absence of them) will result in the same outcomes.
This point is highlighted by more recent empirical work into the impact of growth on inequality, summarised in Lopez (2005). Research undertaken over the last two decades, using a variety of methodologies and case-studies, has all concluded that there is no consistent, predictable relationship from growth to inequality. That is, there is no tendency for growth to result in either an increase or a decrease in levels of inequality.3 Either effect may occur, but this is due to the particular circumstances of the countries concerned, not an inherent feature of the process of economic growth and development.
(b) From inequality to growth: 1
We have seen that levels of inequality can have a strong impact upon the effectiveness of economic growth in reducing poverty levels: the growth elas- ticity of poverty. Other strands of research have also examined the question of how inequality may, in turn, impact upon levels of economic growth.
Box 2.3 Nicholas Kaldor
Nicholas Kaldor (1908–1986) was a Hungarian-born economist who, after studying in Berlin, spent his academic and policy-focused life in the UK, most prominently at the University of Cambridge. As well as his work on inequality and growth, Kaldor’s work on inefficiency led to a more useable modification of standard Pareto efficiency forms. For a change to be Pareto efficient, it must be that someone is made better off, with no one becoming worse off. Kaldor-Hicks efficiency, in contrast, can arise where a change does lead to some people being worse off – which is far more likely to happen in practice – but that those better off could, in theory, compensate the losers and still remain better off them- selves. This approach has been very influential in the policy sphere, forming the basis for much cost-benefit analysis work to judge whether particular policies should be undertaken. As well as his academic work, Kaldor was an advisor to post-war Labour governments and was instrumental in early work to move from direct to indirect taxation – i.e. VAT.
At the same time as Kuznets was developing his work on growth and inequality, his contemporary, Nicholas Kaldor, was formulating a theory that, at least in some regards, can be seen as complementary. Kaldor (1956) argued that relatively high levels of inequality are associated with higher levels of growth. The ‘Kaldor hypothesis’ states that because wealthy people have a higher propensity to save – as much of their income is surplus to the require- ments of satisfying basic needs – unequal societies where income is concen- trated in relatively few hands will have higher savings rates than more equal societies, which is translated into higher investment rates and faster growth.
It is not difficult to see how the Kaldor hypothesis could be combined with the Kuznets curve, with the result not only that would there be no point in aiming to reduce inequality as countries develop, but that growth could be accelerated if inequality was actively encouraged. Furthermore, faster growth would hasten the point at which the Kuznets curve turned down, thus reducing inequality more quickly than would otherwise be the case. In short, the best (and fastest) way to reduce inequality in the long term would be to increase inequality in the short term.