Inequality , Poverty , and Employment : What We Know

D'enormes disparites economiques existent au sein des pays et plus encore dans le monde entier. Les inegalites sont beaucoup trop elevees dans la plupart des pays en developpement et en moyenne (ponderation selon la population) elles ont augmente depuis les dernieres decennies. Cela engendre d'enormes couts sociaux, en parce que cela generalise la pauvrete et en partie par d'autres mecanismes sociaux et psychologiques. L’article met l'accent sur les causes de ces ecarts et tente de determiner dans quelle mesure ils pourraient se justifier pour des raisons economiques ou autres, tout en s’interessant aux mesures que les gouvernements peuvent prendre pour reduire ces ecarts. L'analyse economique suggere un ensemble de politiques, y compris (surtout dans les pays les plus agricoles) la reforme agraire ou d'autres mesures pour assurer une repartition equitable des terres, avec l'appui du public pour accroitre la productivite dans les petites unites familiales agricoles, une politique de l'education qui soit forte et equitable, une politique industrielle offrant des incitations adequates pour les activites a forte intensite de main-d'œuvre, y compris le soutien pour les petites entreprises hors du secteur agricole, de meme qu’un ensemble de politiques visant a diminuer les prejuges anti-pauvres de certains marches et de certaines formes d'intervention publique (par exemple, contraintes administratives) contre les pauvres. Enfin, on pourrait ajouter un systeme fiscal approprie et equitable, les transferts sociaux et les biens publics qui reduisent l'inegalite tout en ne reduisant pas trop considerablement les incitations a travailler ou a creer une entreprise. Un ensemble integre de politiques visant a promouvoir la demande de travail, cle de l'augmentation des revenus sur une large echelle, serait base sur ce type de politiques. Malheureusement, la politique sans doute la plus pertinente et efficace, soit la politique foncee sur les reformes politiques touchant l’affectation des terres et les systemes de soutien a la petite agriculture, une politique de soutien actif a la demande de main-d'œuvre, et un Etat fortement redistributif sont politiquement intenable dans la plupart des pays, et par ailleurs difficiles a atteindre pour des raisons techniques, mais aussi politiques.

Introduction: The Reality and the Challenge 1 Inequality and poverty have been and remain dramatic features of most of the world's dominant societies over their histories.They show up in many forms and bring many ill effects.From a policy perspective it is important to know what factors generate such levels of inequality and poverty, whether inequality brings benefits (say faster growth, as often argued) and what policy instruments offer the best hope to lower inequality and poverty without causing loss on some other front.To shed light on these matters it is helpful to review the historical experience both of the now industrial countries and the developing ones for evidence they may provide on the settings and policies that can foster equality and poverty reduction as well as to put forward a simple analytical framework for thinking about them.

2
It is useful to begin with a feel for recent 1 levels of inequality and poverty, which can be gleaned from figures like the following: 1.In 2001 the top decile (10%) of Brazilians, Paraguayans and South Africans collected 50-55% of their respective national incomes, giving them on average about 40-60 times that of the bottom decile who collected around one percent, 2 and the top 1% probably received about 10-13%, making their average income 100-130 times that of the bottom decile.(Graph 1 shows the income shares of each decile in Brazil for the year 2000).In Sweden the top decile (2000) received 22% of all income and the bottom decile 3.6%, for a ratio of 6.2 to one.
2. In 2000 the top decile of the world's population collected a little over half of total world income, giving them on average about 70-75 times that of the bottom decile, who collected less than three-quarters of one percent of the total (Berry and Serieux, 2007, 81); the top 1% probably received about 10-15% of the total, making their average income 133-200 times that of the bottom decile.
3. Expressed in US purchasing power parity dollars of 2000 the average income of the world's top decile in that year was about $28,000 while that of the bottom decile was about $400-500.
4. At the very top, the 1,153 people (or sometimes families) currently classified as billionaires have a total of about 4,245 billion dollars in wealth.As of 2000, the top decile owned about 71% of world wealth, the top 1% had 31.6% and the bottom decile had 0.1% (Davies et al, 2011, 245).
5. Adopting poverty lines of 500 and 1000 dollars (of 1985) per person per day (approximately the levels commonly used to define extreme poverty and poverty), the share of the world's population in poverty as of 2000 was about 28% and that in extreme poverty was 12%.The latter figure was over 19% in South Asia and about 43% in Sub-Saharan Africa.
6.The poverty gap (the share of national income that would have to be transferred to the poor to pull them all out of poverty varies widely by country.Thus in a low income country like Ethiopia and with a poverty line of $1.25 per day that ratio was 28.9% in 2005, whereas in Brazil it was 7.5%.
7. In countries with high levels of inequality, ending poverty is in purely economic terms something that can be done quite cheaply.Thus if the top decile has 50% of all income and the bottom two deciles are below the poverty line and have between them 2.5% of total income, a doubling of the income of this latter group -pulling most of them out of poverty, would reduce that of the top decile by just 5%, a cut that could be won back in just two years in an economy with per capita income growing at 2.5% per year. 3 Most people consider these levels of inequality and the associated poverty to be unjust, especially when those at the bottom of the income hierarchy suffer from such obviously harmful effects as malnutrition, inadequate housing, and illiteracy, which could be eliminated if the incomes of the rich were a little lower.Most people also accept that a degree of income inequality is appropriate in reflection of the fact that some people work harder than others and some do more important work than others (the only doctor in the town) so it is fair that they be recompensed in accordance with these differences.Others argue that hard work and preparation for important jobs may not be forthcoming in the same degree if not differentially compensated.What is fair recompense for harder or more important work is clearly a subjective matter; how much the work effort and the preparation for important jobs is affected by income differentials in favour of those activities is at least in part an empirical question.Neither of these issues will be dealt with in detail here, on the grounds that there are no serious arguments to justify the sort of income differentials referred to above.Gaps of 3:1 or 5:1 are probably justifiable (on the former grounds) and possibly necessary (on the latter) but gaps of 50 or 100:1 are not; instead they often come at a high cost to overall societal welfare.Economic policy should for the foreseeable future have their reduction as a goal, to be pursued simultaneously with other objectives.
What accounts for the enormous gaps between the better and the less well-off within countries and in the world as a whole?Are they in any sense justifiable?What steps can a government take to reduce the gap and are those steps likely to have negative or positive side-effects on growth or other goals?
How much inequality can be justified depends, among other things, on how it is that people wind up rich or poor, in particular on what share of wealth has been acquired by less than acceptable means.Many of the rich are where they are through some combination of a head start (wealthy family background), luck (in financial investments, for example), good connections (e.g.nepotism) or dishonesty and/or forceful theft or appropriation of resources from others less well placed to defend themselves.Most of the poor have lacked the head start and the good luck and some have been the victims of the dispossession of assets by aggression which fuels the riches of some of the former group.Although these unjust or undesirable routes to wealth are, of course, only part of the story, many careful students are surprised to find how significant a role they play and how little economic "justice" there is in their own societies.The illicit appropriators, as would be expected, tend to cover up well, may not tell their children how they accumulated their wealth, hire top lawyers to defend themselves within the judicial system, and often burnish their reputations by engaging in charities and other positive activities.A second feature of the rich is that much or most of their total income comes from capital.At best, that income is payment for savings, innovation, risk-taking and effective stewardship of resources under their control-the story emphasized by the classical economists and, among development economists, highlighted by Lewis (1955).At worst, a considerable share comes from illicit activities, the exercise of monopoly power (usually legal, but not in the social interest), speculative gains (often not in the social interest) and so on.
The "lacks" that can leave one at the bottom of the income hierarchy begin with a low endowment of income-earning assets (land, other forms of capital, labour, skills).Such lacks often characterize certain groups disproportionately.Poverty is also associated with ethnicity (where colonized or otherwise less powerful groups are under the control of powerful ones), with gender, with age (older people have less income and economic clout), and with various types of disability.Frequently it is the interaction between two features that creates serious vulnerability to poverty, as in the case of female members of an ethnic group which is both marginalized by the society as a whole and machistic.
A normal accompaniment of low average incomes is economic insecurity-the risk that due to factors beyond one's control (a bad harvest, a sickness in the family, a loss of market for ones goods or services) one will sink into poverty or fall deeper into it.Since the poor tend to have little chance to accumulate assets that could tide them over such downturns, a loss of current income is translated quickly into a crisis.The economics of inequality and poverty, and the possibly useful policy responses to them, thus involve also dealing with economic insecurity.

