Democracy, redistribution, and Equality*

The article argues that economic inequality inevitably generates political inequality, which in turn reproduces economic inequality. Basic concepts are introduced first along with strong caveats concerning the quality of the cross-national data on income distributions; historical patterns of income inequality are summarized next, and with these preliminaries, a distinction is made between redistribution of consumption at a particular time and equalization of income earning capacities over time. Following this economic considerations, the article discussion moves to political factors that may block redistributions.


introduction
S oon after Luiz Inacio Lula da Silva was first elected the president of Brazil, I had a conversation with the secretary of the Workers' Party (PT) in a favela of Rio de Janeiro.When I asked my interlocutor what the newly elected government should do, his response was immediate "redistribute.""Ok," I said, "the government should increase taxes on the rich, but what should it do with the revenue?".His answer was still "redistribute."This is how far we got.And I wondered ever since what his answer could have and should have been.
What does it mean to "redistribute"?While redistribution of income continues to constitute the standard slogan of the Left around the world, specific programs do not go far beyond the silence of my interlocutor.When the idea of redistribution first appeared in modern history, in England during the seventeenth century, its meaning was clear for the asset to be redistributed was land. 1 Attacking the Levellers, Harrington's (1977, 460) claimed that "By levelling, they who use the word seem to understand: when a people rising invades the lands and estates of the richer sort, and divides them equally among themselves."Now, land can be carved into pieces and redistributed, thus equalizing the capacity to earn incomes.But what can be redistributed today, in economies in which most production is specialized and socialized, involving cooperation of many in large units constituted by modern firms?Moreover, what can be redistributed and how so as to reduce inequality of income earning capacities?
These are the questions addressed below.I should warn that this text is not more than a summary of the current state of knowledge, offering few if any answers.Persistent inequality is a central puzzle in economics and in political economy: articles that ask "Why the poor do not take it away from the rich?" appear almost every year in academic journals (Putterman 1996;Roemer 1998;Lind 2005;Huber and Stanig 2010).Note that "why not?" questions are notoriously difficult to answer indeed, they are unanswerable unless there are good reasons to expect that the answer should be positive.But in this case the reasons are good: after all, when the majority of the people are poor and when political decisions are made by majorities, incomes should be equalized.Yet they rarely are.All I can do is to list and add to extant explanations.But because there are so many explanations, choosing among them is not a simple matter.
The paper is organized as follows.Basic concepts are introduced first along with strong caveats concerning the quality of the cross-national data on income distributions.Historical patterns of income inequality are summarized next.With these preliminaries, a distinction is made between redistribution of consumption at a particular time and equalization of income earning capacities over time, with a particular emphasis on the technical difficulties of redistributing produtive assets.Following this purely economic considerations, the discussion moves to political factors that may block redistributions.Specifically, I argue that economic inequality inevitably generates political inequality, which in turn reproduces economic inequality.

