UNECA Macro Training 4[1]

United Nations Department of Economic & Social Affairs UNDP Regional Bureau for Africa

Reading Material 4:

Pro-Poor Growth: Equity and Poverty Reductionby John Weeks Centre for Development Policy & Research School of Oriental & African Studies University of London

Table of ContentsPages

Introduction: The Millennium Poverty Goals 1.1 1.2 1.3 1.4 What is Pro-poor Growth? Theoretical Framework: Sources of Growth 8 Demand Determined Growth Growth, Distribution and Poverty Reduction: Theory 1.4.1 Redistribution and the Opportunity Cost of Growth 1.4.2 Comparing and Combining Growth and Redistribution 1.4.3 Inequality in sub-Saharan Countries 1.5 1.6 Policies for an Effective Redistribution with Growth Summary

1 3

10 14 14 17 22 28 38 40 46

Annex: Basic Growth Theory References

Module 1: Pro-poor Growth Equity and Poverty reduction

i1

Introduction: The Millennium Poverty GoalsOf the many issues central to the development process, few have been characterised by the shifts, reversals and re-affirmations that have plagued the analysis of the interaction of growth, poverty and inequality. The mainstream literature has not so much evolved as fluctuated over the past fifty years.1 From the 1950s into the 1970s emphasis was on probable tradeoffs between growth and income distribution. In contrast, work in the 1970s sought to identify redistributive mechanisms for poverty reduction without hampering growth.2 This was a short-lived focus of the literature, reversed with the rise of neo-liberalism and the Washington Consensus in the early 1980s. For the latter, growth itself would be the vehicle for poverty reduction, achieved through trickle-down mechanisms not always clearly specified. In the last decade, both the neo-liberal analysis and the earlier view of a trade-off between growth and equity were challenged by a number of studies.3 In the late 1990s the bilateral and multilateral development agencies placed increasing emphasis on poverty reduction in developing countries.4 This emphasis led to the establishment by the United Nations of the Millennium Development Targets. The achievement of a target requires policies, and policies are most effective within an overall, coherent strategy. Seven of the MDGs cover the various aspects of human welfare, and the eighth addresses international responsibility, especially for the developed countries, to create a framework to achieve them.5 The first MDG calls for1

2 3 4

See Kanbur (1998) for a thorough review. Dagdeviren, van der Hoeven and Weeks (2000) also provide review of the literature on growth, poverty and redistribution.See Chenery, Ahluwalia, et al. (1974). See, Aghion (1999), Alesina (1994), van der Hoeven (2000)

The International Development Targets, set by the Social Summit in 1996, are presented and discussed in Hanmer and Nascold (2000). The UK Department for International Development officially adopted these targets (DFID 2000, and Goudie and Ladd 1999). More modest targets were set by USAID (USAID 2001). The new emphasis of the international financial institutions on poverty is reflected in the inclusion of poverty strategies in loan agreements (see IMF and World Bank 1999). For a sceptical view, see Cramer (2000). 5 The seven welfare-related MDGs are: 1) to reduce by half the proportion of people living on less than a dollar a day and reduce by half the proportion of people who suffer from hunger; 2) Ensure that all boys and girls complete a full course of primary schooling; 3) Eliminate gender disparity in primary and secondary education preferably by 2005, and at all levels by 2015: 4) Reduce by two thirds the mortality rate among children under five; 5) Reduce by three quarters the maternal mortality ratio; 6) Halt and begin to reverse the spread of HIV/AIDS, and Halt and begin to reverse the incidence of malaria and other major diseases; and 7) Integrate the principles of sustainable development into country policies and programmes; reverse loss of environmental resources, Reduce by half the proportion of people without sustainable access to safe drinking water, and Achieve significant improvement in lives of at least 100 million slum dwellers, by 2020. The eighth MDG itself has seven parts: a) Develop further an open trading and financial system that is rule-based, predictable and non-discriminatory, includes a commitment to good governance, development and poverty reduction, nationally and internationally; b) Address the least developed countries' special needs. This includes tariff- and quota-free access for their exports; enhanced debt relief for heavily indebted poor countries; cancellation of official bilateral debt; and more generous official development assistance for countries committed to poverty reduction; c)


1

action to reduce by half the number of people in extreme poverty and those suffering from hunger. This module will focus on the former, to reduce by half the proportion of people living on less than a dollar a day (http://www.un.org/millenniumgoals/#). Thus, setting a specific level of poverty to achieve by a specific date makes comparison of redistribution and growth analytically interesting and relevant for policy. The Millennium Development Goal6 for head count poverty, which we use, was quite specific:The [MDG] for well-being is a practical measure of absolute poverty [US one dollar a day]. It is based on an average of national poverty lines in poor countries, which reflect peoples ability to afford a diet sufficient to meet minimum nutritional requirementsIt thus represents an internationally agreed operational method of identifying the number of people who by any standard have unacceptably low incomes. Thetarget is to reduce by half the proportion of people in developing countries living in extreme poverty by 2015. The base year is 1990 (DFID 2000, p. 11)

It may be that for some countries there is no feasible growth rate, given historical performance and changes in inequality and resource availabilities that would achieve it. The World Bank warned that such might be the case: Progress in reducing extreme poverty during the 1990s was constrained by increasing inequality in a few countries that accounted for a large share of the worlds poor. In looking ahead to 2015, continued increases in inequality coupled with less than robust growth would imply failure to reach the poverty target for developing countries as a group, and in particular substantial increases in the number of poor in Sub-Saharan Africa (World Bank 2001b, p. 7)7 The World Bank went on to conclude that the alternative [growth] scenarios highlight the importance of achieving fast growth, as well as distributing the benefits of growth equitably (World Bank 2001b, p. 10).8 The same point is made by UK DFID, without growth the poverty reduction target will not be achieved, but it is not enough on its own (DFID 2000, p. 11).9Address the special needs of landlocked and small island developing States; d) Deal comprehensively with developing countries' debt problems through national and international measures to make debt sustainable in the long term; e) In cooperation with the developing countries, develop decent and productive work for youth; f) In cooperation with pharmaceutical companies, provide access to affordable essential drugs in developing countries; and g) In cooperation with the private sector, make available the benefits of new technologies especially information and communications technologies. 6 Initially the MDGs were called International Development Targets. The former became the official name. 7 This document was taken off the internet, without pagination. Page numbers given here are based on numbering form the first pages of text (Introduction). 8 Evidence that the pattern of growth in both developed and developing countries became more unequal is presented in Cornia (1999). 9 For further discussion of the achievability of the targets see Demery & Walton (1998) and Hanmer &


2

Readings: http://www.un.org/millenniumgoals/ (Detailed statement of the MDGs and targets) http://ddp-ext.worldbank.org/ext/GMIS/home.do?siteId=2 (MDG monitoring) http://www.undp.org/mdg/ (UNDP role, MDGs at the country level) 1.1 What is Pro-poor Growth?

Empirical evidence shows that in the vast majority of cases, economic growth raises the incomes of those households at the bottom of the distribution. In this sense growth is good for the poor (Dollar and Kraay 2000). The term pro-poor growth is more specific, seeks to differentiate among periods of growth, sometimes called growth episodes, in which growth is more and less advantages to those below the poverty line. There is some disagreement and confusion over the use of the term propoor growth. This module defines the term as follows: Growth is pro-poor when the incomes of the poor rise proportionately more than the incomes of the non-poor.10 This definition, used by the UNDP International Poverty Centre, is demonstrated graphically in Figure I.1, where the vertical axis is household incomes in ascending order of size, and the horizontal axis is the growth rate of those incomes. If all incomes grow at the same rate, the vertical line, growth is distribution neutral, and all households enjoy equal proportional increases. The positively sloped line shows the case in which the rate of growth increases as one ascends the distribution. This would be consistent with some degree of poverty reduction, less so the flatter the slope of the line. Thus, it would not be strictly accurate to designate this line as anti-poor growth, a term which strictly speaking would be restricted to cases in which the growth rate is negative for households below the poverty line. Therefore, it is identified by the precise term, unequal distribution growth. Finally, the negatively sloped line unambiguously represents pro-poor growth, since the increase in the income of those below the poverty line is proportionately greater than for those above. It should be noted that the degree to which growth is pro-poor by this definition can be measured in a variety of ways. In this module an algebraically simple measure is used. At any point in time, the growth rate of household incomes is the weighted

Nascold (2000). 10 Eduardo Zepeda of the UNDP International Poverty Centre in Brasilia contrasts this with a broader definition: Ravillion [2004a] defines pro-poor growth as nay increase in GDP that reduces poverty. Such a definition is too broad: it implies that most real world instances of growth are pro-poor, even if poverty decreases only slightly and income distribution worsens during a period of strong growth. A more appropriate definition has growth as pro-poor if in addition to reducing poverty, it also decreases in equality[Kakwani, Khamdker and Son] propose a simple and sensible definitiongrowth is propoor, relatively speaking, if it benefits the poor proportionally more than the non-poor. (Zepeda 2004)