8
The ultimate relevance of inequality (like that of income) is its impact on human welfare, which implies that it should be seen as an intermediate rather than a final goal.Measurement of such other aspects of inequality as opportunities (favoured by Sen, 1977), discrimination, or failure to be appreciated (Shaffer, 1999) is by nature more difficult, and has received correspondingly less attention.An important question is the degree of correlation between income or consumption (the usual economic indicators of economic welfare and poverty) and these other aspects of welfare; only if it is very strong can we assume that policies that address income and consumption inequality will also assure benefits in those other elements of personal and group welfare.

9
Many aspects of the quality of life are in fact highly correlated with income and wealth, including education, health, and the quality of goods consumed.This is true in part because money buys the private goods people consume and in part because governments are often under the control of rich elites and thus favour these groups in the way they provide public goods.Other potentially important indirect effects of inequality involve the character of the society, in such dimensions as instability, violence and crime, which can form part of the vicious circle within which inequality and its correlates interact.Another important correlation is between poverty and either not having a job or having an unsatisfactory one; both feed directly into low self-esteem.Improving access to decent employment thus constitutes a mechanism to facilitate group identity and inclusion.This, together with the income generated and the self-esteem implicit in being able to carry out some socially valued activity, helps to explain why employment emerges as so important to so many people.
10 Inequality and poverty have many direct social and psychological effects.Because relative income or consumption is a major determinant of welfare for many people (Easterlin, 1974), income inequality can be seen as a form of economic exclusion.But since exclusion takes many other forms, inequality and exclusion are by no means the same thing.For example, each member of a very inegalitarian society could nonetheless feel a strong sense of inclusion within her caste or social class."Psychological inclusion" -the state of feeling part of a social group (as small as the nuclear family or a group of intimate friends or as large as the nation or the world) in ways that are conducive to personal wellbeing can be experienced either at the individual or the group level.Psychologists are unambiguous in their view that group identity is central to the lives of most people.Unfortunately, one person's inclusion in a group is often the other side of the coin from another's exclusion.The overall societal implications of groups capable of providing a sense of personal or collective belonging thus depend on whether and to what degree that sense is defined in reference to an out-group or even an enemy.A key mechanism of group exclusion-often referred to as structural inequality, is the attribution of an inferior status to a category of people.It is especially severe when the stronger group also dominates state institutions and is reproduced through a wide range of mechanisms, among which ethnicity plays a major role.Important to the role of economic structure in the various forms of inclusion/exclusion is their links to economic inequality, as where poverty prevents a person or group from getting access to a public service supposedly available to everyone or to poor people in particular; the more important are these links the more likely it is that raising the relative income of currently excluded groups can weaken the mechanisms and effects of that exclusion.
Inclusiveness can affect economic performance and individual welfare though various types of cooperation.For example, worker cohesion often raises group productivity (Hall and Jones, 1999).Alternatively, association and interaction with others helps build social networks that may provide support when it is needed. 3Many mutual support networks are horizontal, involving people of roughly the same socio-economic status.Very low income families are often precluded from participation in networks because they cannot reciprocate; at the limit their members may have only each other to rely on for support in difficult times (Pahl, 1984).
A worrisome implication of the fact that relative income (or consumption) matters a great deal to most people's sense of wellbeing is that economic growth may contribute rather little to human satisfaction.This fact has immense implications both for growth policy per se and for the environment.If the environment is being despoiled to achieve growth that has only a modest societal payoff anyway, it should be slowed (perhaps even stopped) until societies have found ways to benefit more from it.The importance of relative income to welfare also poses a number of challenges for social policy, including: (i) how can the psychological costs of social exclusion-a phenomenon related in part to individual status-seeking behaviour, be curtailed through policies to foster social inclusion?; (ii) how can people be made less envious, status conscious and otherwise unable to be satisfied except when they feel they are somehow doing better than others?; and (iii) does reducing economic inequality help to alleviate these social problems as well?
Two practical questions relating to the challenge of reducing inequality are (i) whether market forces can be expected eventually to lower inequality in the normal course of events, as Kuznets (1955) proposed, or whether they will tend to work in the opposite direction, and (ii) whether the political economy of inequality constitutes such a powerful force against change in highly unequal societies that not much is likely to happen.Lindert (2000, 208) refers to this as the "Robin Hood paradox" -the fact that in those times and polities where redistribution would be most warranted it happens the least.
Analysis and policy-making with respect to inequality can benefit from two types of information.One is knowing the details of how economies function, (such as how the labour market works), and the cross-section evidence on what features of an economy are correlated with inequality at a point of time.A second is the historical record on inequality in the developing countries, which however suffers from two great weaknesses: (i) the data on capital income and hence on those people at the very top of the income hierarchy are very incomplete and (ii) there are thus far no confirmed examples of major declines in inequality in market economies.As a result, the record of the now industrial countries, which suffers from neither of those weaknesses, takes on more importance.Among the developing countries there are many examples of increasing inequality, especially over the last two or three decades, though in a number of Latin American countries recent evidence reveals declines in the inequality of reported income (made up mainly of labour income and transfers but typically missing the bulk of the capital income of richer people).

How Inequality Changes Over Time
The first notable feature of the pattern of inequality over time is its remarkable inertia under all but very extreme circumstances.Behind this pattern of relative constancy over time are two underlying types of stability.One involves the major determinants of inequality (discussed in more detail below), especially the distribution of factors of production; the distribution of land, physical capital and human capital never change quickly, absent some extreme event; in the case of human capital even an extreme event does not affect it much.Second, the policies and the politics and administrative systems that underlie them also tend to change gradually-there is a heavy dose of path dependency in this respect as well.