Preliminaries: Concepts and Data
Figure 1 is a picture of a distribution of income among individuals (or house-holds).The horizontal axis gives the logarithm of income, measured in 1000's (of whatever currency), while the vertical axis gives the density (relative frequency) of recipients with such incomes.This picture is deliberately designed to represent a typical distribution: there is a large mass of people with low incomes and a fat, long tail that represents recipients of high incomes.
Democracy, redistribution, and Equality bpsr (2012) 6 (1)  13 11 -36 The first vertical line shows the median income, while the second line represents the mean.
Note that the median is lower than the mean -a pattern observed in all known income distributions -which means that the majority of individuals would benefit from incomes becoming more equal.The ratio of the median to the mean income gives one summary of income inequality: the lower this ratio, the higher the inequality (if the distribution is lognormal this ratio is uniquely related to the variance of the distribution).recipients, which will be written below as Q5/Q1.There are several other measures, but these will suffice here.Now, before illustrating these statistics and making cross-national and over-time comparisons, a strong warning is required about the quality and the comparability of the available data.Even in large household surveys, the few very large and the many very small incomes escape the sampling frame.The probability that a billionaire would be included in the sample is almost zero, while many poor people cannot be found.As a result, it is estimated that the income reported in household surveys covers between 60 and 80 percent of income derived from national accounts, which means that even large surveys underestimate the extent of inequality (Córtes 2000, 246-51;Cornia, Atkinson, and Kliski 2004, 31).International and over time comparisons present an even greater problem because of different definitions and methods used.Some data concern individuals while other concern households some report incomes, other expenses, while still other consumption.Most data are based on national household surveys but some do not provide complete territorial coverage and some are based on tax rolls.Finally, the coverage of particular countries is very unequal: the number of annual observations per countries varies from 1 to 45.Given these differences, perhaps it would be prudent just to give up international comparisons based on such data sets (Atkinson and Brandolini 2001).But the lure is irresistible, so all I can do is to warn the reader against believing too much any of the numbers that follow.
Below I use two compilations of cross-national over-time data.One, to which I refer as "WDI+," is based on the most recent edition of World Bank's World Development Indicators (originally collected by Deininger and Squire 1996) and is augmented by whatever numbers could be found anywhere to extend the coverage.Hence, this series is highly heterogeneous with all the issues listed above.The second data set is taken from SWIID (Salt 2011), which is an attempt to address the issues of heterogeneity by constructing homogeneous series using a complex algorithm of multiple imputation.Whether this attempt is successful is doubtful. 2 With these caveats, here is the distribution of the median/mean ratios based on WDI+. 3 For future reference, note that this distribution has two peaks.The ratio of 0.7 means that a distribution is highly unequal, while the ratio of 0.9 indicates that it is quite equal.
To provide intuition about numbers, here are some illustrative values for selected Analyzing historical patterns of income distribution lends itself to several generalizations: (1) Income distributions appear to be stable over time.The strongest evidence, albeit for a relatively short period, comes from Li, Squire, and Zhu (1997), who report that about 90 percent of total variance in the Gini coefficients is explained by the variation across countries, while few countries show any time trends.Earned incomes show almost no variation during the twentieth century (Piketty 2003).Hence, inequality and equality persist over time.Top income shares (those of the top 1 percent of recipients), however, have been highly volatile over the long run, with significant consequences for overall inequality (Atkinson, Piketty and Saenz 2011). 4  (2) Increases in inequality appear to be much more rapid than its declines.Particularly after 1982, some increases of inequality have been dramatic.In Poland, where distribution was quite egalitarian under communism, the ratio of the median to the mean income was 0.82 in 1986, while in Mexico in 1989 it was 0.59.By 1995 this ratio in Poland was 0.62, almost the same as in highly unequal Mexico.In the United States, income inequality hovered around a constant level until about 1970 and then increased sharply (Bartels 2008, 35).(3) It seems that no country rapidly equalized market incomes without some kind of cataclysm: destruction of large property as a result of foreign occupation (Japanese in Korea, Soviet in Eastern Europe), revolutions (Soviet Union), wars, or massive emigration of the poor (Norway, Sweden).The top income shares were particularly effected by the two world wars and the depression of 1929 (Atkinson, Piketty and Saenz 2011).
In sum, it may well be that income inequality tends to increase as a result of the operation of markets -which would constitute a "Newtonian" law of income distribution -unless governments actively counteract this tendency and then it simply persists, unless some cataclysmic events intervene.

redistribution Changing distributions
Consider first a change of income distribution: Distributions can change because of changing market conditions, for example, increasing or falling demand for some skills.They can change as an unintended by-product of some policies, such as regulation of monopolies (Peltzman 1976).They can be affected by the "cataclysms" discussed above.Or the change can result from deliberate redistributive policies.Isolating the effect of all these potential factors is difficult, as attested by the controversies over the relative impact of changing demand for skills, technological change, or fiscal policies.
Just as an illustration of this difficulty, consider long-term patterns of income distribution, as characterized by the median/mean ratio in the United Kingdom and the United States.The United States seems to have had a much more equal distribution in early 19th century but a massive emigration of the poor, mainly from Ireland, reduced the inequality in the U.K. while the massive immigration of unskilled workers to the U.S.
increased it.The two world wars and the economic crisis of 1929 had a powerful impact on equalizing the distributions in both countries.The equality persisted in the aftermath of World War II but was quickly eroded under the impact of neo-liberalism in both countries.I have no idea whether this is the correct story but it serves to show that different factors can be at play at different times and that telling them apart is not an easy task.To put it differently, one should not assume that equalization is necessarily an effect of redistribution.