3

average of the rate for the poor and the non-poor. If the head count poverty ratio11 is H and the growth rates for the poor and non-poor are gp and gnp, then: g = Hgp + (1 H)gnp The extent to which growth is pro-poor over any time period can be measured by the following ratio: Index of pro-poor growth = IPPG = gp/gnp For any rate of growth of aggregate household income, this index is a function of the head count poverty ratio. If the aggregate rate is g*, then: gp = (g - [1 H] gnp)/H Figure I.1 shows the index of pro-poor growth for a five percent overall rate of growth and head count ratios of thirty and fifty percent. Below an index of unity (neutral distribution growth between the poor and non-poor), a higher head count ratio results in a lower measure of pro-poor growth, and above unity a higher measure. In Figure I.2 the index is presented for two different aggregate growth rates, three and five percent, and the same head count ratio. The diagram shows that the degree to which growth is pro-poor increases for any rate of growth as the head count ratio declines. This measure of pro-poor growth has the advantage of being easily measured from household surveys, assuming they are consistent in design, and allowing a cardinal measure of the degree to which growth is pro-poor. Its disadvantage is that the index is defined for positive values for the growth rates of each group. Readings: Barungi, Barbara and Eduardo Zepeda 2005 The Challenge of Pro-Poor Growth in Uganda, International Poverty Centre One Pager, Number 11 http://www.undp-povertycentre.org/newsletters/OnePager11.pdf Hanmer, L., and F. Nascold

In a reply, Ravallion defends his broad definition with an ad hoc argument. The policy implication of that argument is that distribution does not matter for poverty reduction (Ravallion 2004b). While agreeing that Ravallions definition is too narrow, Osmani takes Kakwani to task because he takes the [pro-poor] bias to the extreme (Osmani 2005). Osmanis alternative, a country and time specific benchmark against which a subsequent growth episode is judged, would seem ah hoc as Ravallions but less operational. With no clear rule for choosing a benchmark time period, no definitive conclusions could be reached as to whether growth in any period was pro- anti- or neutral with regard to the poor. 11 The head count ratio is the proportion of the population below a given poverty line.


4

2000 Attaining the International Development Targets: Will growth be enough? Development Policy Review 18, 1, pp. 11-36 Kakwani, Nanak, Shahid Khandker and Hyun H. Son, Pro-poor growth: concepts and measurement with country case studies, Working Paper # 1, August 2004 http://www.undp-povertycentre.org/newsletters/WorkingPaper1.pdf Ravallion, M. 2004b Defining Pro-poor Growth: A reply to Kakwani, International Poverty Centre One Pager, Number 4 http://www.undp-povertycentre.org/newsletters/OnePager4.pdf Son, Hyun H. and Nanak Kakwani, Economic growth and poverty reduction: initial conditions matter, Working Paper # 2, August 2004http://www.undp-povertycentre.org/newsletters/WorkingPaper2.pdf Zepeda, Eduardo 2004 Pro-poor Growth: What is it?, International Poverty Centre One Pager, Number 1 http://www.undp-povertycentre.org/newsletters/OnePager1.pdf


5

Box 1.1: Pro-poor Growth in Uganda, Success and ReversalA recent paper by Kappel et al (2004) provides a wealth of information and insights. The proportion of people living below the national poverty line had declined from 56%in 1992/93 to 34%in 1999/00; between 1999/00 and 2002/03,poverty increased to reach 38%.The growth performance of the economy during the second period was clearly inferior to the first one. Using changes in GDP per capita as the measure of growth, for example, the annual rate of growth declined from an average of 4.3% between 1993 and 2000 to 2.9% between 2000 and 2002. Assuming everything else remained the same, this lower rate of growth should have been responsible for the slow down in the pace of poverty reduction in Uganda, but it can hardly account for the reversal in poverty trends something must have changed drastically. For one thing, while inequality in Uganda measured by the Gini coefficient increased throughout the whole periodfrom 36.4 in 1992/93 to 39.5 in 1999/00 and to 42.8 in 2002/03, it did so at a much faster rate during the last years; in fact, the speed of increase almost doubled, from an average annual change of 1.2% in the first years to 4.1% in the last three years. Fast growth and poverty reduction during the 1990s were due to the immediate benefits of recovery from civil war and from overcoming the economic mismanagement that prevailed during much of the 1980s. According to Kappel et al, the two main factors explaining the rapid reduction of poverty and the strong growth of the 1990s were increases in the production of cash crops and high international prices for Ugandas export products...The economic reforms of the 1990s implied greater reliance on market conditions. When market conditions are favourable, as in the 1990s, particularly in the second half of the decade, the economy fares well, but when markets do not perform well, the economy suffers, especially the poor. By November 2001, the price of robusta coffee had decreased by almost 90%relative to its peak in 1994[and] the prices of cotton, tobacco and tea also decreased. The economic environment had changed drastically, the pace of the economy slowed down and poverty increased. Rapid growth and the substantial reduction in poverty of the 1990s are a welcome outcome for Uganda, especially for Ugandas poor. According to a minimalist definition,* the performance of the Ugandan economy was clearly pro-poor during the 1990s and not pro-poor after 2000, because there was poverty reduction in the first years but not in the second. However, a more demanding definition of pro-poor would tell us that the 1990s were not pro-poor and that the years after 2000 are a case of immiserising growth. But whether the 1990s should be considered as pro-poor or not pro-poor is a question that can lead to different policy conclusions. Accepting that the performance of the 1990s qualifies as pro-poor would, most likely, lead to a continuation of the same policy framework. In this scenario, one risk is to be unpleasantly surprised, as happened with the poverty reversion of the 2000s. If, instead, the informed dominant view holds that the 1990s were not benefiting the poor sufficiently, as a stricter definition of pro-poor suggests; then, policy makers and stakeholders are forced to look more carefully into ongoing policies. The poverty outcomes of the years between 2000 and 2003 will only reinforce such a stance. *That all growth that reduces poverty is pro-poor. See Section I.1 and Ravallion (2004). Text taken from Barungi & Zepeda (2005).


6

Figure I.1: Graphical Demonstration of a Strict Definition of Pro-Poor Growth income leveldistribution neutral growth

Unequal distribution

Growth

Income

poverty

line

Pro-poor Growth

0

growth rate

Figure I.2:

Indices of Pro-poor Growth for 5 percent Growth and Head Count Ratios of 30 & 50 percent4.5 4.0 3.5 3.0

g=5, H=50g=5, H=30

IPPG

2.5 2.0 1.5 1.0 .5 .0


.0

1.0

2.0

growth rate of the poor

3.0

4.0

5.0

6.0

7.0

8.0

7

1.2

Theoretical Framework: Sources of Growth

Fostering pro-poor growth in an economy is a practical policy process, and theory provides guidance to the effectiveness of the policies involved. Prior to considering how policy can make growth more pro-poor as measured by our pro-poor index, this section briefly reviews basic growth theory. The standard approach is to treat economic growth as a long term phenomenon which is supply determined. Most growth discussions are derived from or can be treated as special cases of the general Neoclassical model. The models applicability to actual economies is quite limited due its restrictive assumptions. However, it is useful tool for organising a typology of growth models. What follows is a non-technical presentation, with the algebra and diagram placed in an annex to this module for those who might find it useful. Those trained in economics should skip this section and go directly to the annex. Strictly speaking, the model is one of the potential growth rate. At any moment, an economys potential output or supply is determined by the quantity and quality of two basic inputs, capital and labour, and the prevailing technology, within an aggregate production function. Land is treated as part of the capital stock on the grounds that investment in its quality makes it unlimited. The algebraic forms that the aggregate production function can take are quite limited, since it must conform to several necessary conditions: that no output is possible without at least one unit of each input (the function must be multiplicative, not additive); if one input is held constant, the other must exhibit diminishing productivity (the principle of diminishing returns to the variable input); and, not required but typically a condition, a constant ratio of the prices of the inputs should be associated with a constant ratio of their physical use (a given wage-profit ratio implies a given labour-capital ratio). If these conditions hold, the growth rate becomes the sum of the proportional changes in the two inputs, each weighted by a coefficient less than one (the ensures the operation of the principle of diminishing returns), plus the rate of technical change. The potential growth rate is rendered the actual rate through perfect competition that ensures no idle labour or capital. The wage rate and profit rate which result from the clearing of the input markets dictates the labour-capital ratio (which on the potential growth path is equal to the ratio of the aggregate supplies of the inputs). The rate of change of technology is considered exogenous in the simple version of the model. The growth of the labour force is treated as naturally determined, equal to the rate of growth of population if one ignores changes in demographic composition. The rate of growth of capital is determined by the economys aggregate saving rate. The assumption of perfect competition ensures that all saving becomes investment, which by definition is equal to the change in the capital stock. One might reasonably assume that the growth rate could be increased by raising the saving rate, but such is not the case. The wage-profit ratio is determined by the relative supplies of the inputs. Were the saving rate to rise and the growth of the capital stock to increase, the aggregate supply of capital would rise relatively to the exogenously given growth rates of labour and technology. As a result, the profit rate would fall, lowering aggregate saving, which would lower the investment share in national income. This scenario would result in a return to the initial growth rate. Thus, a fundamental