The Record of the Industrial Countries
The overall trajectory of pre-fisc inequality (inequality prior to the effects of taxes and public spending) has been strikingly similar across the now industrial countries, typically involving a pre-20 th century increase, a decline during some or much of the 20 th century and a levelling off or increase during the last several decades.The 20 th century upsurge in transfer payments (like pensions) and in social spending more generally has meant that post-fisc inequality has fallen more than its pre-fisc counterpart, and that the former is probably now less in all or nearly all countries than was the case a century or more ago.
At their peaks, both pre and post-fisc inequality appear to have been very high in most of these now industrial countries -comparable say to levels in some of today's Latin American nations.Given that, far back enough in time, agriculture was the major economic sector, the distribution of land was necessarily a major determinant of income inequality, just as it has been in the history of today's developing countries.Another frequently significant source of inequality in the 18 th century and much of the 19 th was the ability of well-placed groups to derive rents from an institutional framework that biased markets in their favour; 4 by contrast, the wages of the unskilled were set in a competitive market.Many of the guaranteed incomes benefiting the middle class or the well to do in the 18 th century had disappeared by 1970 in the industrial countries (Morrisson, 2000, 252).The state had by then intervened on behalf of the lower strata to fix a minimum wage, provide unemployment compensation, etc.The concentration of human and physical capital decreased substantially over the 19 th and 20 th centuries due to the diffusion of education and of savings as incomes rose.There has also been increasing access to family housing ownership.
Although timing and pattern varied, the first half of the 20 th century saw a significant net reduction in pre-fisc inequality in all of the industrial countries, and a sharper fall in post-fisc inequality.Notable declines in the pre-fisc income shares of those at the top were usually linked to the trauma of war or economic crisis -in all cases, inequality fell either during or after each World War (Piketty and Saez, 2006).Meanwhile, both tax and social spending patterns in the industrial countries went through dramatic alterations over the 20 th century, leaving the post-fisc distribution of (disposable) income quite different from the pre-fisc one in nearly all cases.Typically the greater redistributive effect came from the spending side.By the end of the 18 th century no country had even 3% of its GDP devoted to redistributive social programs; now such programs claim over a third of GDP in many countries.Until nearly the end of the 19 th century there were only two forms of social spending 5 -poor relief and public schools.The rise in social spending accelerated between 1880 and WW11 and then boomed between WW11 and about 1980, after which its share of GDP has levelled off.The World Wars and the Great Depression pushed continental Europe toward progressive taxation and the expansion of social programs (Lindert, 2004, 11, 15); popular recognition of the fragility of economic fortunes led to Roosevelt's programs in the US, the 1942 Beveridge Report in the UK, and the victory of the Social Democrats in Sweden.At the end of WWII the threat of communism frightened both the Church and the Christian Democratic parties into acceptance of social democracy.The big increase in social spending took place in the 1960s and 1970s.By 1995 social transfers ranged from 33% of GDP in Sweden down to 12.2% in Japan with public education adding another 4-6% in most countries (Lindert, 2004, 177-8).Post-fisc inequality was also lowered by more progressive tax systems, usually built around the income tax.Before WW1 this effect was absent or unimportant, but by the 1920s and 1930s it was significant in Scandinavian countries and the UK.After the Second War political factors greatly raised tax progressivity, with the British tax structure being copied in a number of European countries.Still, the welfare state with its greatly increased public social expenditures lowered inequality much more than did the tax system.
The early evolution of the social spending and tax systems is, according to Lindert, explained by a few key factors, including (i) the extension of the franchise (to women and to men in the lower part of the socio-economic hierarchy and (ii) decentralization, which allowed local government greater freedom to fund in areas like public education. 6In various countries, including the United States, localities raised most of the taxes for their schools.
Among now industrial countries, the experience of Japan is of particular interest.Its large decline in inequality, entirely concentrated in the Great Depression and World War II years, saw capital owners sustain severe shocks to their assets and permanent declines in their shares of wealth and national income.Japan then soon embarked on a dramatic post-war burst of growth which allowed it to catch and surpass many of the earlier industrializing countries.Piketty and Saez' (2006) explanation for the failure of wealth to reconcentrate in a few hands is the presence of progressive income and estate taxation after the initial shocks occurred, a view consistent with the fact that in Switzerland, which neither went through the stress of war nor instituted very progressive taxes, the wealth share of the top 1% has remained in the 35-40% range since 1915, and that in Ireland, which did not suffer the severe shocks of the wars but did institute progressive taxes, top income shares did (with a lag) fall.Dell (2005, 412, 417) reports that top incomes in Germany quickly reconcentrated after the second war, and attributes this to the much lower inheritance taxes there.
The phase of declining pre-fisc inequality was reversed in nearly all of the industrial countries in the 1970s and 1980s, a reversal that appears generally to have involved both the very top income recipients, whose share has risen in most countries and dramatically in some, and also the rest of the income hierarchy.The jump in the shares of the very rich is of concern to the developing countries if there is reason to believe that it might be occurring in them contemporaneously or might appear soon.If further evidence demonstrates that the patterns are similar to those of the rich countries, one hypothesis will relate to globalization and its tendency to raise upper-range incomes of developing

The Record in the Developing Countries
Useful, albeit incomplete, statistical information on income inequality started to become available in the developing countries just a few decades ago with the appearance of periodic household surveys.Apart from the already noted inertia of income distribution over time under normal conditions, the main conclusions to emerge to date are that: 1. inequality, especially post-fisc, tends to be higher in most developing countries than in the industrial ones.Measured by the Gini coefficient, the range for pre-fisc income is roughly from 0.3 to 0.6 or a little higher and for post-fisc incomes (including government transfers) probably from 0.25 to 0.55; 7 2. many of the countries of Latin America have typically been at the upper end of the inequality spectrum, while a number of those of East Asia have been at the lower end; 3. measured inequality has been increasing in more countries than it has been decreasing (e.g.Cornia and Kiiski (2001); 4. there have been increases in recent decades in the two largest developing countriessignificant in India (Deaton and Dreze 2002) and enormous in China (Benjamin et al, 2008); 5. primarily as a result of the cited trends for India and China, the population-weighted trend in inequality in developing countries has been upward for several decades (Berry and Serieux, 2007, 83) 6. over recent decades the capital share of income has been rising in developing as well as in developed countries (Guerriero, 2012); 7. recent evidence from several countries of Latin America suggests that inequality may have begun to fall (Lustig et al, 2011) -remaining uncertainty relates to the fact that the data on capital incomes are very incomplete.If true, may these initial albeit thus far modest declines be the beginning of a significant downward trend of inequality in the developing world, perhaps first in Latin America, a relatively rich region by developing country standards? 8 The inequality of educational attainment across the economic hierarchy has been falling in many -perhaps nearly all -developing countries as the result of rapid growth of primary and secondary educational coverage (Holsinger and Jacob, 2008).Similarly, it appears that the inequality of health outcomes has been falling.

Inequality Trends at the World Level 9
The huge gap between the world's rich and its poor has made trends in world inequality a matter of much interest.That gap widened markedly during a period beginning in the early nineteenth century or probably earlier 10 and continuing until at least the middle of the twentieth; the bulk of this widening had taken place by the early 20 th century and thus substantially coincided with the period during which the Industrial Revolution saw the now industrial countries create and greatly widen the gap in average income between them and the rest of the world.The bottom three quintiles of the world population lost systematically until 1980 (from 25.7% of total income in 1820 to 12.5% in 1980) and the top decile gained continuously through to 1992.At the world level this, then, was a long period of highly exclusive growth.
Apart from its dramatic scale-considerably more severe than that of any but the most inegalitarian countries, and the major increase over a long period leading up to the middle of the twentieth century, the literature underscores two other main features of world inequality.First, that inequality comes mainly from inter-country income differences rather than intra-country ones. 11Second, the level of world inequality did not change markedly, in either direction, between 1950 and 1980 or so (Berry, Bourguignon and Morrisson, 1991).As for the 1980s and 1990s, although there has been some disagreement about the trends most studies conclude that (i) changes in the standard indicators, in whichever direction, were not large relative either to the absolute level of inequality or to the changes over the preceding couple of centuries, but that (ii) the bottom three quintiles and the top one gained at the expense of the upper-middle group, especially deciles 8 and 9.For the bottom three quintiles, therefore, their long decline in the share of world income was reversed, growing from 11.3% to 14.0% (according to Berry and Serieux, 2007, 81).Meanwhile the share of the top decile continued an upward trend that appears to have been in evidence for upwards of two centuries.
What is perhaps most intriguing, and at first glance paradoxical, about the inequality outcome for 1980-2000 is that, though no previous post-war period has been characterized by as general a pattern of intra-country worsening of distribution, overall world inequality appears if anything to have moved a little in the opposite direction.
Expressed in terms of the Theil index, which can be decomposed in a straightforward way, inter-country inequality accounted for over three-quarters (75.8 per cent) of overall world inequality in 1980, while just 24.2 per cent reflected intra-country inequality (Berry and Serieux, 2007, 83), but by 2000 these proportions had changed to 66.1 per cent and 33.9 per cent respectively, reflecting the general deterioration of intra-country distribution evidenced by the rising Gini coefficients mentioned earlier, along with the rapid decline of the between-country component.