redistribution of current consumption
The central and complicated issue is whether redistribution of consumption has the effect of equalizing income earning potentials.Almost the entire literature in political economy focuses on the redistribution of current consumption through taxes and transfers ("the fisc"). 5The story goes as follows: Individuals who work or own capital assets receive "gross," "market," "factor," or "pre-fisc" incomes.These incomes are taxed and then transferred in the form of private income or public services.After the transfers, individuals receive "net" or "post-fisc" incomes or consume equally valued public goods.The rate at which incomes are taxed and transferred is the "rate of redistribution." 6 Redistribution of current consumption is often an urgent task, necessary to protect people from destitution.It is costly -it bears administrative expenses and may affect incentives to invest -but these costs are minor.The Mexican Progressa, the Brazilian Bolsa Família, the Argentine universal child subsidies and old-age pensions made a big difference in the lives of millions of people at a negligible cost. 7 Just calculate: if the ratio of the incomes of the incomes of the top to the bottom quintile is 26 (in Brazil in 1989 the share of Q5 was 0.652 and of Q1 0.025), taxing the incomes of the top 20 percent of recipients at an additional 4 percent would double the incomes of the bottom 20 percent.Even if this tax would generate inefficiency, the marginal effect would be minor and the welfare consequences enormous.
Yet what is the impact of current redistribution on the future inequality of incomeearning potentials?Note that unless current redistribution increases the capacities of the poor to earn incomes, it must be repeated again, year after year, as in Marx's concept of "simple reproduction.".This is why this question is crucial.
In principle, the answer should be simple to obtain: we could analyze statistically the impact of current income transfers or of current social policies on the future distribution of gross incomes.But given the poor quality and the scarcity of data, not much credence can be given to such results.Hence, we can only speculate.
One may expect that redistributive programs that consist exclusively of transfers to households of money or in kind differ in their effects from programs that imitate the Mexican Progressa by also including access to health and educational services.Consider first transfers to poorer individuals or households.Even when these transfers are in the form of money, hence potentially allowing saving and investment, there are good reasons to think that they are entirely consumed. 8Hence, as the result the rich consume (or invest) less, the poor consume more, but nothing else changes, so that the inequality of income-earning potentials remains the same the next year.With a reminder about all the caveats concerning the data, it appears that if anything, a pure redistribution of current consumption during a particular year increases the inequality of gross incomes as measured by the Gini coefficients. 9 I am not quite willing to believe this result and I cannot think of a plausible mechanism that would generate it. 10But the conservative interpretation of this pattern is that redistribution of current consumption has no effect on the distribution of future gross incomes, which is sufficient to support the theoretical claim.Social policies that include access to health services, education, and sometimes housing present more complicated issues.It may well be that having better health makes someone able to work 48 hours per week instead of 30 or that having completed ten years of education makes someone more productive than having completed six.But, first, the magnitude of these policies may be insufficient to pull people out of poverty and, second, the effect of these increased productive capacities on income depends on the general economic situation.Banarjee and Duflo (2011) provide evidence that the rate of return to investment is very high when the amount of investment is very small, but it is low when investment is somewhat larger, to become high (and declining) again when investment is large.This implies that even if the productive capacities of the poor increase somewhat, the rate of return does not warrant further investment, so that poverty remains to be a trap.In turn, there is evidence that while the effect of Oportunidades was to increase the human capital of children in comparison to their parents, the kids do not find employment given the depressed market conditions.Moreover, note that if everyone increases their potential, the general equilibrium effect may be that wages remain the same: one of the big puzzles of development economics is that individual rates of return to education are high but aggregate levels of education do not affect aggregate growth rates.
Hence, while some social policies may have productivity enhancing effects 11 , the scale of these policies is most likely insufficient and their effects depend on the general state of the economy, in particular on the demand for the particular assets and skills.
Instead of expecting that equalization of income-earning potentials would result as a byproduct of redistributing consumption, the inequality of income-earning potentials must be attacked directly.
Thinking about redistribution cannot be limited to a focus on taxes and transfers.All policies pursued by governments affect income distribution.While it may make sense to think of "market," "pre-fisc" incomes, there is no such thing as "pre-government" incomes.
Incomes are earned in markets but markets are constructed by the state (Przeworski 2003).All policies affect the distributions of income.Not only all policies affect distributions of incomes but some policies concentrate incomes while other policies equalize them, and probably all governments pursue such policies simultaneously.This is why the net effect of public policies on redistribution of income is impossible to determine.The counterfactual cannot be an economy in which there is no government: this is a figment of the imagination of some economists.Moreover, partial equilibrium effects -effects of particular policies in the presence of other policies taken as fixed -are difficult to unravel.