8

conclusion of the basic model is that the potential growth rate of the economy is determined by the rate of growth of the labour force and technical change. The discussion so far has assumed that a proportional increase in both inputs results in an equal proportional increase in output if there is no technical change, constant returns to scale. There are two other cases, decreasing and increasing returns to scale. The former case is sometimes called the Ricardian case, because of David Ricardos model with a limited land input. Proportional increases in the two inputs results in an increase in output that is proportionately less than the increase in the inputs. This is typically ignored because it is considered to be empirically refuted. With regard to the second, various authors have suggested that developing economies are characterised in the aggregate by increasing returns. This is a theoretical possibility of no empirical relevance, because no one has discovered a useful specification of the Neoclassical production function that can distinguish between returns to scale and technical change. A further and quite serious practical problem with the Neoclassical model is that it is purely supply-driven, with no consideration that growth might be demand determined. This characteristic follows by definition from the assumption that growth occurs at full employment of the inputs due to perfect competition. The last ten years has seen the rapid growth of empirical work using the so-called endogenous growth theory. As Fine has shown (Fine 1998), this empirical work is typically ad hoc in its specification of estimating equations. As a consequence, endogenous growth theory represents little if any advance on the basic Neoclassical model. In 1952, W. A. Lewis proposed a special case of the Neoclassical model in which there are unlimited supplies of labour (LM); i.e., over the foreseeable future labour can be drawn into production at constant wage. As the economy expands at a constant wage rate, profits increase disproportionately. If all profits are reinvested, the growth rate rises continuously. When the labour surplus is exhausted, real wages rise and the growth rate slows, for the profit share reached its maximum value and begins to fall. Many problems beset the Lewis model, both theoretically and empirically (Weeks 1973). Perhaps most serious for empirical work, it predicts 1) that the growth rate should decline as countries develop (exhaustion of the surplus labour), in the absence of ex machina interventions such as accelerating technical change; and 2) other things equal, across countries growth rates should be inversely proportional to level of development. There is no compelling evidence for either.12 More promising for policy work is the Harrod-Domar Model (HDM). Like the Lewis Model, it employs a constant marginal Y/K (output-capital ratio), but not the problematical assumptions of constant real wages that yield a rising profit share. The HDM has empirical advantages over the Neoclassical model which are particularly appropriate to developing countries. First, it allows for an empirical differentiation between the potential (warranted) and actual rate of growth; second, and a consequence of this, it is possible to specify it to incorporate the possibility thatThere is a literature on convergence of growth rates across countries, in which some econometric specifications suggest that growth rates are negatively related to the initial level of per capita income. Since the econometric estimations are cross-country, this finding must be viewed sceptically.12


9

growth is demand determined. The warranted growth path derives form the rate of investment, which determines the maximum increase output that can be achieved when capital is fully employed. Whether the warranted path is achieved depends on the level of demand in the economy at any moment; in other words, the economy is demand determined in the HDM as applied to African countries, in contrast to the Neoclassical model in which potential output is achieved through relative price adjustments (price determined). These points are elaborated in the following section of the module. Readings: Dollar, David 1992 Outward-Oriented Developing Economies Really Do Grow More Rapidly: Evidence from 95 LDCs, 1976-85, Economic Development and Cultural Change, 523-544. Fine, Ben 1998 Endogenous Growth Theory: A Critical Assessment, SOAS Department of Economics Working Paper 80, London: http://mercury.soas.ac.uk/economics/workpap/ adobe/wp80.pdf Frankel, Jeffery and David Romer 1999 Does Trade Cause Growth, American Economic Review, 89 (3): 379399 Geda, A. 2002 Finance and Trade in Africa: Modeling Macroeconomic Response in the World Economy Context (London: Palgrave-Macmillan). Kanbur, R. 1999 Income Distribution and Growth, World Bank Working Papers: 98-13, (Washington: The World Bank) Weeks, John 2001 Orthodox and Heterodox Policy for Growth for Africa South of the Sahara, in Terry McKinley (ed.), Growth, Employment, and Poverty in Africa (London: Palgrave)

1.3

Theoretical Framework: Demand Determined Growth

The previous section considered the sources of growth, or the determinants of the potential growth rate. The extent to which an economys long run potential for growth is realised is determined by demand conditions in the short term. There are two broad theoretical approaches to macroeconomic analysis, demand determined and price determined frameworks. A price determined economy is either in a unique full employment general equilibrium, or prevented from achieving that general equilibrium by private or public price distortions. An economy is demand determined when its level of output is limited by one or all of the components of aggregate demand: consumption, private investment, government expenditure, or exports.


10

The theoretical differences between the two frameworks imply fundamental policy differences. Consider the simple case of a closed economy with no public sector that produces only one product (see Weeks 1989). In the price determined framework, all markets clear in an instantaneous process, in which no exchanges occur at prices other than those in the price set which would prevail at full employment general equilibrium (no false trading). In this theoretical circumstance, consumers and producers take prices as signals to determine the quantities they buy and sell. If all markets clear automatically at the unique full employment price set, then it follows that any action by private or public agents to inhibit market adjustment in prices will result in an outcome below full employment. Assuming there to be no private constraints to instantaneous price adjustment (e.g., no market power by private agents), the full employment equilibrium price set is unique, and all markets are fully developed, it follows that the role for public policy is extremely limited. The price determined framework implies that fiscal and monetary policy should be neutral and passive. Fiscal policy would be neutral in that it should not alter the incentives of private agents from making the choices that would yield the general equilibrium price set: 1) taxes should not affect the decision of private agents between income and leisure (e.g., no income tax); 2) neither taxes nor expenditures should affect the relative profitability of commodities (no tariffs, export levies or subsidies); 3) expenditures should not impact on the consumption allocation decisions of households (no sales taxes that discriminate among commodities and no subsidies to commodities); and 4) government should not distort capital markets by competing with private agents for funds (no funding of the fiscal deficit through bond sales). As a practical matter, governments must tax, spend, and sometimes run deficits. The price determined framework accepts this and counsels that the inherently-distorting operations of the public sector should be minimised: levy taxes on a uniform basis (a single tariff rate for all imports, for example); minimise fiscal deficits; and restrict government operations to national security, social services, and general administration It should be noted that an active monetary policy, that involves inflation targeting, for example, is inconsistent with this, the fundamental neoclassical macro framework. In this framework, inflation should be zero, so that nominal and real interest rates are equal to each other and to what is required for the general equilibrium price set.13 Inflation targeting is discussed below. The theoretical basis for the price determined framework is quite weak. It cannot be demonstrated that there is a real world process that ensures the realisation of the full employment price set. Nor can it be demonstrated that the full employment price set is unique. The latter is a quite serious problem, because if the price set is not unique, the concept of distortions is called into question. A distorted outcome is defined inSince the general equilibrium price set is unique, there is only one real interest rate that is not distortionary, which we can call r*. Equally important in terms of formal logic (though of little practice significance), a nominal interest rate above r* would distort the transactions demand for money from its general equilibrium value (it would be too low). As a result, at full employment output, the price level would be too high. As a result, real balances (or real outside wealth) would be too low.13