World Poverty Levels
Historical trends in the incidence of poverty and extreme poverty have mainly reflected growth rates of the countries with many people near the poverty lines.Poverty incidence naturally depends on where poverty lines are drawn, and it is now customary for analysts to present at least two.By 1850 the Bourguignon-Morrisson lines for poverty and extreme poverty showed only a small decline from their 1820 levels of 94.4 and 83.9 respectively, but the reduction accelerated thereafter, and was rapid over the 20 th century -especially since 1950, such that by 1992 they had fallen to 51.3% and 23.7% respectively.Berry-Serieux (2007, 93) estimate poverty incidence in 1980, 1990 and 2000 for 500 and for 1000 international dollars of 1985 per year (Table 1), 12 and report that extreme poverty continued to decline rapidly during the 1980s, after which the pace slackened markedly in the 1990s.During the 1980s, the share of people in this category fell sharply in East Asia, mainly reflecting the growth of China and also in South Asia, while remaining about constant in Africa (Table 2).In the 1990s, such extreme poverty was again roughly halved in East Asia, though the rate of decline slowed sharply in South Asia due to increasing inequality in India, and incidence rose markedly in Sub-Saharan Africa.These two decades saw only a modest reduction of extreme poverty in the world outside China and none at all in the world outside China and India.The experience of the 1980s and the 1990s is dramatically different with respect to poverty.For the world as a whole and for the world minus China, the percentage point decline was considerably greater in the 1990s (Table 2).In summary, East Asia and South Asia, the two regions with the largest poor populations, reduced poverty rapidly during these two decades while the third one (sub-Saharan Africa) was going in the opposite direction.For the most recent and naturally most interesting decade (2000-2010) there appear to be no detailed analyses of the evolution of world inequality.From the fact that developing country growth has been faster than that of the leading countries, it is clear that by some criteria inter-country inequality has diminished. 13Remaining uncertainty relates to serious data problems at the bottom and at the top of the world distribution.The main cause of pessimism about what is happening at the bottom is that many of the countries with low average incomes are now found in sub-Saharan Africa.Although the region has experienced good growth over this decade, it has to a large extent taken the form of natural resource-based booms, which create the possibility for serious increases in inequality with the implication much of the world's bottom quintile has not been experiencing the sort of income increase that the aggregate growth figures suggest; stories abound of theft and nepotism at the top, as Nigeria's sad natural resource curse story (Sala-i-Martin and Subramanian, 2003) runs the risk of being repeated elsewhere in the region.At the top end of the world distribution the long-standing inability of developing country surveys to capture much capital income, together with the fact that the world's super-rich now include many people from developing countries, could cause an increasing downward bias over time in this decile's measured share.Failure of the statistics to capture income from asset appreciation, which at times can constitute a quite substantial share of total capital income, especially that concentrated at the top of the distribution, could also constitute a downward biasing element in the estimated trend of concentration at the top.
The Determinants of Inequality; a Closer Look together with a set of market "imperfections" that leave incomes of individuals either higher or lower than their contributions to production. 14Each of these categories of determinants shows up directly in a person's income, and a key practical question is which have the biggest impact on the distribution of income among people.
Personal Correlates of Pre-Fisc Inequality: a Starting Point With respect to the proximate determinants/correlates of income -and bearing in mind that in a high inequality country the income gap between the top and the bottom deciles lies in the range 50:1 to 60:1 or higher, several points stand out.First, human capital, as approximated by level of education and degree of work experience, explains much of the earnings differentials across people that show up in household surveys.In statistically explaining 25-40% of such differences, this variable dominates all the others that have been included in such analyses.Average income of university graduates can be as much as 10-15 times that of illiterates.The traditionally high estimates of the payoff to education in developing countries may be seen as the other side of the coin from its high correlation with income.Recent studies (Rosenzveig, 1998;Pritchett, 2001) point to probably serious upward biases in most estimates of the true causal impact of education on income, as opposed to that of other personal characteristics linked to the level of education (e.g. the value of one's family network).Second, differences in physical and financial capital are the other main direct determinant of inequality, in particular inequality at the top of the income hierarchy.This 20-40 percent of income (excluding that from asset appreciation) is quite concentrated at the top, but its precise distribution is not yet known.
Other important correlates of income are gender, ethnicity (often), region of country (sometimes), rural vs. urban residence, sector of activity, size/modernity of firms, and whether union member or not.That such characteristics should be related to income usually suggests either discrimination or some other form of market imperfection.Since there are many intercorrelations among these variables and between each of them and level of education or human capital, average income (or gross) differences between categories of people defined by differences in these variables typically overstate their net causal contribution to inequality.Whereas the gross differentials can be 2:1, 3:1 or sometimes higher; the net ones are seldom more than 50%, as with gender, and (sometimes) region.Though often emphasized because of the unfairness involved (as in the case of gender or ethnic income gaps) they do not loom nearly as large either in gross terms or as contributors to overall income inequality as do human and physical/financial capital.
The role of native talent (intellectual, business, artistic aptitude, looks, etc) is something of a puzzle.No one doubts its role, together with that of hard work, in determining who gets high incomes from sports, the arts and a few other activities where performance is relatively easy to measure.But attempts to ascertain whether it plays a significant role in the creation of population-wide income differentials have thus far come up with little.Taken literately, most suggest a very small role for native ability in explaining income differentials within a population (Boissiere et al, 1985).This may reflect the difficulty of measuring native skills in an adequate way; further, the fact that success in different areas may rely on different skills suggests than any native ability test needs to be geared differently for different people.In any case, there is still no solid evidence to the effect that native ability is a significant factor in a country's income distribution.

Structural and Policy Determinants
Many aspects of the setting within which the personal differences among people play themselves out also have an impact on the level of inequality.We here consider four-the extent and pattern of technological change, population level and growth, market imperfections and the degree of openness of an economy.