redistribution of capacities to earn incomes
What policies, then, would have the effect of equalizing income earning potentials, the capacities to earn incomes?Because incomes are generated by efforts applied to productive assets -whether land, physical capital, education, or skills -to equalize the capacities to earn incomes, we must think in terms of distribution of these assets.
But what assets can be equalized in modern societies?When the idea of equal property first appeared productive assets meant land.Land is relatively easy to redistribute.It is enough to take it from some and give it to others.Hence, agrarian reforms were frequent in history of the world: there were at least 175 land reforms entailing redistribution between 1946 and 2000 alone.But today the distribution of land plays a relatively minor role in generating income inequality.In turn, other assets resist such a simple operation: (1) Communists redistributed industrial capital by placing it in the hands of the state and promising that uninvested profits would be equally distributed to households.Although this system generated a fair degree of equality, for reasons that cannot be discussed here, it turned out to be dynamically inefficient: it inhibited innovation and technical progress.
(2) Alternatively, one could redistribute titles to property in the form of shares.But this form of redistribution has problems of its own.One is, that as the Czech privatization experience shows, they could be and likely would be quickly reconcentrated.People who are otherwise poorer would sell them to those who are wealthier.
(3) Many countries equalized human capital by investing in education.Not only is this process slow, but moreover people exposed to the same educational system acquire different income earning capacities as a function of their social and economic background.
(4) Finally, income earning capacities can be generated by policies that are narrowly targeted on increasing the productivity of the poor ("pro-poor growth"), such as relaxing credit constraints, training for specific skills, subsidizing the necessary infrastructure, focusing on diseases to which poor people are most vulnerable, etc. 12 Such policies, however, require a high level of administrative competence to diagnose the needs and to target the policies.Moreover, they can be easily used for clientelist purposes.
Finally, even if productive assets were equalized, perfect equality cannot be sustained in a market economy.Suppose productive assets had been equalized.But individuals have different and unobservable abilities to transform productive assets into incomes.Moreover, they are subject to vicissitudes of luck.Assume that particular individuals (or projects they undertake) are subject to slightly different rates of return: some lose at the rate of -0.02 per annum and some gain at the rate 0.02.After twenty-five years, the individual who generates a 2 percent return will be 2.7 times wealthier than the individual who loses 2 percent per year, and after fifty years (say from the age of 18 to 68) this multiple will be 7.4. 13Hence, even if productive assets were to be equalized, inequality would creep back in. 14  None of these difficulties imply that governments are unable to counteract the unequal distribution of wealth and the consequences of this inequality for the distribution of current consumption.Benhabib, Bisin, and Zou (2011)  and Saenz (2011) provide historical evidence that progressive incomes taxes prevented the reemergence og grand fortunes after they were destroyed in the 1914-1945 period.
Yet equalizing productive assets seems to be difficult for purely technological and administrative, not just political or economic, reasons.It may well be that when and where they did occur, reductions of inequality of gross incomes were due to the gradual removal of barriers to access by the poor to the use of productive resources they already commanded rather than to distribution of productive assets (Przeworski 2012b).These barriers historically included attachment to land, monopolies and monopsonies, guilds, access to professions and occupations, access to education, protection of particular technologies, and access to credit.Banerjee and Duflo (2011) report that the cost of credit continues to be much higher for the poor around the world.Hence, some policies of "market liberalization", which acquired a deservedly bad reputation as the result of the failure of neoliberal policies, may in fact have equalizing effects on the access of the poor to productive resources.

