Political Rights, Property Rights, and Economic Development* Adam Przeworski Department of Politics New York University Abstract Some part of modern economic growth must have been due to a reduction of deadweight losses associated with upward redistribution of income. In turn, upward redistribution was reduced when new groups won political rights and used them to defend their interests. I construct a model in which extensive redistribution from the poor to the rich occurs when few people enjoy suffrage, redistribution declines as new groups are enfranchised, reaches a minimum when the median income among the enfranchised equals the average income in the population, and then changes direction and increases. As a result, growth rates first increase and then decline as the enfranchised proportion of the population increases. This model is supported by econometric results covering the world from the inception of representative institutions until today. Note: This version was doctored by placing the formal model on which the results are based in the Appendix and relying on an intuitive example to motivate them. Hopefully, in this way the text is comprehensible without the mathematics. I would obviously appreciate, however, if those among you who can looked at the model. Some of the footnotes are based on the model. *I am grateful for comments to Neal Beck, Jess Benhabib, Kathy Hafer, Peter RosendorfF, David Stasavage, as well as several participants in the Workshop on Inequality and Development, Universite de Namur, Belgium, May 11-13, 2007. I appreciate the assistance of Tamar Asadurian, Carolina Curvale, Sunny Kuniyathu, Yingying Na, and Anjali Bolhken Thomas in collecting the data. This work was supported by a grant from the National Science Foundation. 1 1 Introduction Some part of modern economic growth must have been due to a reduction of deadweight losses associated with upward redistribution of income. In turn, upward redistribution was reduced when new groups won political rights and used them to defend their interests. Since economic models often require a poetic license, I analyze redistribution in terms of "taxes" and "transfers." But this is just a reduced form of representing the redistributive effect of any kind of policies. As Stigler (1975) observed, all policies - from credentialing nurses, to issuing taxi medallions, to prohibitions of noxious products - affect incomes differentially. A woman with two years of vocational education has a different earning capacity when anyone can become a nurse and when becoming one requires this training. In turn, incomes of everyone using nursing services are different when entry into nursing is open than when it is regulated. And the moment one thinks of policies other than those that pass through the fiscal system, it becomes obvious that they may have regressive, not only progressive, effects (See, for example, Peltzman 1976). Redistributions occur in both directions: upward, from those relatively poorer to the wealthier, and downward, from the wealthier to the poorer. In fact, all societies redistribute simultaneously in both directions. For example, patent laws concentrate incomes but anti-monopoly laws diffuse them, free primary education redistributes incomes downward but free tertiary education redistributes it upward, minimum wage legislation raises low wages but anti-union laws depress them. Such broad conception of redistribution is crucial for understanding the historical patterns of growth. Specifically, focusing exclusively at the tax rate in the narrow, formal sense of this term captures only one aspect of the deadweight costs of redistribution. The arguments of Just-man and Gradstein (1999), Acemoglu and Robinson (2000), as well as of Lindert (2004), that extensions of suffrage increased downward redistribution through the fiscal system generate a puzzle - the "puzzle of a free lunch" in the language of Lindert (2004: 16-19) - namely that between roughly 1880 and 1980 growth accelerated as tax rates mounted dramatically. Clearly, one possibility is that tax revenues were used to finance investment in productive resources, notably education (Saint Paul and Verdier 1996, Bourgignon and Verdier 2000, Mariscal and Sokoloff 2000, Lindert 2004), health and sanitation (Lizzeri and Persico 2004), or public goods that enter into production (Barro 1990). But another possibility is that extensions of franchise liberated productive capacities of the broad masses by removing the most onerous mechanisms of upward redistribution, so that growth accelerated in spite of an increase in 2 downward redistribution through the fiscal system. In turn, the claim that economic growth was due at least in part to a reduction of upward redistribution is not inconsistent with findings such as those of Lizzeri and Persico (2004: 710), who observed with regard to England that "the expansion of the franchise does not seem to be associated with a large redistribution of resources from the elite to the disenfranchised," of Aidt, Dutta, and Loukoianova (2001), who found that franchise extensions had no impact on the size of government in twelve European countries between 1830 and 1938, or of Galor (2004: Section 3.3.3.C), who maintains that "political reforms that took place in the 19th century had no apparent effect on education reforms over this period...." All that was needed for growth to accelerate was for policies to become less regressive. The story goes as follows: When people who owned productive property acquired political rights, their property became secure from encroachments by the monarch, they began to invest, and growth was launched (North and Thomas 1973). But barriers to development continued to be politically maintained because they generated a distribution of income beneficial for the narrow elite. Thus, Lindert and Williamson (1985: 342) observe that "since only the top economic classes had political voice and literacy in the 18th- and early 19th-century Britain, policy tended to be regressive." As political rights became more extensive, many fetters to development were removed and growth accelerated. Yet when people without productive property or reasonable prospects of acquiring it conquered political rights, they used these rights to redistribute incomes downwards and growth slowed down. The conclusions is thus that when political rights were highly restricted, redistribution was from the poor to the rich, its rate was high, and growth was slow. As political rights became more extensive, the rate of upward redistribution declined, and growth accelerated. Finally, as political rights became nearly universal, redistribution changed direction, its rate increased, and growth slowed down. Note immediately that this pattern is visible in the raw data (growth rates from Maddison 2003, suffrage from own data). The rate of growth accelerated as the extent of political rights, measured as the proportion of the population that is enfranchised, increased up to some level and then slowed down as suffrage became almost universal. 3 Suffrage and growth co W_-Isy, where a stands for "after," b for "before," y is the average income, Is indicates whether an individual is enfranchised, and 0.25t2 is the deadweight loss. In the formal model these deadweight losses originate from the effect of taxes on investment. 8 of the voter with the median income among the enfranchised is then tm = 0.5. When suffrage is extended to two additional agents, s = 5/7 and the median voter has an income of 5, his optimal rate is rM = 0.21. Hence, the redistribution rate declines as franchise is broadened. While a higher rate would generate more revenue, there is nothing to extract from the disenfranchised, and the revenue has to be now shared among five agents while the income of the decisive voter is still sufficiently high that his own tax bill would not be compensated by a larger transfer. Finally, when suffrage is universal, the median voter has an income of 3 and revenue is shared among everyone. Now the median voter is sufficiently poor to benefit from higher redistribution, and her optimal rate is again rM = 0.5. Hence, when only some among property owners decide how much to extract from others, incomes are redistributed from the poor to the rich and at a high rate. As the proportion enfranchised increases, the rate of redistribution falls. But when the proportion enjoying political rights increases further, the direction of net redistribution changes and its rate increases again. Note that the redistribution rates are the same when s = 3/7 and s = 1, but their distributional consequences are very different. Figure 2 shows how the ratio of the median income among the enfranchised to the average income in the population changes as a function of the proportion enfranchised (assuming that the distribution of income is exponential). When only few people enjoy suffrage, the median among them has assets about five times larger than the average in the entire population. When suffrage is universal, the ratio of the median to the mean is 0.69, which is about the average ratio of the income of the middle quantile to the average income in the Deininger and Squire (1996) data set. 9 5 Tl 4 — 3 2 — 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 S Figure 2 To see the behavior of the tax rate as a function of the proportion enfranchised, I calibrate the model given in the Appendix. The solid line is for the case when everyone has some productive assets, the dashed line for the case where 20 percent of the population does not. 0.07] 0.06: 0.05: 0.041. tax O.OJ 0.02 0.01 0 0.2 0.4 s 0.6 Figure 3 0.8 Tax rates decline, reach a minimum (when the voter with the median income among the enranchised has an average population income, see Proposition 2), and then increase. 