11

relation to a non-distorted one. If there is more than one non-distorted outcome, one cannot be sure that the prices in an economy with public sector interventions are substantially different from some non-distorted outcomes.14 While price determined systems may seem abstract curiosities, they are the basis for any statement that governments distort the economy. One cannot allege the existence of distortions without simultaneously asserting the existence of a unique non-distorted economy. Consider the apparently simple statement, tariffs distort profitability between importables and exportables. The validity of this statement requires the prior demonstration of the existence of a unique full employment general equilibrium. Since this cannot be demonstrated generally, even in theory, the correct statement would be, tariffs alter profitability between importables and exportables. The practical difference between using the two words, distort or alter, is the core of policy debate. If public sector actions distort the economy, that results in inefficiency that should be avoided or minimised. If the actions alter the economy, then a subjective policy assessment is required to determine whether the alteration is net beneficial to society.15 If one moves from the world of the abstract to the characteristics of African economies, it should be obvious that the price determined framework is not applicable.16 First, during two decades of stabilisation and structural adjustment, most of these economies have been demand determined by high real interest rates, fiscal austerity, or heavy debt burdens, and in some cases all three. Second, many of the economies have suffered from declining terms of trade, which have had a net contractionary effect on external demand. Third, and frequently ignored, major prices in many African economies are not primarily market determined. It is obvious that the nominal interest rate is an administered price if the monetary authorities practice inflation targeting. In addition, in almost every sub-Saharan country official aid flows and debt servicing represent a substantial portion of the balance of payments, and neither is directly exchange rate sensitive. As a result, the value of a floating exchange rate is determined to a great extent by non-market flows. If economies are demand determined, then the existence of a general equilibrium price set becomes a moot point, because relative prices derive from the level of aggregate demand, and change as aggregate demand rises and falls. Therefore, relative prices are not signals to producers and consumers, but result from their production and consumption decisions. Since prices do not determine quantity choices by consumers and producers (they are derivative from them), they are not indicators of efficient allocation. This implies that public sector interventions should be judged on a pragmatic basis in terms of social cost and social benefit. In the caseIf the general equilibrium price set is unique, it is not necessary to know what it is, since by definition any price intervention prevents it from being realised. 15 Subjective assessment is used in the sense that the economics profession defines positive and normative statements. 16 In our view, the slow progress of [structural adjustment programmes] in reviving the HIPC [countries] is related more to the fundamental problem of the theoretical construct than to the weak capacity of African states and institutions in implanting and carrying through structural adjustment to completion (Nissanke and Ferrarini 2004, 47).14


12

of fiscal policy, active macroeconomic interventions are justified to move the economy towards full employment and to foster growth. Taxes and expenditures should be similarly judged, and not by whether they violate non-relevant abstract principles of efficient allocation. The criterion for judgement should be whether taxes and expenditures achieve the goals set by society, and when those goals conflict, an empirical analysis of trade-offs is required. The economic justification of the UNDPs approach to economic policy and poverty reduction is that economies are demand determined. Inequality and poverty would reflect allocative efficiency, so that interventions to promote equity would be distortionary. The role of the government would be to provide safety nets for those unable to benefit from market opportunities. In contrast, the demand determined framework predicts and accounts for unemployment, it concludes some inequalities to be inefficient, and it calls for effective public sector intervention to achieve social objectives. The demand determined framework considers the social objective of sustained growth to be one in which the public sector provides the residual stimulus to maintain the economy at its productive potential. In the short and medium run this involves counter-cyclical policies, and in the long run investment that increases aggregate supply. Thus, a growth fostering policy package that recognises economies to be demand determined would have the following components: 1. an expansionary fiscal budget, consistent with the rule that the overall deficit not exceed public investment; 2. an accommodating monetary policy that tolerates moderate inflation to achieve higher growth, by a) targeting the Golden Rule real interest rate (equal to the sustainable growth rate of per capita income), b) allowing the money supply to expand to accommodate growth and monetary deepening, and c) providing subsidised credit for poverty reduction programmes; and 3. a managed exchange rate regime that seeks to maintain a rate is expansionary by promoting tradeables. The demand determined framework can be applied empirically by employing a partial adjustment model, in which there is a desired level of investment that corresponds to the investment rate of the warranted rate of growth (see annex for details), and an actual rate which partially but incompletely adjusts in each period to that desired rate. Such models were common thirty years ago, but abandoned with the rise of the rational expectations approach to macro modelling (see discussion in Weeks 1989). In effect, rational expectations modelling presumes the economy to be price determined, because of the presumption of instantaneous adjustment to equilibrium. A partial adjustment model is presented in detail in the annex to this module. Its essential elements are a desired rate of investment determined by the capacity to import, export growth, the real exchange rate, and the rate of inflation which along with exports and net government expenditure are the major sources of aggregate demand. These three elements of demand determine the actual growth of the economy.


13

Applying this model to cross-section and time series data for the sub-Saharan countries, Weeks (2001) reached the following conclusions: 1. the statistical significance of real interest rates (negative) and the fiscal balance (negative) provide strong evidence of policy-fostered demand compression; thus, high real interest rates and reduction of fiscal deficits toward zero and into surplus are not sound policies for generating growth; 2. exchange rate devaluation tends to be contractionary, by reducing imports necessary for investment; if devaluation is necessary to improve the trade balance, it should be combined with demand expansion policies to maintain growth, such as increased public investment; and 3. as others have found, there is no evidence that moderate inflation has anyimpact on growth, positive or negative.

On the basis of this theoretical discussion, one can conclude that the GNGM provides a useful tool for considering growth as a purely supply side phenomenon, and it will be used for this purpose at the sectoral level; and, modelling of the actual rate of growth can be done employing a country-specific version of the HDM. With these points in mind, we turn to issues of growth and distribution. Readings: Collier, Paul and Jan W. Gunning 1999 Explaining African Economic Performance, Journal of Economic Literature, 37(March): 64-111. Weeks, John 1989 A Critique of Neoclassical Macroeconomics (London & New York: Macmillan & St Martins)

1.4

Growth, Distribution and Poverty Reduction: Theory & Evidence

1.4.1 Redistribution and the Opportunity Cost of Growth Despite the wide-spread recognition that GDP growth should be combined with mechanisms of redistribution to achieve the international poverty target, one finds little quantitative evaluation of the relative impact of the two poverty determining mechanisms, either in the abstract or for specific countries; i.e. what would be the reduction in poverty for a given rate of growth and a given redistribution? Were this question answered, one could then assess the growth and redistribution mechanisms in light of the resource cost of their poverty reducing impact. To calculate the poverty-reducing impact of growth and redistribution, one can use a simple analytical framework that formulates two abstract possibilities, distribution neutral growth and redistribution with growth:


14

Poverty reduction through distribution-neutral growth (DNG) is trickle down under the most favourable conditions, in that there are equal percentage increases in income across the entire distribution, implying an Index of pro-poor growth of unity; this benchmark can be compared to Poverty reduction through a more equal redistribution of each periods growth increment (redistribution with growth, RWG), implying an IPPG greater than unity. Without a dated poverty target such as mandated by the MDGs, the question of whether growth or redistribution is more effective in reducing poverty would be analytically trivial. If one ignores the practicalities of the redistribution process, and a countrys per capita income lies above the designated poverty line, poverty can be eliminated by a one-off redistribution in any current time period, while per capita growth would take several or many periods to achieve the same result. The imposition of a specific target on the poverty agenda makes comparison policy-relevant. For purposes of presentation, the possibility of growth with a worsening distribution of income is not considered (IPPG < 1). This outcome characterised many if not most of the countries in transition from central planning to market regulation during the 1990s (and China perhaps as early as the 1980s). Clearly, such a growth pattern would reduce the poverty reducing potential of growth. In effect, it represents a dynamic transfer from the poor to the non-poor. In what follows this scenario is noted but not analysed. Its consequence for poverty is clear, and the purpose of the discussion is to consider policy measures whose purpose is to reduce poverty more than would be the case in a poverty-neutral policy framework. The evaluation of the effectiveness of growth and distribution for poverty reduction would be required even were it the case that for the vast majority of countries historical growth rates would achieve the poverty target (see van der Hoeven 2000). Any target growth rate, in this case for poverty reduction, has an opportunity cost in foregone consumption compared to lower rates. This real resource cost can be compared to the cost of achieving the same poverty reduction at a lower growth rate. Economic growth is a means, and raising the rate of economic growth without considering the opportunity cost would be the domestic equivalent of mercantilism. The relevance of the opportunity cost of raising growth rates passes from academic to practical interest because, for the vast majority of countries, maintaining historical growth rates would not be sufficient to meet the international poverty target.17 Table I.1, taken from Hanmer and Nascold (2000) and up-dated, demonstrates the inadequacy of past growth performances for the major developing regions. Only the East Asia and Pacific countries have consistently grown at above the rate necessary to reach the poverty target, though the South Asian countries on average have been quite close to the necessary rates. For the sub-Saharan region, the Middle East and North Africa, and Latin America, both longer term run rates (1985-2000) and growth during 2000-2004 were below17

A discussion of this issue is found in Demery and Walton (1998).