Democracy and redistribution
Let me begin with a fact, as always to be taken with some dose of skepticism because of the quality of the data: income distributions (of gross income as measured by the quintile ratio) do not seem to be more equal under democracy than under other regimes. 15Where the gray areas in this figure overlap, inequality is not statistically different under the two regimes.Note that the low inequality in wealthier autocracies is due exclusively to Singapore, while the increase of inequality in developed democracies is due to the United States and Switzerland.Statistical analyses that control for non-random selection of regimes confirm that the two regimes do not differ on the average (Przeworski 2011) Now, this fact is surprising because one can expect that democracy, via political equality, must lead to economic equality.Indeed, at some moment, political and economic equality became connected by a syllogism: Universal suffrage, combined with majority rule, grants political power to the majority.And because the majority is always poor, it will confiscate the riches.The syllogism was perhaps first enunciated by Henry Ireton in the franchise debate at Putney in 1647: "It [universal male suffrage] may come to destroy property thus.You may have such men chosen, or at least the major part of them, as have no local or permanent interest.Why may not these men vote against all property?"(quoted in Sharp 1998, 113-14).It was echoed by a French conservative polemicist, J. Mallet du Pan, who insisted in 1796 that legal equality must lead to equality of wealth: "Do you wish a republic of equals amid the inequalities which the public services, inheritances, marriage, industry and commerce have introduced into society?You will have to overthrow property" (quoted in Palmer 1964, 230).James Madison warned that "the danger to the holders of property can not be disguised, if they are undefended against a majority without property." 16  Conservatives agreed with socialists 17 that democracy, specifically universal suffrage, must undermine property.The Scottish philosopher James Mackintosh predicted in 1818 that "if the laborious classes gain franchise, a permanent animosity between opinion and property must be the consequence" (Collini, Winch and Burrow 1983, 98).David Ricardo was prepared to extend suffrage only to "that part of them which cannot be supposed to have an interest in overturning the right to property."(Collini, Winch and Burrow 1983, 107).
Thomas Macaulay (1900, 263) in the 1842 speech on the Chartists vividly summarized the danger presented by universal suffrage: The essence of the Charter is universal suffrage.If you withhold that, it matters not very much what else you grant.If you grant that, it matters not at all what else you withhold.If you grant that, the country is lost... My firm conviction is that, in our country, universal suffrage is incompatible, not only with this or that form of government, and with everything for the sake of which government exists that it is incompatible with property and that it is consequently incompatible with civilization.
Nine years later, from the other extreme of the political spectrum, Karl Marx (1952Marx ( [1851]], 62) expressed the same conviction that private property and universal suffrage are incompatible: 11 -36 The classes whose social slavery the constitution is to perpetuate, proletariat, peasantry, petty bourgeoisie, it puts in possession of political power through universal suffrage.And from the class whose old social power it sanctions, the bourgeoisie, it withdraws the political guarantees of this power.It forces the political rule of the bourgeoisie into democratic conditions, which at every moment jeopardize the very foundations of bourgeois society.From the ones it demands that they should not go forward from political to social emancipation from the others they should not go back from social to political restoration.
According to Marx, democracy inevitably "unchains the class struggle": The poor use democracy to expropriate the riches the rich are threatened and subvert democracy, by "abdicating" political power to the permanently organized armed forces.The combination of democracy and capitalism is thus an inherently unstable form of organization of society, "only the political form of revolution of bourgeois society and not its conservative form of life" (Marx 1934(Marx [1852]], 18), "only a spasmodic, exceptional state of things (...) impossible as the normal form of society" (Marx 1971(Marx [1872], 198)], 198).
And yet democracy turned out to be compatible not only with private property of productive assets but also with a fair dose of income inequality.Contrary to all the predictions, universal suffrage did not result in confiscation of property, not even in equalization of incomes.Why not?
Because the issue is burning, explanations abound. 18I can only list generic varieties: (1) One class of explanations maintains that for a variety of reasons the poor do not want to equalize property, incomes, or even opportunities.The reasons come in several variants: (1.1) False consciousness due to a lack of understanding of the distinction between productive and non-productive property.
(1.2) Ideological domination due to the ownership of the media by the propertied (Anderson 1977).
(1.4) Expectations of the poor that they would become rich (Benabou and Efe 2001).
(1.6) Poor information about the effects of particular policies even among people holding egalitarian norms (Bartels 2008).
(1.7) The belief that inequality is just because it is a consequence of efforts, rather than of luck (Piketty 1995).
(2) Another variety of explanations claims that even if a majority holds egalitarian norms, formal political rights are ineffective against private property.Some distinctions are again relevant: (2.1) Wealthy people occupy positions of political power, which they use to successfully defend themselves from redistribution (Miliband 1970;Lindblom 1977).The "power elite" is the same as the economic elite.
(2.2) The democratic institutions are in fact not majoritarian but super-majoritarian: more than a simple majority is required to alter the status quo.Unless the two houses of the legislature are identically elected, bicameralism effectively generates a super-majority rule (Przeworski 2010). 19Executive veto power provides another countermajoritarian device.
Non-majoritarian institutions, such as constitutional courts or independent central banks, are not supposed to follow simple majorities.
(2.3) Political projects of radical redistribution may meet violent op position, including military interventions (Przeworski 2012a).It is not easy to explain why the military would intervene in defense of an inegalitarian status quo but the instances in which they did -Chile in 1973 is a prominent example -are familiar to everyone.
(2.4) Independently of their class composition, governments of all partisan stripes must anticipate the trade-off between redistribution and growth.Redistributing productive property or even incomes is costly to the poor.Confronting the perspective of losing their property or not being able to enjoy its fruits, property owners save and invest less, thus reducing future wealth and future income of everyone.This "structural dependence on capital" imposes a limit on redistribution even on those governments that want to equalize incomes (Przeworski and Wallerstein 1988).
None of these explanations remains unscathed when exposed to counter-arguments and to evidence.Personally, I am not taken by the claim that the poor would not want to live better, even if it were at the expense of the rich. 20I am more inclined to believe that the constraints on redistribution are structural, both institutional and economic.