Technology Choice and Biased Technological Change
Technology choice and biased technological change(i.e.biased towards the use of capital and against that of labour) may be the most important structural contributor to the level and trend of income inequality, since it can play a powerful role in determining both the capital share as well as the distribution of capital and labour incomes.There is no doubt of its relevance but the magnitude of its impact is hard to measure precisely and, since technological advance is essential to growth, there is a possible trade-off between growth and labour demand.In principle technological advance can be employment creating, for example when it involves the improvement of intermediate technologies (those with medium-level capital/labour ratios) which when applied in small or medium sized enterprises (SMEs) can make them more competitive with large firms, as occurred in Korea from the mid-1970s during a period of very rapid growth of manufacturing SMEs (Nugent and Yhee, 2002).How often policy and structure can be combined to produce this sort of result is the key empirical question.At the pessimistic end of the spectrum of possibilities, it may be that larger-scale capital intensive firms will continue to raise their share of output but not of employment and will squeeze the micro, small and medium enterprise (MSME) sector with the result that the generation of decent employment remains weak, wages low and inequality high.
A large role for technological change as a determinant of inequality is suggested by (i) the microeconomic evidence of large differences in the capital/labour ratio across firms of different sizes together with the high degree of polarization or dualism in the distribution of firms by size and type of technology; (ii) the accumulated evidence that much technological change is either moderately or strongly labour displacing; (iii) the coincidence in a number of countries of sharp increases of inequality with improved access to foreign exchange and thence to more modern foreign machinery and the technologies embodied therein; and (iv) the evidence over recent decades that the capital share of income has been rising in both developing and developed countries (Guerriero, 2012).Such empirical evidence is consistent with the theoretical prediction that the shift towards more capital intensive technologies would raise capital incomes and hold wages down.But the magnitude of the effect, what variables that magnitude depends on, and the extent of any trade-off between achieving a high rate of economic growth and a low aggregate level of labour displacement remain open to discussion.Reasonable guesses are that technological change is the most important single factor in raising inequality or keeping it high and that it takes quite good policy to avoid a significant trade-off between such displacement and growth.

A Dense or Rising Population
A dense or rising populationhas long been seen as a source of high or rising inequality.By increasing the supply of labour relative to that of other factors of production, it exerts downward pressure on the wage rate.This may or may not lower the wage share of GDP (ceteris paribus, a lower wage pushes this share down but more workers push it up) but it does widen the gap between the income of the average capitalist and that of the average worker.In the Ricardian growth model, where labour was seen as homogeneous, proletariat population growth was essentially a regulator which guaranteed that the wage rate stayed close to the level of subsistence, through a mechanism whereby a rising wage led to higher child survival rates and greater completed family size, but the resulting population growth then pushed the wage rate back down (Malthus, 1926).A second link between demographic patterns and inequality has its roots in the fact that family size has often been larger in the working class than among the middle classes; this increases the size of the lower income groups while also leading to a more rapid divvying up of the limited capital they have than occurs among the rich.By the time human capital becomes important as a source of income, their large families also push down the average amount they can invest in this way.
There are no very precise tests for the role of these demographic features in the level and evolution of inequality, but some conclusions are possible.First, the Malthusian prediction of a devastatingly negative impact of population growth on GDP per capita turned out not to hold for the now industrial countries, 15 which in effect outran this threat to reach a threshold level of income, education and other determinants at which family size began to decline, eventually falling below replacement levels.Similarly, many developing countries are in the process of outrunning the threat, leaving only Sub-Saharan Africa (at 2.5% per year in 2011) with population growth above 2% per year.An ultimate escape from the Malthusian peril does not, of course, mean that many countries have not paid (nor are still paying) a heavy price for their fast population growth.It seems likely that such growth has with some frequency both slowed growth of income per capita and raised inequality, in which case its impact on poverty could have been (or could be) quite negative.How negative remains a matter of debate. 16In currently heavily populated countries the damage has been wrought over the long period during which population density rose so far as to shrink average farm size to one-half hectare or so.In others, whose resources to population ratio is better, the impact may come from the fact that very fast growth over a shorter period makes it difficult for the economic system to absorb that labour increase rapidly and productively, leading among other things to serious employment problems.That demographic pressure is not generally one of the dominant sources of inequality is, however, suggested by the fact that inequality has typically been lower in the densely populated countries of Asia than, for example, in less densely populated Latin America.

Market Imperfections
Market imperfections are likely to affect inequality when the most important of them are orchestrated by well-off beneficiaries to in effect transfer income to them at the expense of the poor; this is most typically true of the capital market but occurs in many product markets as well.In the labour market a degree of market power is exercised by unions and a degree of state intervention comes through labour legislation.Their net impact on inequality has been greatly debated, but is unlikely to be large except in exceptional cases, since it is constrained by the facts that (i) usually only a minority of workers are union members and/or are effectively covered by labour legislation and (ii) wages cannot be pushed too far above the market equilibrium level without seriously deterring employment.The capital market has much greater potential to affect (raise) inequality, by favouring better off people vs. the less well-off and larger firms vs. small ones.The same is true of product markets, since extensive monopoly power raises the profits of the rich (including favoured workers) at the expense both of consumers and of labour as a whole, in the latter case through its downward pressure on the demand for that factor.Most of the few reasonably competitive markets (including that of unskilled labour) involve low income people on the supply side, with the lack of "imperfections" thus tending to help the better off.Many other imperfections also have clear impacts on inequality, usually tending to raise it.The most blatant "imperfection" is the theft of factors (e.g.land), usually by the rich (or those in the process of becoming rich) with the aid of weak legal systems and generally dysfunctional institutions.
Among policy issues involving freedom of markets from intervention, the possible distributional effects of trade policy and capital flows, and more generally of globalization and market liberalization, have been a topic of much debate and little consensus. 17The simplest static trade theory suggests that the distributional effects of freer trade will reflect the factor intensities of the export and import-competing products; if the former are labour intensive and the latter capital intensive wages should rise relative to the returns to capital and if the opposite is true the reverse should happen.For most of the East Asian tigers the first burst of exports involved labour intensive goods and at least for Taiwan it appears that rising trade improved income distribution (Fei et al, 1979).For many other developing countries whose comparative advantage lies in the export of primary goods (minerals or their products, agricultural goods, etc.) the distributional effect depends on the combination of the factor intensities of the tradable involved and who owns or who ultimately gets the revenues from the natural resource base.At one end of the spectrum are those minerals produced using very capital intensive technologies and generating high capital incomes for a few people.At the other end are agricultural products like coffee and tea which, when produced on small farms, generate high levels of employment and also provide capital and land-based income to relative low income families (Bourguinon and Morrisson, 1989); since dispersed ownership of the land typically goes with substantially greater use of labour, these two determinants of distribution both contribute in a positive direction.All of these impacts are, however, complicated by the role of government, which often owns and operates mineral industries and in other cases collects large tax revenues which provide a basis for spending whose allocation forms part of the final impact.
The distributional impacts of capital flows are also hard to sort out.When such flows are simply translated into an increase in the country's capital stock, 18 their raising that stock relative to the supply of labour should raise wages and lower the returns to local capital, hence producing a positive effect on income distribution.When the inflow brings a technology shift towards the capital intensive, as FDI often does, the impact is more likely to be negative; when it flows into labour intensive industries in order to take advantage of lower wages, the impact is likely to be positive.
The relative frequency of positive and negative distributional impacts of openness to trade and capital flows and their quantitative dimensions have been matters of great controversy, partly because there is no easy way to judge them 19 and partly because economic openness has been a major topic of debate more generally. 20The wide range of conclusions emerging from the different attempts to analyse this issue may reflect the fact that none has gone far enough in the direction of distinguishing settings and groups of countries where different outcomes might be expected.Still, the lack of general consensus suggests that the impact is not normally very large, unless much of it comes with a lag and/or involves those components of income that are not properly measured.
Over time country-level evidence tends to support the theoretical prediction that increased trade will lower inequality in some countries, especially those exporting labour intensive goods, and raise it in others, especially those exporting primary products based on assets whose ownership is concentrated or those where freer trade leads to rapid labour-displacing technological change.Taiwan exemplifies the former, and a number of Latin American countries the latter.The higher inequality in mineral dependent exporting nations also fits this broad prediction (Berry, 2008).But detectable effects are modest in relation to the inequality gaps across countries.It seems likely that if major effects exist they occur through the dynamic processes of investment and technological change or perhaps through the process of growth acceleration, to which openness can contribute.The rapid turning outward in India and China and the enormous increase in inequality in the latter are consistent with an important role for those dynamic mechanisms.Confirmation of their greater role than that of differences in factor proportions comes from Meschi and Vivarelli (2009, 296), whose state of the art analysis focuses on over time changes in inequality within countries (rather than cross-country differences).They find that while changing total trade flows (in relation to GDP) are not significantly related to inequality, trade of middle income developing countries with the industrial countries does push that inequality up, while trade among developing countries pushes it down.This suggests that the former trade facilitates the diffusion of new technologies, in particular in the middle-income developing countries where the capacity to absorb such technologies is greater, with associated labour-displacing and inequality-increasing effects.Trade among the developing countries is presumably weighted towards goods that are not capital or skill-intensive.
43 The above cited determinants plus the interventions of government are the core mechanisms at work to determine the level of inequality.Government enters the picture in various ways, most obviously by taxing the population and engaging in spending but also through a wide range of other policies, including those on international trade and capital flows.The public-private transfers that take a monetary form (taxes from people and monetary transfers to them) account for the difference between pre-fisc or primary income and "disposable" income.Much government activity takes the form of provision of goods and services (like education, health services, food); allocating the associated benefits to individuals or families requires a complicated calculation to arrive at a distribution of "post-fisc" income.
44 Behind the set of "proximate" causal factors discussed above are the underlying or more basic determinants of inequality and poverty that explain the proximate ones, e.g.why and to what degree markets are rigged for some groups and against others, why the distribution of assets is as it is, etc.