15

what would be required to reach the poverty target with distribution-neutral growth. Performance for the Central and Eastern European countries and central Asian countries would be more difficult to assess. The pre-1990 rates were sufficient, but the post-reform performance was far below target. It is probably the case that some of the Central and Eastern European countries would achieve the growth target, while the central Asian countries could not. Table I.1: Growth rates estimated as necessary to halve poverty by 2015 and Income SharesRegion Item: Region Sub-Sahara ME&NA EAP South Asia LAC EE&CA Per capita growth rates: To meet Actual Actual target 198520002004-15 2000 04** 6.0 -0.6 1.3 2.7 0.4 0.7 3.3 6.1 6.0 3.6 3.4 4.2 7.0 1.1 0.0 4.0 -0.7 2.1 Target minus* Actual 1985-00 6.6 2.3 -2.7 0.2 5.9 4.7 Actual 2000-04 6.6 2.1 -2.7 -0.6 7.0 1.9 Income share, top 20% 52 na 44 40 53 na

Notes: ME&NA Middle East and North Africa, EAP East and the Pacific, LAC Latin America & the Caribbean, EE&CA Eastern Europe and Central Asia *A negative number indicates that the region grew faster than the rate necessary to meet the poverty target. **The rates for 2004 are estimates. Sources: Growth rates, Hanmer & Nascold (2000), World Bank website by region; income share, Deininger & Squire (1996), for the 1990s; and DFID (2000, pp. 16 & 22), where the numbers are reproduced. Similar calculations can be found in World Bank (2000) and World Bank (2001a).

For all the regions the opportunity cost of the target growth rates appears relevant in light of the substantial degree of income inequality (last column of the table). To consider this further, an analytical framework is presented in which growth and redistribution are specified rigorously. In order to do so, it is necessary to simulate the income distribution in each country. In the simple case in which income was equally distributed and per capita income higher than the poverty line, a country would have no poverty (every one receiving the same above-poverty income). As the income distribution becomes more and more unequal, an increasing portion of the population falls below the poverty line. Thus, for two countries with the same per capita income, the one with the greater inequality would have a higher poverty rate. It is this simple proposition that is generalised below. Readings: Demery, L., and M. Walton 1998 Are Poverty and Social Goals for the 21st Century Attainable? IDS Bulletin, 30 Hanmer, L., and F. Nascold 2000 Attaining the International Development Targets: Will growth be enough? Development Policy Review 18, 1, pp. 11-36


16

1.4.2 Comparing and Combining Growth and Redistribution For most countries the income distribution statistics are not detailed enough to estimate the change in poverty when income or distribution changes. However, many countries have summary measures of inequality, such as the Gini coefficient, based on income distribution by quintiles or deciles of the population. A standard approach is to take one of these summary measures and use a distribution function that generates it. That is, the method is to use a simple formula to assign incomes to households from the poorest to the richest. The formula is assigned various parameters (fixed numbers), until it generates the summary measure of inequality. In this exercise we use the simplest distribution function, the Pareto. It has two parameters, per capita income and an exponential inequality coefficient (below we call this coefficient the Greek letter ). The latter determines the increase in income between any two households, for example, from the poorest to the next poorest. If the coefficient were two, the second poorest household would have an income double that of the poorest. For an equal distribution of income, the coefficient would be zero.18 Using this simple function, one can generate a family of curves that show a constant degree of poverty, of decreasing level as they shift to the right, shown in Figure I.3. For each curve inequality is constant, and poverty varies as a result of changes in per capita income. Because of the assumption of a poverty line of one US dollar per day, all the curves begin on the horizontal axis at US$ 365. The diagram clarifies the policy alternatives: redistribution of current income (RCY) involves a vertical (downward) movement, distribution neutral growth (DNG) a horizontal (rightward) shift, and RWG is represented by a vector lying between the two. The diagram also shows the case of increasing inequality growth (IIG), in which the growth of per capita income so worsens the distribution of income that it leaves poverty unchanged (movement along the constant poverty level curve for P = 20 percent). The diagram yields several generalisations. First, because the schedules converge to the left, the impact of redistribution on poverty declines as per capita income declines. At low incomes, both redistribution and redistribution with growth are less effective, relatively to distribution neutral growth. Second, for a given per capita income, the lower the level of inequality, the greater is the impact of redistribution on poverty reduction. In other words, when the poor are clustered close to the poverty line, the

The general form of the function is, Yh = Ah , where Y is the income household number h, so that h is the ordinal rank of all households from 1 to n. The letter A is a scalar, to ensure that (Y1, Y2Yn) = total household income. The inequality coefficient is , which has a lower limit of zero (any number raised to a power of zero is one, so that all households have the same income), and no upper limit.18


17

income transfer necessary to raise them out of poverty is less than if the same number of households were unequally distributed. The growth-distribution interaction on poverty reduction can also be shown for growth rates. In Figure I.4, the percentage reduction in poverty is on the vertical axis and growth rates on the horizontal. Three lines are shown, for increasing degrees of inequality as they rotate clockwise (increasing values of holding initial per capita income constant). The figure shows that for any initial per capita income, growth reduces poverty more, the less the inequality of the initial income distribution. From the initial position at point a, distribution neutral growth increases the rate of poverty reduction along the schedule = 1.3 to point b (an increase in the growth rate with distribution unchanged), redistribution of current income involves a vertical movement to point c, and a shift from a to d is a case of redistribution with growth.Figure I.3:

Relationship between Inequality and Per Capita Income for Constant Levels of Headcount Poverty5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 .5 .0 0N = 40

N = 30

N = 20

IIDNG

Inequality coefficient

PCY = 365RCY RWG

1000

2000

3000

4000

5000

per capita income

Note that using a head count measure of absolute poverty has an inherent bias towards the effectiveness of growth alone (DNG). Assuming all income distributions to be relatively continuous,19 any distribution neutral growth in per capita income, no matter how low, will reduce poverty. However, by defining poverty to be below a certain income level, redistribution reduces poverty only to the extent that it moves a person above that level, in this case above a per capita income of US$ 365. To put the point another way, redistributions that reduce the degree of income poverty for those below the absolute poverty standard do not qualify as poverty reducing.20 Even19 20

That is, we assume there are no gaps in the distribution below and near the poverty line.

A redistribution of one percentage point of GDP from the richest ten percent of the population to the poorest ten percent, equally distributed among the latter, would improve raise the incomes of all those in


18

confronted with this strong condition, we show that simple redistribution rules result in powerful outcomes for poverty reduction. The redistribution we propose, in the Chenery, et. al. (1974) tradition,21 is equal absolute increments across all percentiles, top to bottom. This could be viewed as relatively minimalist, with alternative redistribution rules considerably more progressive. This, a special case of the redistribution with growth strategy, we call equal distribution growth (EDG). Assuming that the absence of a distribution policy implies distribution neutral growth, the proposed equal distribution growth implies income transfers, or an implicit tax. Assuming that per capita income lies above the poverty line (US$ 365 in this case), the redistribution tax would be negative up to the point of mean income (positive income transfer), then positive above (negative income transfer). If income were normally distributed, the tax would be negative through the fiftieth percentile. The proposed marginal redistribution has characteristics that derive automatically from the nature of income distributions. First, and most obvious, the relative benefits of the equal absolute additions to each income percentile increase as one moves down the income distribution. Second, and as a result of the first, for any per capita income, the lower the poverty line, the greater will be the poverty reduction. As a corollary, when a policy distinction is made between degrees of poverty, with different poverty lines, the marginal redistribution will reduce severe poverty more than it reduces less severe poverty. Third, the more unequal the distribution of income below the poverty line, the less is the reduction in poverty for any increase in per capita income, or redistribution of that increase. Readings: Aghion, P.; Caroli, E.; Garcia-Penalosa, C. 1999 Inequality and Economic Growth: The Perspective of the New Growth Theories, Journal of Economic Literature, Vol. XXXVII, December, pp. 1615-1660 Dagdeviren, Hulya, Rolph van der Hoeven and John Weeks 2002 Poverty with Growth and Redistribution, Development and Change 33, 3, pp. 383-413 Dollar, David, and Aart Kray 2000 Growth is Good for the Poor, (www.worldbank.org/research: World Bank) Goudie, Andrew, and Paul Ladd 1999 Economic Growth, Poverty and Inequality, Journal of International Development, 11, 2, pp. 177-195 Lubker, Malte, Graham Smith and John Weeks 2002 Growth and the Poor: A comment on Dollar and Kraay, Journal of International Development 14, 5, pp. 555-571

the lowest decile, but might shift none of them above the poverty line. 21 This volume was path breaking, in that it focused World Bank policy on strategies of poverty reduction. Particularly important were two papers by Ahluwalia (1974a and 1974b), and by Ahluwalia and Chenery (1974a and 1974b). A good review of the distribution literature of the 1960s and 1970s is found in Fields (1980).