Economic inequality, political inequality, economic inequality
One reason income distributions persist may be that economic inequality generates political inequality, which in turn reproduces economic inequality. 21  Decisions made by governments affect the welfare of particular groups and individuals.
Hence, it is only natural that those whose well-being would be influenced by these decisions seek to influence them.Indeed, the essence of democracy is that citizens exert influence over governments by freely exercising their equal rights to participate in elections, public speech, peaceful demonstrations, and other forms of political activity.But in any market society, the resources which the participants can bring to the competition for political influence are unequal.Democracy is a mechanism that treats all participants equally.But when unequal individuals are treated equally, their influence over collective decisions is unequal.Imagine a game of basketball.There are two teams, perfectly universalistic rules, and an impartial referee to administer them.But one team consists of players who are 2 meters tall and the other of people who barely exceed 1.6 meter.The outcome of the game is predetermined.The rules of the game treat everyone equally, but this only means that the outcome of the game depends on the resources participants bring to it.Equality of rights is not sufficient to sustain the equality of political influence in economically unequal societies.
This observation is almost as old as democracy itself.Already in 1844 Marx characterized the duality between universalistic rules and unequal resources as follows: The state abolishes, in its own way, distinctions of birth, social rank, education, occupation, when it declares that birth, social rank, education, occupation, are non-political distinctions, when it proclaims, without regard to these distinctions, that every member of the nation is an equal participant in national sovereignty (...) Nevertheless the state allows private property, education, occupation to act in their way -i.e., as private property, as education, as occupation, and to exert the influence of their special nature.
This duality was repeatedly diagnosed ever since.The idea that political equality is not possible without social and economic equality was the cornerstone of Social Democracy.
Jean Jaures (1971, 71) thought that "The triumph of socialism will not be a break with the French Revolution but the fulfillment of the French Revolution in new economic conditions," while Eduard Bernstein (1961) saw in socialism simply "democracy brought to its logical conclusion."Yet economic and social inequality persist and their impact on political inequality continues to be a burning issue of democracy.
Tracing the impact of economic resources on democratic politics and the resulting policy outcomes is not a simple task for several reasons: (1) The point of departure must be that this impact is to some extent inherent in the capitalist economic system in which the decisions affecting the entire society, primarily those concerning investment and employment, are a prerogative of private owners of productive resources.The democratic process, even when it operates perfectly, is constrained by these private decisions.
(2) Socioeconomic inequality may cause political inequality without any expenditures of money or other costly efforts by wealthy individuals or groups if either poverty or inequality directly affect political participation of lower income groups.Except for the United States, and to a lesser extent France and Switzerland, the rates of electoral participation do not differ much by income and education (see Przeworski 2010, 93-94 for a summary of findings).Yet Salt (2006) found that among the countries for which data on inequality are available from Luxembourg Income Studies, greater inequality depresses political interest, the frequency of political discussions, and electoral participation of all but the most affluent income quintile.Hence, economic inequality may be sufficient to generate political inequality without any actions by special interest groups.
(3) The impact of money on politics cannot be reduced to "corruption."True, corruption scandals abound: suitcases filled in cash are found in the prime minister's office, government contracts are awarded to firms co-owned by government ministers, public officials exit politics to cushy jobs in private companies they favored, insider trades are rampant, political parties are found to have bank accounts in Switzerland, local governments operate systematic bribe schedules on contractors, the list goes on and on.Moreover, such scandals are by no means limited to less developed countries or to young democracies: these examples are drawn from Germany, Spain, France, Italy, United States, and Belgium.
But reducing the political role of money to instances of "corruption" is deeply misleading.
Conceptualized as "corruption," the influence of money becomes something anomalous, out-of-ordinary.We are told that when special interests bribe legislators or bureaucrats, democracy is corrupted.And then nothing needs to be said when special interests make legal political contributions.In order to exist and to participate in elections, political parties need money.Because election results matter for the private interests, they understandably seek to befriend parties and influence results of elections.The logic of political competition is inexorable.That the same acts are legal in some countries and illegal in other systemssome U.S. political financing practices would constitute corruption in several democracies -is in the end of secondary importance.The influence of money on politics is a structural feature of democracy in economically unequal societies.
(4) The information about the uses of money in politics is scant.To some extent this lack of knowledge is due to the very nature of the phenomenon: legally or not, money infiltrates politics in ways that are intended to be obscure.A general conclusion of surveys conducted in twenty-two countries by the National Democratic Institute for International Affairs (Bryan and Baer 2005, 3)  (5) Even when the information is available, the causal impact on money on the electoral process, on legislative outcomes, and on bureaucratic and regulatory decisions is difficult to identify because the direction of causality is often unclear.Estimates of these effects vary broadly across different studies, perhaps due to their different methodological designs.Even if methodological problems, in particular determining the directions of causality, are formidable, there is widespread statistical evidence as well as conclusions from case studies in several countries to the effect that (1) in elections, money is more productive for challengers than for incumbents, it matters most in close races, and has a significant effect in open races (those in which there is no incumbent), (2) lobbies have a powerful effect on legislation, (3) special interests influence regulatory decisions and their implementation (See Przeworski 2011 for an overview of evidence).Money has endless ways to infiltrate politics.When limits are imposed on political advertising by candidates, advertising is conducted by "independent" groups in favor of positions that the candidates are known to stand for.When corporations are prohibited from contributing to candidates, their employees are urged to do it as individuals.When regulation caps contributions to political parties, special interest groups spend more money trying to persuade voters directly (Hogan 2005).
It should not be taken as obvious, however, that the impact of money on democratic politics must have the effect of preserving or increasing in equality.Most competition for political influence is competition among interest groups (Becker 1983) and the particular interests may benefit from policies that have equalizing effects: construction companies want governments to build schools or public housing, pharmaceutical companies want governments to finance access to medicines.Moreover, companies that cannot compete because of barriers to entry and other market rigidities want to remove protection of various types of monopolies.Hence, some special interests lobby for policies that have equalizing effects.But they share the interest in keeping unions weak, wages low, and redistribution of consumption low.
The relation between money and politics can be to some extent mitigated so that the impact of economic inequality on political inequality varies across countries.Various Something is wrong when a plurality of citizens in a democracy answer the question about which institutions have most power in their country with "banks. 23Perhaps the most plausible explanation of the persistence of inequality is the feedback from political to economic inequality.High economic inequality generates high political inequality, disproportionate political influence of the rich perpetuates the inequality. 24