Building a Theory of Inequality and its Evolution
The main factors of production are land, physical capital, human capital (i.e.skills, among which entrepreneurial and innovative capacity are sometimes singled out) and pure labour (basic skills).As development proceeds, land and other natural resources become relatively less important and physical and (later) human capital relatively more so.The distribution of physical capital is normally quite concentrated, that of land is in some countries but not in others, and that of what we may call basic labour (the skills that are mainly manual and quickly acquired and hence not too different among people) relatively equal.Human capital-the skills and capabilities that result from education and training may or may not be unequally distributed, depending on the country.The more unequally each of these factors is distributed among people or families and the higher the correlation between a person (or family's) holdings of one factor and of another, the more unequal the primary distribution of income is likely to be.
The technologies in use, together with the relative endowments of the factors of production, determine the prices of those factors and hence the "functional" distribution of income among basic labour, human capital, physical capital and land (or, more generally, natural resources).Overall income distribution is more equal when the functional distribution favours those factors that are least unequally distributed, especially basic labour.When the technologies favour capital, such that the physical and human capital shares of national income are high, then the pre-fisc distribution of income tends to be unequal.Three mechanisms tend to bias technological change in favour of the better off owners of capital.One is the importance of borrowing technologies from the industrial countries; such technologies are usually capital intensive in relation to the factor endowments of developing countries and thus depress the demand for less-skilled labour and increase the demand for higher skills and the higher incomes that go with them.A second is the fact that the resources to generate new technologies or forms of production are greater in larger rather than smaller firms, and the former tend to use and prefer capital intensive technologies.Finally, public efforts to develop technologies for smaller scale enterprises are often far below their optimal level.Historically, R&D for agriculture has been either small-farm focussed on more or less size neutral; as a result the Green Revolution brought widespread benefits across the agricultural sectors of the countries where it took hold.Recently agricultural research has in many countries been increasingly privatized with an associated focus on use by large, capital-intensive farms.All of these mechanisms may be expected to raise inequality in a country, but it remains hard to ascertain how powerful their effects are.
Political and sometimes social mechanisms underlie much of what happens within these domains, and tend to create the vicious or virtuous circles that explain the striking inertia in the level of inequality in countries with either high or low inequality.The vicious circle usually begins with a politically and economically self-reinforcing concentration of all major forms of capital.Usually it also involves market imperfections, some of them policy-induced or uncorrected by policy, that leave the return to a given factor higher for the better off and hence better connected.Examples include the capacity of some large capitalists to achieve monopoly profits, while small informal enterprises may realize only very low returns on their capital because of the competition they face.Political power upholds monopoly profits by blocking anti-trust efforts.That political process channels public funds for education mainly to the better off, and cohabits with a financial system that also favours the better off, larger firms.
At the other end of the spectrum of possibilities is a country with low levels of asset concentration, an egalitarian educational system, and public support for unbiased technology development.The total impact of differences in asset concentration, the character of technological change and the pattern of market imperfections may be approximated in terms of the Gini coefficient as the difference between about 0.30 and about 0.65 (or the difference in the income ratio between the top and bottom deciles of say 55:1 vs. 6:1), but the contributions of each of these factors and of the ways they combine has not been sorted out.

Effects of Inequality on Growth
Apart from the ethical issue of justice and fairness, various other effects of inequality warrant attention, most prominently its possible impact on growth.In the early days of development thinking it was often held that the goals of growth and equality were at odds; 21 allowing and fostering the wealth of a few was seen as a necessary incentive for entrepreneurship and the high savings rate needed to finance investment.Since capitalists were the savers, innovators and inventors, 22 a premature pursuit of equity would come at the expense of growth.This view may have held considerable validity at some points during the economic history of the now industrial countries. 23However, the setting within which the growth-equity relationship plays itself out has changed in various ways over the economic history of the now industrial countries and varies considerably across today's developing countries, so it is unlikely that there would be a universal pattern across countries and over time. 24  It has been argued frequently over the last couple of decades 25 and is probably now the majority view among students of the issue, that an equal income distribution has on average been associated with and promotive of faster growth A larger majority probably hold that redistributive policies can be growth enhancing when reasonably well designed and implemented -a more relevant question than whether, on average, good growth and improvements in distribution have tended to go together.Whether or not these claims stand the test of time, it is clearly true that fast growth can occur with either low or high levels of inequality, and that it can occur as inequality increases (as recently in India, and more dramatically, in China) or as it decreases (Taiwan, earlier and some countries of Latin America more recently).In short, theory suggests that some policy packages may achieve equality in ways that are promotive of growth and others in ways that deter growth; in the latter case it may be that neither objective is attained.Given the policy levers available in principle to governments, achieving growth with equity (of primary distribution) is usually not a very difficult challenge in technical/economic terms.
The overwhelming explanations of why it is the exception rather than the norm are political and institutional-the pressures in favour of growth paths that favour the few.