19

Box I.2: Redistribution and IncentivesWhen redistribution of income or assets is proposed, the wisdom of such a policy is commonly called into question on the grounds that it would have a negative effect on the incentive to work and invest. With regard to working effort, the analytical basis of this argument is that income provides the inducement to work, and reducing income reduces the motivation to work. There are several reasons why this line of argument should be viewed sceptically. First, if there is a positive relationship between working effort and income, the relationship must be asymptotic. That is, as income rises, the working effort it induces at the margin must decline. The physical limit to the working day, week or year ensures this. Therefore, one would expect that reducing the income of the highest income earners would have little effect on the incentive to work. Second, if it is the case that less income induces less work, than it should be the case that more income induces more work. On the basis of this symmetry, consider the case in which US$ one million is taxed from the richest household in the distribution, and offered to the employed members of the one thousand poorest households in the form of wage increases. If there is symmetry, the wage earners in the one thousand poorest households will have their incentive to work increased, while the one rich household would have its incentive decreased. The net effective would be an increase in the overall incentive to work in the society. Third, the type of redistribution proposed in this module, associated with pro-poor growth, does not involve redistribution of current household income, but changing the distribution of the increment in growth. In principle, no household loses; rather, some win more than others. In other words, every household enjoys an increase in income (assuming there is growth), but the gains are shifted down the distribution. Following the argument in the paragraph above, this should result in a greater increase in the incentive to work than would be the case for distribution neutral growth (because the incentive effect of income increases declines at the margin as income rises). Complementary to the incentives argument is the allegation that redistribution would discourage investment, by creating uncertainly about the business climate. It is not clear why this would result from a more pro-poor growth policy, in which redistribution occurs at the margin. In summary, the incentive arguments against redistribution arise from an earlier era in which redistribution typically referred to radical measures taken to alter current income and asset patterns. It is inappropriate to transfer these uncritically to the pro-poor growth context.


20

Box: I.3: Decomposing Poverty ReductionBetween two moments in time, the change in poverty is the result of changes in aggregate income (growth) and changes in distribution across households. Several measures have been proposed, such as that of Ravallion and Datt (1991, see also Kakwani 1990. Consider two years, t and t+n (such as t+1), and a reference year r.

Pt +1 Pt = G (t , t + 1, r ) + D(t , t + 1, r ) + R(t , t + 1, r )In words, Poverty Change = Growth Component (G) + Redistribution Component (D) + Residual (R) The growth and redistribution components are given by the following equations,

G (t , t + n, r ) P (Z t + n , Lr ) P (Z t , Lr )

D(t , t + n, r ) P(Z r , Lt + n ) P(Z r , Lt )

A residual results if the index used is not additively separable between (mean per capita income) and L (the Lorenz curve). When mean income and the Lorenz curve jointly determine the change in poverty, a residual results. The treatment of the residual in the decomposition exercise involves differences in analytical interpretation. Datt and Ravallion interpret the residual as the difference between the growth and redistribution components evaluated at the terminal and initial Lorenz curves (mean incomes), respectively. In other words, for them the residual is an index number problem, such as arises with measuring inflation over time. The Datt-Ravallion decomposition can be only used when there are comparable estimates of poverty for at least two points in time. Alternatively, one can measure the elasticity of poverty with respect to growth and a summary measure of distribution. The estimated elasticity is sensitive to the specific poverty line used. In most exercises, cross-country regressions are used to obtain the elasticities of poverty with respect to growth and income inequality, as is done in this module in the section on inequality in Africa. See Ravallion 1999.


21

Figure I.4:

Poverty Reduction and GDP Growth for Degrees of Inequality10.0 9.0 8.0% poverty reduction = 1.2

7.0 6.0 5.0 4.0 3.0 2.0 1.0 .0 .0 2.0 4.0 6.0RCY c a RWGd

= 1.3 DNG b = 1.4

8.0

10.0

GDP growth rate

1.4.3 Inequality in sub-Saharan Countries The foregoing discussion of distribution and growth leads to the issue of inequality in Africa. In the 1960s when most sub-Saharan countries received their independence, the consensus of experts was that the countries in the region had relatively low levels of inequality compared to countries in Asia and Latin America. In part this view derived from the absence in all but a few African countries of large landed estates, associated semi-feudal labour relations, and landlessness, as well as the small size of the urban-based capitalist class. Whether this judgement was correct, and there are almost no distributional data for the period to support or refute it, it can no longer be maintained. In the first data column, Table I.2 provides data on distribution in twenty-eight of the forty-eight sub-Saharan countries in the 1990s, listed in descending order of inequality using the Gini coefficient (the statistics in the rest of the table are discussed below).22 Even considering that these twenty-eight may be biased towards the more inequitable countries, the Gini coefficients suggest a quite extraordinary degree of inequality. Figure I.4 demonstrates this by showing the frequency distribution of Gini coefficients for over one hundred countries. The average for the twenty-eight sub22

The table includes all countries with data from World Development Indicators 2004. has more than one entry during the 1990s.

No country


22

Saharan countries was forty-nine (see Table I.2), and for the other seventy-six countries it was thirty-nine. One might consider this an invalid comparison, because Figure I.4 includes developed countries whose Gini coefficients are clustered at the low end of the distribution. Perhaps the more relevant comparison is to point out that fifty-two countries reported Gini coefficients of forty or more, and twenty-three of them were in the sub-Saharan region (the first twenty-three in Table I.2). Thus, the sub-Saharan region represented twenty-six percent of the 104 countries, and accounted for forty-four percent of the countries with Gini values greater than forty. Even more striking, sub-Saharan countries were slightly over half of the countries with Gini coefficients in excess of fifty, and Namibia can claim the dubious distinction of having the highest measured value in the world.A statistical inspection of Table I.2 reveals a strong positive correlation between the value of the Gini coefficient and per capita income;23 i.e., the poorer countries had more equal distributions.. However, there are several reasons to think that this may be a statistical illusion or open to alternative interpretation. First, if one excludes three countries, Namibia, Botswana and South Africa, the correlation becomes non-significant. Second, if one uses a binary variable (dummy variable) for the countries with substantial white populations during the colonial period,24 per capita income is not longer statistically significant. The income variable is also rendered non-significant if a binary variable is used to identify countries in which the petroleum and mineral sectors are overwhelmingly important.

The evidence on inequality in sub-Saharan countries leads to the following conclusions: 1. the region has an extremely unequal distribution of income, and includes countries with the most unequal distributions in the world; 2. the degree of inequality not significantly correlated with level of development; and 3. inequality appears to be correlated with exploitation of minerals and hydrocarbons, and the legacy of white settler regimes. These conclusions imply that inequality represents a formidable barrier to poverty reduction in the region, both in terms of the elasticity of poverty reduction with respect to growth, and raising the growth rate (Rodrik 2001). At the same time, inequality offers possibilities for substantial poverty reduction through policies of redistribution and makes these policies relevant. The next section considers whether there are policies of redistribution that would be feasible for the countries of the subSaharan region. Using the analytical tools presented in the previous sub-section, we can investigate the impact of inequality on poverty reduction in sub-Saharan countries, referring again to Table I.2. By definition, an absolute poverty head count ratio is determined by a countrys per capita income and the size distribution of income (that is, across persons or households). Therefore, with these two statistics one should be able to estimate the share of households in poverty without need of a household level survey. This would require a functional form which exactly generated the size distribution of income of a country, and with per capita income known, the equation would calculate precisely the share of households in poverty. However, actual distributions differ23 24

The adjusted R-square is .29, and the F-statistic 12.0. These countries are Kenya, Mozambique, South Africa, Zambia and Zimbabwe.