Conclusions
As I warned, this paper offers no answers.With so many possible reasons it is difficult to identify a single one.Having to anticipate the decisions of private owners of productive resources is a structural feature of capitalist societies but the extent to which this structural dependence binds is not clear.Equalizing income earning potentials appears to be difficult for purely technical reasons discussed above.Which policies would increase the earning capacity of the poor is not obvious 25 and few states have the administrative capacity to implement them.I agree with Banerjee and Duflo (2011) that until we know which policies would work it makes no sense to introduce politics: the question of political economy is "Why are policies that would work not pursued?" or "What political factors would lead to policies that work?".But political factors may also play an important role.Specifically, there are grounds to think that there is a vicious circle from economic to political and back to economic inequality.
Redistribution of current consumption is an urgent need in societies in which some people are not able to earn incomes sufficient to protect them from destitution.Indeed, as the recent experience of Argentina and Brazil shows, social policies that subsidize consumption of the poor may have a powerful impact on reducing poverty.But redistribution of consumption has no discernible impact on the distribution of income earning potentials.
We need to reformulate the agenda of research concerning the distribution of income.
The almost exclusive concentration on the redistribution through taxes and transfers is misguided both intellectually and politically.What we need to understand is how distributions of earning capacities change over time, why does inequality of earning potentials persist over long periods of time.We need to focus not on "redistribution" but on equalization.And such a focus calls for examination of all public policies, not just the deliberately redistributive ones, for their impact of the distributions of future incomes."Pro-poor growth" (UNDP), "development with equity" (CEPAL) are the correct slogans: we need to think of development policies that focus on increasing the earning potentials of the poor. 2 Particularly suspicious are high Gini coefficients of gross incomes reported for the OECD, including the Scandinavian, countries.Although this data set claims to be based on the highquality Luxembourg Income Studies (LIS), the Ginis of gross income reported in SWIID are on the average seven points higher than the corresponding numbers in LIS.
3 There are four observations in which the ratio of the median to the mean has the value above 1.These observations are clearly mistakes.
4 These data are based on tax returns, so the volatility is not due to sampling errors.Moreover, top income shares show long-term trends: they peak before World War I, experience a deep decline between the two world wars, and in some countries climb steeply in the recent period.5 The central issue in this literature is whether it is true that a higher inequality of gross incomes leads to more redistribution, as the median voter model (Meltzer and Richards 1981) implies.Redistribution is measured either as a difference between Gini coefficients of gross and net incomes of individuals or households or by social spending.The findings diverge, depending on the samples, periods, and estimators.I decided not to delve into this topic because I am too skeptical about the data.
6 The data issue which emerges in assessing the rate of redistribution is the treatment of oldage pensions.They are typically considered to constitute delayed wages and as such they are included in gross income.The pension systems in most countries, however, already entail some, varying, degree of redistribution.
7 On the equalizing effects of the Argentine Assignación Universal por Hijo, see Gasparini and Cruces (2010), and of the old-age pensions Repetto and Potenza Dal Masetto (2011).
8 Bertola (1993) shows analytically that if the utility function is concave and the production function is linear in capital, individuals who have no capital endowments consume the entire wage and transfer incomes.The data confirm this result.
9 Redistribution is measured in Figure 5 as the difference between Gini coefficients of gross and net income divided by the coefficient of gross income, which the "rate of redistribution" as defined above.With fixed country effects Gini coefficients of gross income increase when there is large upward redistribution through the fisc but also when the redistribution is larger than 10.
19 Cuttrone and McCarthy (2006, 184) conclude that "If the median legislators in the two chambers differ in their preferences, then no status quo that lies between them can be defeated by two separate majority votes." Iaryczower, Katz, and Saiegh (2009) show that to pass both houses of the US Congress a bill requires the supermajority of 74.4 percent.
20 Note, moreover, that the rich may actually support redistribution when inequality has strong negative externalities on their welfare.In an analysis of 330 surveys in approximately 50 countries between 1985and 2008, Dion (2010) ) shows that the slope of the relation between household income and support for redistribution varies across countries depending on the extant degree of inequality, the extent of already effectuated redistribution, and several cultural factors.
21 An alternative causal channel is that inequality induces conflict and/or weak institutions, which in turn promote inequality.A vast literature along these lines is summarized and tested with regard to Latin America by Gasparini and Molina (2006).As they emphasize, however, the causal chains are murky and next-to-impossible to test empirically with the available data.