Policy Response
Inequality, poverty and the insecurity that goes with them can be reduced both by wellchosen economic policies (which affect the primary distribution and thus prevent these maladies from arising) and by what are often called "social policies" designed to react to and offset poverty and vulnerability.Historically most gains have taken place through the former route, but by the 20 th century social policy was becoming increasingly important in the now industrial countries and by the late 20 th and the 21 st it has taken on considerable significance in the developing countries, especially the higher income among them.As noted, theory suggests that some policy packages may achieve equality in ways that promote growth and others in ways that deter growth.
There is a considerable degree of consensus that improvements in the primary distribution are best achieved through the following (partially overlapping) elements: 1. land reform or other steps to achieve an equitable distribution of land, and public support to raise productivity in the small family agricultural units; this is especially important in countries that are still mainly agricultural; 2. an industrial policy providing adequate incentives for relatively labour intensive activities, which may be seen as including 3. appropriate support for smaller, more labour intensive businesses outside agriculture; 4. a strong and equitable educational policy that leaves a low variance in the level of education/training; 5. a set of policies to diminish the biases of markets and of ill-chosen forms of government intervention (e.g.unneeded red tape) against the poor.
Elements (I) to (III) all involve encouraging the demand for labour which, as noted below, is usually the major broad challenge to policy The post-fisc distribution depends on the pre-fisc one, plus 1. a tax, transfer, and public goods system that lowers inequality and provides protection against economic insecurity while not significantly lowering incentives to work or pursue business enterprise.Such a system involves many components, one of the most important being the sharing of rents from natural resource exploitation broadly within the population.
None of policies (I) to (V) is, per se, likely to slow growth, when applied in the appropriate manner and measure.But in most countries none of them are pursued very effectively, with the result that the primary distribution of income remains quite unequal unless and until the tax and transfer system comes into play as a partial palliative, as has recently occurred in a number of mostly middle-income developing countries.Where policy is good, it tends to be good on most of these fronts, a fact that suggests either that the technical and political prerequisites (including the nature of the society and the social contract) are similar for all, or that progress on some of them facilitates progress on others. 26Which policies are adopted and how successfully naturally depends on the underlying social and political institutions.Which have the greatest potential depends on the economic considerations involved.

The Key Role of Policy to Foster Labour Demand
A strategy to enhance equity must operate substantially through the generation of an adequate supply of employment opportunities in both quantity and quality terms.In most cases a good employment strategy brings faster as well as more equitable growth, along with the direct psychological benefit people get from having a decent job.In virtually all the countries of the world there is frequent reference to "employment policies" or an "employment strategy" but in almost none does one find a well designed and implemented set of policies that deserves the term.Labour outcomes (wages, levels of employment, unemployment and underemployment, working conditions, etc.) depend on the demand for each of the different categories of labour, on the supply of each, and on how effectively the labour market functions in bringing demand and supply together.Facilitating a healthy demand for labour is almost always the most important of these three determinants of labour outcomes as well as being the biggest challenge to a better income distribution and the reduction of poverty; policy too often meets that challenge poorly.Though it is arguable that the demand for labour is the key determinant of labour outcomes it has probably received less theoretical and general attention than have supply and labour market functioning. 27This weakness has probably been quite costly in terms of foregone distributional and poverty-reduction benefits.The main (interacting) factors lying behind it appear to be five: 1. a lack of interest (even an active disinterest) by decision-makers in these variables as objectives of economic policy, because their own welfare is unaffected by them; at worst leaders have an aversion to either attention to or measurement of such variables 28 2. the lack of a persuasive composite labour outcomes indicator, comparable to GNP for the economy's production level.A main advantage of such an indicator is that it can focus attention on this dimension of economic performance.(A danger associated with such an aggregate variable is that it may invite lack of separate attention to the components on which it is calculated); 3. the lack of a complementary theoretical framework like that used to buttress the contention that under largely free markets GNP will be maximized; 4. the technical or objective difficulty of designing good policies that target labour outcomes, as compared for example with policies that target GNP growth, if only because the determinants of the former include all those of the latter plus a number of additional ones; 5. the institutional difficulty of carrying out a good set of policies even if it has been well formulated, because of the number of disparate institutions that need to be involved, among which the Ministry of Labour is not usually one of the more important.
58 Probably all five of the above factors have come into play historically to explain why there is a generally accepted composite variable for the level of economic activity but not for labour outcomes.Decision-makers have been more interested in output than in labour outcomes; GNP as a composite variable has a clear economic meaning; with time the technical problems of measurement and the learning about which policy instruments promote higher GNP have yielded and evolving economic theory built around GNP has proven useful.There are causal links between the lack of priority assigned to employment and labour incomes as goals of socio-economic policy and the technical and institutional problems in the implementation of such policies, since these latter problems and challenges are typically resolved only with time and an effort proportional to their perceived importance.Developing target variables that are useful in practice involves both defining them in a conceptually meaningful way and measuring them with enough statistical precision and in a sufficiently timely way.Failure to do so can reflect both the lack of interest and the genuine obstacles to doing so. to achieve stability in the economy are.Many first-best options are precluded by the distribution of political power.To exemplify, in many low income countries it is arguably the case that the most effective (highest benefit/cost ratio) policy to cut poverty and inequality is a strong small family farm strategy, including land reforms (shifting land from larger to smaller units) where necessary and a strong support system of R&D, diffusion of new technologies and investment in complementary infrastructure and human capital.In many other countries the best option is a strategy to generate a strong demand for labour.This would normally include support for the sort of small nonagricultural enterprises that have high employment generating capacity, through a set of policies ranging from microfinance systems to technical support for small and medium firms, appropriate limitations on patent privileges, control of the monopolistic tendencies of some large firms, and support for mutually beneficial interactions between other large firms and small and medium ones.Finally a third option may be a system of transfers to the needy (welfare).In fact, all these types of policies are likely to be part of an optimal package.But land reform almost never happens since it involves taking an asset (land) in fixed or relatively fixed supply from richer people and allocating it to the less well-off, and at a time when rural elites still have much political power, as reflected in the last two coups in Latin America -Honduras and Paraguay.Support for small and medium enterprise and other elements of a good labour strategy do not face the strong political opposition that land reform does, since they are not so often perceived as being against the interests of the rich.But policy-making in this area is complicated, as noted above, and the requisite dedication to doing it well has not been present.Finally, welfare payments may garner even more political support since they elicit the feelings of generosity and humanity that part of a country's population has, and are often seen by the elites as a way of avoiding social unrest.But they are more likely to be palliatives, with shorter-duration effects than the other two policies, especially the first.

Broad Conclusions
60 Inequality is unacceptably high in most developing countries and has, on average (weighting countries by population) been increasing for the last several decades.It has enormous social costs, partly by making poverty unnecessarily widespread and partly through other social and psychological mechanisms.Remedies are thus urgently needed.Understanding provided by economic analysis suggests a package of policies including (especially in the more agricultural countries) land reform or other steps to achieve an equitable distribution of land, and public support to raise productivity in the small family agricultural units; a strong and equitable educational policy; an industrial policy providing adequate incentives for relatively labour intensive activities, including support for smaller businesses outside agriculture; a set of policies to diminish the biases of markets and of ill-chosen forms of government intervention (e.g.unneeded red tape) against the poor; and a tax, transfer, and public goods system that lowers inequality while not significantly lowering incentives to work or pursue business enterprise.An integrated set of policies to promote the demand for labour, key to raising incomes on a broad basis, uses several of the above policies as its building blocks.Unfortunately the potentially most powerful of these policies-land reform/small farm support systems, a generally strong policy to create labour demand, and a strongly redistributive state are, in the first case politically untenable in most countries, and in the other two difficult for technical as well as political reasons.
61 The historical record of the industrial countries shows that inequality reduction has come substantially from wars and depressions (unacceptable routes) and, in the absence of these two factors, from strong tax systems that over time lower inequality or prevent its increase.This is bad news for the majority of developing countries whose tax systems have been shifting towards the use of indirect and not very progressive taxes.It means that, if the state is to have a significant redistributive function the great bulk of it will be carried out by expenditure policy.