23

substantially from those generated by equations, especially at the bottom and top of the distribution. That is, distribution functions tend to estimate inaccurately the incomes of households in the bottom and top ten percent of the distribution. Because the poverty shares in most sub-Saharan countries are so large, inaccuracies at the bottom of the distribution are likely to be relatively unimportant; that is, if the estimating equation can generate the income share of the bottom thirty percent with accuracy, errors at the bottom of the distribution would have minor impact on the poverty estimate. In data column three of Table I.2, head count poverty shares have been estimated using each countrys Gini coefficient and per capita GDP for 2004. The estimating equation is reported at the bottom of the table. Among others, the poverty estimates assume that the Gini is an accurate measure of distribution, that it did not change, and that there is a common distribution function across countries, which varies only by its inequality coefficient. The poverty estimates vary from a low of ten percent for South Africa, to a high of seventy-four percent for Ethiopia. The four middle-income countries in the table (estimates in bold) had an average poverty share of seventeen percent, compared to over fifty percent for the twenty-four low-income countries. The difference distribution makes is shown by a comparison of Malawi and Burundi, which both had per capita GDP of about US$ 150 in 2004. However, with its less unequal distribution, Burundis estimated poverty share was eighteen percentage points below that of Malawi (the Central African Republic and Kenya show the same point). If one assumes no change in distribution, the equation used to estimate poverty in 2004 can also be used to estimate the level of per capita income that would have to be reached in 2015 in order to meet the MDG of reducing head count poverty by half. This statistic is shown in column five, followed by the growth rate during 2004-2015 required to achieve that per capita GDP. The last column, the growth deficit, is the difference between a countrys growth rate during 1994-2003 and the MDG rate for 2004-2015. Only the four middle income countries had growth rates during the previous ten years had growth rate close to the necessary level, Botswana, Namibia, South Africa and Swaziland. For the twenty-four low-income countries, the growth deficit ranged from minus 3.6 for Senegal to almost minus twenty-three percentage points for Ethiopia. The large deficits, over ten percentage points for fourteen of the countries, result from a combination of low or negative growth rates during 19942003, initially low per capita GDP, and high inequality. The effect of low growth rates is obvious, but less obvious is the impact of an initially low level of per capita income. Holding distribution constant, this implies that the required growth rate to reach the MDG target will be high because while the relationship between per capita GDP and the poverty share is linear, it is not homogenous; that is, for a doubling of GDP the poverty share falls by less than half. Ethiopia can be taken as an example (see Figure I.5). With an initial per capita income of just over US$ one hundred, and a poverty share of seventy-four percent, the MDG target calls for reduction of the share to thirty-seven percent. However, it would be numerically impossible for a country to have a per capita GDP of US$ 200, and only thirty-seven percent of the population with annual incomes less than US$


24

365. With an unchanged distribution (Gini = 57), the thirty-seven percent would be reached at a GDP per capita of US$ 1230 (the DNG line in Figure I.5). However, if the Ethiopian income distribution were to incrementally approach that of Burundi (Gini = 33), the poverty share would be halved at GDP per capita of US$ 790, about thirty-five per cent lower. It is also the case that this per capita income could not be reached by 2015, but poverty reduction by the date would be considerably greater than for no change in distribution. This discussion produces an important conclusion: few, if any low-income subSaharan countries are likely to achieve the head count poverty MDG by growth alone. The lowest growth rates necessary to reach the MDG among these countries are those achieved by the East and Southeast Asian miracles, five to seven percent, and a majority are beyond the realm of possibility. This conclusion is strengthened by the absence from Table I.2 of several countries with very low growth rates during 19942003. Achieving the MDG for poverty in the sub-Saharan region requires policies to make growth more pro-poor as defined at the outset of this module. The next section considers policies to do this. Readings: Cornia, G. A. 1999 Liberalization, Globalization and Income Distribution, WIDER Working Paper Series, No. 157, March 1999 Cornia, G. A., and S. Reddy 1999 The Impact of Adjustment Related Social Funds on Distribution and Poverty, WIDER Project Meeting on Rising Income Inequality and Poverty Reduction, 16-18 July 1999, Helsinki Geda, A. and Abebe Shimeless 2004 Trade Liberalization, Poverty and Inequality in Africa: A Survey of the Theoretical and Empirical Literature (forthcoming, University of Gotenberg, Department of Economics Working Paper) Geda, A. 2004 Openness, Inequality and Poverty in Africa: Exploring the Role of Global Interdependence, Background Paper Prepared for UN International Forum for Social Development on the issue of Social Development Equity, Inequality and Interdependence, June 2004


25

Figure I.4:

Distribution of Gini Coefficients, 104 Countries, 1990s30 25 20 15 10 5 0 25 30 35 40 45 50 55 60 65 70Gini coefficient 52 countries < 40 mean = 41.3 median = 39.8

52 countries > 40


26

Table I.2: Estimation of Distribution Neutral Growth Rates Necessary to Reduce US$ 1 Dollar a Day Head Count Poverty by Half by 2015 in 28 African Countries Required PCY annual Annual Gini: Calculated for .5 Growth Growth Country Year Value HCPS* 2004* Poverty 2004-15 Deficit** Namibia 1993 70.7 2608 3125 3.3 -1.8 19 Botswana 1993 63.0 4893 5100 3.7 -.2 12 CAR 1993 61.3 71 323 1402 14.3 -14.8 South Africa 1995 59.3 4295 4337 .1 .9 10 Swaziland 1994 59.3 1586 2001 2.1 -1.7 28 Ethiopia 2000 57.2 74 110 1230 24.5 -22.8 Zimbabwe 1995 56.8 59 563 1450 9.0 -10.3 Lesotho 1995 56.0 56 608 1459 8.3 -6.4 Zambia 1998 52.6 58 435 1318 10.6 -11.0 Nigeria 1997 50.6 62 276 1206 14.3 -13.7 Niger 1995 50.5 63 218 1176 16.6 -16.3 Mali 1994 50.5 59 359 1247 12.0 -8.9 Malawi 1997 50.3 65 157 1142 19.8 -19.3 Burkina Faso 1998 48.2 60 250 1155 14.9 -13.1 Gambia 1998 47.8 55 372 1209 11.3 -10.9 Cameroon 1996 47.7 43 750 1397 5.8 -4.2 Guinea-Bissau 1993 47.0 61 187 1104 17.5 -19.4 Madagascar* 1999 46.0 57 271 1130 13.9 -13.0 Kenya 1997 44.5 53 328 1134 11.9 -12.3 Senegal 1995 41.3 38 683 1260 5.7 -3.6 Guinea 1994 40.3 38 662 1234 5.8 -4.3 Ghana 1999 39.6 45 430 1106 9.0 -6.7 Mozambique 1997 39.6 51 232 1008 14.3 -8.8 Tanzania 1993 38.2 50 225 981 14.3 -12.2 Uganda 1996 37.4 43 399 1056 9.2 -5.6 Mauritania 1995 37.3 39 535 1122 7.0 -5.5 Cote d'Ivoire 1995 36.7 33 694 1192 5.0 -5.8 Burundi 1998 33.3 47 150 865 17.3 -20.1 Average, all 48.7 48 848 1541 10.8 -9.7 17 4 Middle Income 63.1 3345 3641 2.3 -.7 53 24 Low Income 46.3 384 1191 12.2 -11.2Notes: * HCPS Head count poverty share of the population. This is estimated from a cross section regression using data set in Dagdeviren, van der Hoeven & Weeks (2002). The same equation is used to estimate the per capita GDP at which poverty would be reduced by half (assuming distribution neutral growth). The details of the estimating equation are as follows, with the level of significance of coefficients in parenthesis: HCPS = 16.68 - .03(PCY) + 1.07(Gini) (.07) (.00) (.00)

Adjusted R-square = .788 F = 73.27 (.00) Degrees of Freedom = 35 **The growth deficit is the difference between each countrys growth rate for 1994-2003 and the distribution neutral rate necessary to reach the level of per capita income to reduce poverty by half. Sources: Gini coefficients, base year per capita GDP and growth rates are from World Development Indicators 2004.


27

Figure I.5:

Growth and Redistribution: Example of Ethiopia80 70HC Poverty ShareRWGDNG

PCY = 1230 Gini = 57

60 50 40 30 20 10 0

Gini declines from 57 to 33

PCY = US$ 790

Note: PCY is per capita income.

105

225

375

525

per capita GDP

675

825

975

1125

1275

1.5

Policies for an Effective Redistribution with Growth

The major element required to introduce and effectively implement a re-distributive strategy in any country is the construction of a broad political coalition for poverty reduction (see Bell 1974). The task of this coalition would be the formidable one of pressuring governments for redistribution policies, while neutralising opposition to those policies from groups whose self-interest rests with the status quo. How such a political coalition might come about is specific to each country and its discussion beyond the scope of this module. What this module can do is to identify policies that would present options for governments considering redistribution measures. First, the question of policy effectiveness should be considered on the macro level, by returning to the question raised above: what are the opportunity costs of reducing poverty by increasing the growth rate and implementing redistribution? The opportunity cost of implementation will be determined by the specifics of the programme to achieve redistribution, the size of the redistribution, and the administrative capacity of the public sector. None of these can be determined in the abstract. However, the opportunity cost of raising the growth rate can be quantified within broad limits. The consumption foregone to achieve any growth rate y is determined by the Harrod-Domar warranted rate of growth, y = sv, where s is the net saving rate and v is the output-capital ratio. The opportunity cost of lowering poverty through growth alone can be indicated using the calculations for the sub-Saharan countries in Table I.2.25 Calculations show that a distribution neutral growth rate ofThe calculation is based on using the equation, Yh = Ah , explained in a footnote above. This equation can be solved for the income of the percentile in which the poverty line falls, and this implies25