10 If these tax rates appear low, it is because these are taxes on wealth. In terms of taxes on income, in a society in which everyone has some assets, the redistribution rate is about 70 percent when only 4 percent has the right to vote (United Kingdom before the 1832 reform); the redistribution rate reaches the minimum of 33 percent when suffrage reaches 73 percent and the median among the enfranchised equals the average in the population; and it is 38 percent when suffrage is universal. Finally, if the deadweight losses consist of reduced investment (see the model in the Appendix), the growth rates accelerate when suffrage becomes extended from low levels, reach a maximum, and then decline. Under an illustrative calibration of the model, they increase from 1.1 when s = 0.04, to 2.9 when s = 0.73, and decline to 2.7 when s = 1. This is, then, the central result. It says that as political institutions evolved from representing only the rich to mass democracies, redistribution, which was from the poor to the rich, declined, changed direction, and increased somewhat again. As a result, growth first accelerated and then slowed down again. As long as political rights were highly restrictive, property of the wealthy was secure and they could use their political rights to exploit everyone else, whose property or incomes were insecure. Even though those enfranchised had to be concerned about investment by everyone who also had property and about the capacity of the poor to perform labor services, the rate of exploitation was high, investment was low and growth was slow. When the middle sectors gained political rights, they could defend themselves from the exploitation by the rich while participating in exploiting the poor. Since most property was now secure, investment and growth increased. Finally, when the poor conquered rights, they could use them not only to defend themselves but also to take some incomes away from the rich. But the poor are constrained by the investment decisions of the rich (Przeworski and Wallerstein 1988, Bertola 1993), so that downward redistribution remained moderate and growth decelerated only modestly. 4 The Political Economy of Suffrage 4.1 Suffrage and Welfare When first established - in England, the United States, France, Spain, and the newly independent Latin American republics - representative government was not a "democracy" as we would now define the term, nor was it seen as such by its founders (Manin 1997, Dunn 2005). In spite of their egalitarian pronouncements, the problem of "founders," pretty much everywhere, was how to construct representative government for the propertied while protecting it from the poor. As a result, political 11 rights were everywhere restricted to wealthy males. The road from representative government to democracy took a long time to traverse. As of 1900, only one country had fully universal suffrage while seventeen enfranchised all males. Only during the second half of the twentieth century, more than 150 years after representative institutions were first established, did universal suffrage become an almost irresistible norm. Figure 4 The model implies (see Proposition 3) that all current holders of political rights had an interest in guarding them for themselves. This proposition is illustrated in Figure 5. Let sl stand for the percentile in the income distribution at which agent with relative wealth A* is enfranchised, where A* is the ratio of the wealth of i to average wealth. Consider four agents, each characterized by {A*, s1}. The agents are the Rich or {3,0.05}, the Average or {1,0.37}, the Median (in the population) or {0.69,0.5}, and the Poor or {0.22,0.8}. Figure 5 shows the function V(t*(s)) for each of these agents. (V(t*(s)) is the value of future flow of income when the proportion enfranchised remains for ever at s). 12 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 S Figure 5 Note: The timing of enfranchisement is indicated by the vertical segment of each line. The Rich has 3 times the average income and becomes enfranchised at s = 0.05. The Average becomes enfranchised when s = 0.37. The Median (in the population) has 0.69 and is enfranchised at s = 0.5. The Poor has 0.22 and is enfranchised at s = 0.8. Here then is what happens as new individuals become enfranchised: (1) Since taxes decline until the agent with the average income becomes the median among the enfranchised, everyone who is still not enfranchised benefits when others are enfranchised. Those in the bottom 27 percentiles of income distribution, however, suffer from further extension until they become enfranchised. Since this may be not visible in Figure 5, here is a snippet of the value function of the poor agent with sl = 0.8 when the 7 percent of agents with incomes higher than her's become enfranchised: 13 (2) Enfranchisement causes a discrete improvement for each newly enfranchised agent. Note that the improvement is larger for poorer agents. Political rights are a powerful equalizer: they drastically reduce the inequality of post-redistribution incomes. (3) As each one more agent becomes enfranchised, the welfare of those already enfranchised declines. This decline is small, because enfranchisement has mixed effects. Although the entrants join the incumbents in sharing tax revenue as taxes decline, investment increases and growth accelerates. In turn, when s is larger, tax rates increase, so that there is more revenue to share, but the proportion sharing this revenue increases and growth slows down. Yet this analysis was conducted in marginal terms, while the expansion of suffrage was due historically mostly to discrete changes of qualifications, rather than drift caused by increasing incomes, inflation, or increasing literacy. Larger extensions are more costly to the incumbents. Hence, incumbents should always resist further extensions. These results speak in favor of the class of