NOTES
1.The year 2000 or one close to it is chosen to depict the structure of inequality because (a) more recent estimates at the world level are less reliable due to the time lag between the appearance of country data and their being appropriately adjusted and integrated to provide world level figures, (b) it is desirable to present individual country figures for about that same date, and (c) except in very unusual situations (like wars) inequality changes very slowly, so the patterns in 2010 are likely to be very similar to those of 2000.Section 3.3 elaborates on the limitations of the data for the most recent decade.

Figures from World Bank
Databank, online, with upward adjustment for the top decile in accordance with the suggestions of Altimir (1987).
3. See Mauss (2011Mauss ( , 1950) ) for a classic treatment of reciprocity as a form of association.

4.
Thus in Norway until 1860 many entrepreneurs and merchants benefited from rights that prevented competition; only merchants in towns could sell, forcing the rural population to do their shopping there Soltow, 1971).

5.
The term "social transfers" usually refers to all social spending to help the lower income people except the direct provision of education and health services."Social spending" usually refers to those transfers plus spending on education and health.
6.In Europe's early "elite" democracies, the criterion determining one's right to vote was a minimum amount of land, income or direct taxes paid.The retreat of these restrictions in the19th century opened the door to greater taxes and social spending.Defining such "elite democracies" as ones where under 40% of men had the vote, Lindert (2004, 23) concludes that they were more opposed to overall social transfers and to public pensions than was the average authoritarian non-democracy.
8. Ortiz and Cummins (2011) report declining inequality for a number of African countries but the very doubtful over-time comparability of sources in most of these countries leaves this result open to question, as does the erratic behaviour of the indices in some cases.
10. Estimates by Bourguignon and Morrisson (2002) go only as far back as 1820.
11. Country figures are aggregated using international prices to effect the most meaningful comparison.
12. The $500 and $1000 poverty lines can be considered roughly comparable to the $1 and $2 a day consumption lines used by the World Bank, lines that can be reasonably considered as ceilings for extreme poverty and poverty respectively.
13. World income growth per capita averaged 2.27% per year; that of high income countries grew only at 1.0% while that of middle and low income countries respectively grew at 5.0 and 3.7% per year.

14.
In centrally planned systems, personal incomes do not bear the same relationship to productivity as they do (at least in theory) in a market economy.With the state owning the great bulk of capital, the income attributable to that factor, rather than flowing to capitalists and swelling the income share of the top decile or so of the distribution, is allocated according to the dictates of the state, either as wages, as public goods, or (usually to a significant degree) in the form of delayed future income (via high investment rates).Labour income is also typically more equally distributed in these countries.As a result the communist countries (prior to their transitions towards a more market orientation) typically displayed Gini coefficients on average 10-20 points below those of capitalist market economies.In this sense, a major determinant of inequality is the economic system.
15. Marx and the classical economists both came to pessimistic conclusions on this point, influenced in part by contemporary demographic trends and in part by their limited optimism with respect to the rate of growth of productivity.Happily that pessimism has been superseded by events.The industrial revolution, already in motion as they wrote but with its transformative capacity still not evident, undid their doubts with respect to growth potential and although the negative potential threat of an expanding population remained a dark cloud for much longer, it too has now begun to subside as world population growth recedes and zero population growth is now a light at the end of the tunnel.
16.For a recent review, see Ahlburg and Cassen (2008).Analyses are complicated by the fact of two-way causation; regions with high growth dynamic attract immigration and vice versa.Demographic developments are no doubt linked to economic growth and inequality by many causal mechanisms.
17. Since industrial policy has historically been implemented to a large degree by trade policies (tariffs, quotas, antidumping laws), the distributional impacts of trade and industrial policy largely overlap.For example, in countries where more capital-intensive industries were protected, the freeing of trade might be expected to lead to increased production of labour intensive goods and to improved income distribution.

18.
As with a portfolio inflow unaccompanied by other factors of production or by the introduction of new technologies.
19.This in turn is due in part to the difficulty of sorting out immediate effects from longer run ones.The latter may occur, for example, through a gradual upward effect on the size of firms and the share of world GDP produced in very large (usually multinational) enterprise.Effects may also occur through changes on the relative prices of goods consumed principally by the well-off vis a vis those consumed more by the poor, which are not often identified in studies of inequality trends over time.

20.
A useful review is Robbins (2003).The strongest views and disagreements occurred during the neoliberal assault on protectionism (Krueger, 1983), as the proponents of openness and liberalization argued that not only would this policy reform accelerate growth but also that it would diminish inequality.The underlying theory was mainly static, building on the plausible idea that under free trade developing countries would export labour intensive goods and services and import capital intensive ones.While this pattern has been at work in many developing countries, especially densely populated Asian ones, it was not generally dominant across the developing world and has in most cases been outweighed by some of the other mechanisms noted above.The main error of this body of thought was the overconfidence it exuded on the basis of partial analysis and in advance of any empirical confirmation of the predictions.The result was one of the more striking mispredictions by a group of development economists over the last few decades.
23.This was a plausible view when most of the rich were capitalists, but not when they were mainly rentiers.As the former gradually replaced the latter at the top of the income hierarchy its overall validity rose.

25.
For example, Alesina and Rodrik (1994).Suggesting that the issue should be framed in a more complicated way and that methodological differences need to be taken carefully into account in comparing results are studies like , who found that inequality was bad for growth in poor countries but good for it in rich (i.e.developed) ones, and who, using non-parametric methods, find that a change in inequality in either direction is associated with a reduction in growth.

26.
In the classic case of Taiwan, for example, the land reform contributed to an already strong legacy of equality in educational attainment as well as strength in and support for smaller enterprises that dominated the labour intensive industries.

27.
Achieving a high demand for labour is partly a matter of having a high level of aggregate demand (macroeconomic policy) but, especially in developing countries, it is mainly a matter of keeping the demand for labour high relative to that for other factors of production, the aspect on which we focus here.The former component involves Keynesian economics, the latter neoclassical economics.

28.
It is understandable that elite-serving governments would give at best modest priority to labour variables and there have been many such governments.Many specific models have been proposed to explain their modus operandi (e.g.Acemoglu, Ticchi and Vindign, 2006).Seers (1983, 6) noted that "Those who hold power rarely have much interest in such matters, still less in attention being drawn to them.It is preferable to shelter behind the 'growth rates' that are commended in the reports of international agencies."Enormous economic gaps exist within countries and even more so in the world as a whole.
Inequality is unacceptably high in most developing countries and has, on average (weighting countries by population) been increasing for the last several decades.It has enormous social costs, partly by making poverty unnecessarily widespread and partly through other social and psychological mechanisms.The paper focuses on the causes of these gaps, the extent to which they could be justifiable on economic or other grounds and the steps governments can take to reduce them.Understanding provided by economic analysis suggests a package of policies including (especially in the more agricultural countries) land reform or other steps to achieve an equitable distribution of land, together with public support to raise productivity in the small family agricultural units; a strong and equitable educational policy; an industrial policy providing adequate incentives for relatively labour intensive activities, including support for smaller businesses outside agriculture; a set of policies to diminish the anti-poor biases of some markets and of ill-chosen forms of government intervention (e.g.unneeded red tape) against the poor; and a tax, transfer, and public goods system that lowers inequality while not significantly lowering incentives to work or pursue business enterprise.An integrated set of policies to promote the demand for labour, key to raising incomes on a broad basis, uses several of the above policies as its building blocks.Unfortunately the potentially most powerful of these policies-land reform/small farm support systems, a generally strong policy to create labour demand, and a strongly redistributive state are, in the first case politically untenable in most countries, and in the others difficult to achieve for technical as well as political reasons.