Figure 2 .
Figure 2. Distribution of the ratios of the median to the mean across countries and time

Figure 4 .
Figure 4. Income inequality in the UK and the US

Figure 5 .
Figure 5. Redistribution of consumption and change of Gini coefficients of gross income

Figure 6 .
Figure 6.Inequality under different regimes is that "Little is known about the details of money in political parties or in campaigns.Political party financing patterns are extremely opaque [...]".With regard to Latin America,Zovatto (2003, 10)  comments that "information about finances of political parties is scarce because the culture of transparency and the obligation of these forces to give accounts to the State and the civil society has been in general absent from the political-partisan scene of the region.".Information about political contributions is available only for a recent period in a handful of countries.Information about lobbying expenditures exists only in the United States.
regulatory schemes have been proposed and various are in use but we have no systematic knowledge of their effects.Perhaps instead of legal regulation, more effective are mechanisms by which poor people can pool their resources in order to counterbalance the influence of the rich.Unions provided this mechanism in the past and still do in some countries: income inequality is lower in countries which continue to have encompassing unions(Scheve and Stasavage 2009) 22 Non-governmental organizations now play some of this role and, as campaign has shown, perhaps the internet will provide an alternative mechanism.But perfect political equality is impossible in economically unequal societies.

Table 1 .
Some illustrative statistics of income inequality Democracy, redistribution, and Equality bpsr (2012) 6 (1) 15 11 -36 countries and dates, organized from the most equal to the most unequal distribution observed in the WDI+ data set (all the numbers concern gross incomes, m/µ stands for median/mean): Bulgaria in 1976 had the most equal distribution observed, Namibia in 1993 most unequal.Note for future reference the rapid increase of inequality in Poland and the United States.

Democracy, redistribution, and Equality bpsr
Consider a few examples: In some countries in order to practice as a nurse one must complete two years of post-secondary education and in other countries none.Clearly, nurses will have higher incomes if they must Stiglitz (1994).Taxi medallions are a monopoly created by, typically, local governments: the current price of a medallion in New York City is $750,000.Again the medallions transfer incomes from users to owners of taxis (and finance local governments).These are just minor examples but the same is true of more consequential policies: regulation of natural monopolies, regulation of labor markets, laws regarding consumer protection, environmental regulations, (....), the list is endless.Even when the state does not enter directly into private transactions, the terms of these transactions depend on public policies.Consider the example, due toStiglitz (1994), of buying car insurance against theft.Consumers pay premiums and if theft transpires receive benefits.But the amount of premiums and benefits -the terms of this private transaction between individuals and insurance companies -depend on the probability that the insured event would occur and this probability depends in turn on the number of policemen the government puts on the street.The state is present in all private transactions.

Democracy, redistribution, and Equality bpsr
Interest group activities include influencing and mobilizing the electorate, financing electoral campaigns, lobbying legislators and the executive branch, and using the judicial system.Schematically, one may think that money biases political outcomes in favor of the donors if (1) political contributions affect the platforms offered by parties in elections, Rodrik (2008) may combine removal of some barriers to access, selective industrial policies, and targeted policies oriented specifically at the productivity of the poor.Along withRodrik (2008), I do not believe that there are universal recipes, blueprints to be applied everywhere.Also following Rodrik, I believe that to be successful, policy formulation should entail "self-discovery": experimentation that entails trials and errors, with the understanding that errors are inevitable.What my interlocutor, who sparked these reflections, should have asked himself is what did his community need most urgently to escape the grip of poverty.And the answer would have probably included Bolsa Família but also access to clean water, a better primary school, security on the streets, public transportation to place where jobs were, a local branch of a bank, and jobs, jobs, jobs.But it should have also included political institutions that are responsive to such needs, mechanisms which make governments attentive to such needs.Notes1 Demands for a redistribution of land were made intermittently in Latin America, most notably in Mexico by Hidalgo and Morelos in Mexico in 1810 and Artigas in Uruguay (then Banda Oriental) in 1813.