28

one percent reduces poverty by .30 percentage points, while equal distribution growth would achieve the same reduction with a growth rate of .25 percentage points; that is, equal distribution growth could achieve the same poverty reduction as distribution neutral growth with a growth rate only one-quarter of that required by the former. To double the poverty reduction through distribution neutral growth would require an increase of the saving rate equal to the capital-output ratio. If the capital-output ratio is approximately four, then increasing the annual rate of poverty reduction by one percentage point calls for an increase in the saving rate of four percentage points. Equal distribution growth would achieve the same poverty reduction with one percentage point increase in the saving rate. The difference in the required changes in the saving rate implies that equal distribution growth would have a substantially lower opportunity cost of poverty reduction (three percentage points of GDP). To summarise: 1) The opportunity cost of lowering poverty through growth alone rises with per capita income; and 2) The likelihood that the administrative costs of redistribution would render EDG as or more expensive than DNG decreases with per capita income because the capital out-put ratio raises. 3) Arguments that assert that redistribution would be too expensive appear unfounded when one considers the opportunity cost of reducing poverty through growth alone.26 Turning to specific policies for redistribution, perhaps the most important determinant of the effectiveness of the various measures of each redistribution strategy is the structure of an economy. This structure depends on the level of development, which will to a great extent condition the countrys production mix, the endowments of socio-economic groups, the remuneration to factors, direct and indirect taxes on income and assets, prices paid for goods and services, and transfer payments. These elements of the distribution system are initial conditions that delineate the scope for redistributive policies. The implementation requirements of redistributive policies can be summarised in a simple algebraic framework (see Hanmer et.al. 1997). We use the following terms: Y denotes the income of a household, V is transfer payments, T is taxes, k is a vector of assets, w is a vector of rates of return (including wages), p is the price vector of goods and services, q is the quantity vector of those goods and services, and S is household saving. The discussion which follows will refer to the secondary distribution of income, which is the distribution after taxes (after tax income), and the tertiary distribution,

26 For any particular programme, administrative costs would have to be carefully calculated and compared to those of alternative policies. There is relatively little work on this topic. For a case study, see Grosh (1995).

the headcount ratio.


29

which we define as the distribution after all expenditures by government, including transfer payments.27 By definition, Househol d income = Househol d income = Y= all income govt transfers minus taxes (V T) Transfer payments (unemploym ent compensatio n, pensions, child benefits, aid to disabled) & progressive taxes (on income and wealth) Effective in middleincome countries sources = expenditures household expenditures pq Subsidies for basic needs goods, public sector infrastructure investment (p); child nutrition programmes (q) + saving plus household saving

plus = wages, selfemployed income, dividends, etc. + wk = Minimum wages, lowwage subsidies, other labour market regulations, public employment schemes (w); credit programmes for the poor; land reform, education (k); Effective in middle-income and some lowincome countries

+ S Facilitate asset acquisition: village banks & other financial services for the poor

Effective in most countries

Effective in most countries

The effectiveness of tax and expenditure policies (V and T) To generate secondary and tertiary distributions more equitable than the primary distribution depends in part upon the relative importance of the formal sector.28 This is because governments can most effectively apply progressive income taxes to wage employees and corporations. All empirical evidence shows that the formal sector wage bill and profit shares increase with the level of development. Along with the importance of the formal sector goes urbanisation, and working-poor urbanThis is a broad definition of the tertiary distribution, for it includes all services in kind, such as education and health services. More narrowly, the tertiary distribution is defined as income after taxes and cash transfer payments. A transfer payment is defined as a payment for which there is no productive employment associated. 28 For a review of fiscal policies for redistribution, see Chu, Davoodi and Gupta 1999).27


30

households are more easily targeted than either the rural poor or urban informal sector households. The experience of a number of middle-income countries has demonstrated the effectiveness of basic income payments for poverty reduction, with an example being the pension paid to the elderly in South Africa.29 A redistribution strategy based on taxation and transfer payments is most appropriate for middle-income countries, first because the capacity to tax tends to raise with income, and second, because the ability to make transfer payments increases with the size of the public sector. There are sub-Saharan countries that despite low per capita incomes have a disproportionately high capacity to tax due to mineral wealth extracted by large corporations. However, even in these countries the small size of the formal sector limits the ability to use transfer payments as an instrument of redistribution. A few low income countries, Vietnam being an example, have the administrative capacity for use of transfer payments effectively,30 but there are no examples in the sub-Saharan region. Interventions to change the distribution of earned income (wk in the equation above), Mechanisms to affect the distribution between wages and profits will also tend to be more effective in middle-income countries (ILO 1992). The most common intervention is a minimum wage, though there are many other policies to improve earnings from work (see Rogers 1995). Further mechanisms include public employment schemes and tax subsidies to enterprises to hire low-wage labour. It is unlikely that any of these would be effective in low-income countries, because of enforcement problems, targeting difficulties, and narrowness of impact. These measures apply to countries with a large formal sector, and of limited relevance to sub-Saharan countries. Land reform The relationship between land redistribution and level of development is a complex one. On the one hand, low-income countries are predominantly rural, so if land ownership is concentrated, its redistribution could have a substantial impact on poverty. The more underdeveloped a country, the less commercialised tend to be poor rural households. Therefore, the benefits to the poor from land redistribution in lowincome countries are less likely to be contingent on support services. On the other hand, lack of administrative capacity and so-called traditional tenure systems represent substantial constraints to land redistribution, especially in the sub-Saharan countries. The usual approach to land redistribution presupposes that it is legally clear from whom the land will be taken and to whom it will be given. There are few subSaharan countries in which private ownership is widespread, making redistribution difficult or impossible without prior clarification of ownership claims (Platteau 1992, 1995). In Ethiopia a radical land redistribution was carried out in the 1970s, but itsWhile relatively low, the pension in the 1990s was an important income source for the both urban and rural poor, especially for female-headed households (see Standing, Sender and Weeks 1996, Chap 6). 30 Rural Vietnam, especially in the North, has had an effective administrative system down to the local level for several hundred years, as a result of the manner in which rice production has been organised.29


31

association with repressive economic and social policies has undermined benefits it generated. Experience in Latin America has shown that governments can effectively implement land redistribution, though subsequent poverty reduction in the region was dependent on provision of rural support services that were not delivered (Thiesenhusen 1989), including agricultural extension and marketing facilities.


32

Box I.4: Land Reform in RwandaHistorically land has been an issue of contention in Rwanda. It has been centralised in the hands of various elites, which have fluctuated between Hutu and Tutsi, who have manipulated it to establish population control. This has included weekly volunteer work by all Hutu citizens in exchange for access to land, or more recently, politicising land through the denomination of the Tutsi minority to retain Hutu power. The last decade in Rwanda brought with it the recognition from the international community, the Rwandan government and civil society that land reform was crucial for social stability, both to settle historical disputes and initiate economic growth. With a poverty head count of 71 percent, and a Gini co-efficient of 59.4, Rwanda is ranked as one of the poorest country in Africa (Verpoorten and Berlage, 2007, 12). The National Unity and Reconciliation Council found that land disputes are the greatest factor hindering sustainable peace (Musahara and Higgins, 2004, 6). Land inequality and high density characterise land patterns in Rwanda. In the mid-2000s, it had about eight million people and a land density of 350 people per square kilometre (Musahara, 2001, p. 8), which has been exacerbated by the influx of refugees since 1994. At the beginning of this century, 35 percent of rural households held one half hectare of land or less, and 6.6 percent of households farmed three hectares or more, with the latter accounting for almost one third of total arable land (Newbury and Baldwin, 2000, p.7). Until February 1994, land concentration was exacerbated by the lack of a clear legal land law and or administration. Land was ruled by customary law, which varied by region. The occupier had usufruct rights, but legally the state owned the land. Expatriates and church missionaries were the few to hold land with formal title. The Statutory Order no. 09/76 of March 1976 remained the only written law in place regarding land. It sought to avoid the development of a land market, and was successful in this regard (Mushasara and Higgins, 2004, 2). The new government began changing the land system in 1996. In 1997 the first concrete study on land reform, funded by the FAO, was carried out (Musahara and Higgins, 2004, 2). The essence of the new land policy would be to provide tenure security to households (Rurangwa, 2004, 1). After nine years and many parliamentary select committees, a reform bill passed in February of 2004. It stated that, 1) households should consolidate plots to ensure that each holding is not less than 1 hectare, 2) there would be a ceiling of 50 hectares for each holding, 3) all land would be registered, 4) tenure titles would be tradable, and 5) local communities would be involved in allocating titles. Among the more controversial elements of this bill is the policy of imidugudu, described by some as forced villagisation. A provision of the bill states that no one could resettle in an area unless approved by the government, and no areas could be approved for living outside of settlement communities. The government argued that this would mediate disputes, provide security from both physical violence and hunger, and begin the healing process of ethnic reintegration. In a

UNECA Macro Training 4[1]

Documents

Transcript of UNECA Macro Training 4[1]