stories in which franchise was extended only in response to revolutionary threats: political rights were conquered by the insurgent masses (Bendix and Rokkan 1962: 30, Przeworski and Cortes 1971, Freeman and Snidal 1982, Acemoglu and Robinson 2000, Conley and Temini 2001). In turn, there are stories in which elites granted suffrage voluntarily, in their own interest, either because additional enfranchisement has a positive externality for the already enfranchised (Justman and Gradstein 1999), or because the elite 14 prefers public goods over transfers (Lizzeri and Persico 2004), or because the elite needs to prepare for war (Ticchi and Vindigni 2006), or because a particular elite wants to obtain an electoral mandate for some economic policies supported by the disenfranchised (Llavador and Oxoby 2005). Another way to see this distinction is that in the first type of explanation suffrage is extended even though the extension would make the elite worse off than they are under the status quo, while in the second type extensions occur only if they would make the elite or at least a majority thereof better off. The model developed here implies that the incumbents of political rights are always worse off when these rights are extended. Hence, they resist extensions as long as they can. Przeworski (2006) shows statistically that extensions along class lines tended to occur under the threat of revolution, while extensions to women followed a partisan logic. 4.2 Winners and Losers Consider the proportion of the population that consists of net losers from redistribution at each level of suffrage. This proportion is composed of two groups: those who have no political rights and those among the enfranchised whose incomes are sufficiently high that a part taxed away is not compensated by the gains from exploiting the disenfranchised. The first proportion is simply (1 — s). The second proportion is exp(l/s).6 Note that while the proportion disenfranchised declines in s, the proportion that consists of wealthy losers increases from 0 when S = 1 (decisions about redistribution are the prerogative of one person, the monarch) and s ~ 0, to include all those with incomes below the mean when s = 1. Figure 6 plots these proportions, as well as their sum (thick line), which is the total proportion of losers in the population. 6The net losers among the enfranchised are those for whom A1 > 1/s. Let AL(s) be the relative income of the marginal enfranchised, where AL(s) = — logs. Then 1/s = exp(AL(s)). Hence, the proportion with incomes higher or equal to 1/s is exp(—exp(AL(s)) = exp(exp(—log s)) = .exp(l/s). 15 Now, since the poorest and the wealthiest agents are net losers, the beneficiaries are those with incomes in between. This is not quite Director's Law as Stigler (1970) thought about it: only the very wealthy gain when s ~ 0 and only those with incomes below the mean when s = 1. But if the "middle strata" comprise those who are in the middle three quantiles, they are the beneficiaries throughout most of the range of s. Figure 7 shows that the beneficiaries consist of only the wealthy until s 0.2, of the wealthy and the middle in 0.2 < s < 0.63, of only the middle quantiles in 0.63 < s < 0.8, and of the middle and the poor when 16 Beneficiaries Figure 7 Note: The vertical axis is the percentile (from the top) in income distribution. Hence, those below 0.2 are in the top quantile and those above 0.8 in the bottom. The net beneficiaries are those between the two lines. Those above the upper line are not enfranchised; those below the lower line are the enfranchised who are net losers from redistribution. Finally, the model implies that the distribution of income should become more equal as enfranchisement becomes more extensive. This consequence, however, is due to the assumption that all assets earn the same rate of return at every level of enfranchisement. Yet one could easily imagine that as methods of redistribution change, the relative rates of return to different assets change as well. Under the aristocratic systems, for example, access to public positions was conditioned on birth. When this restriction was abolished and talented soldiers could become marshals, as in Napoleon's army, the return to human capital must have increased relative to land or physical capital. Hence, I prefer to remain agnostic about the effect of enfranchisement on income distribution. Note that our knowledge of income distribution during the nineteenth century is extremely limited. Even the longest series on income distribution available today reach at most only few years before the advent of universal suffrage. Earlier data are scarce. Concentration of wealth in France (Piketty, Postel-Vinay, and Rosenthal 2006) increased from 1807 until 1914 and declined sharply afterwards. In England, the income share of the top 10 percent peaked in the second half of the nineteenth century and inequality began a secular decline (Justman 17 and Gradstein 1999: 109). Bourguignon and Morrison (2002), in turn, who reconstructed income distribution data for 33 countries show that within-country inequality increased slightly from 1820 to 1910, only to fall sharply as of 1919 and again as of 1950. These data, however, are too fragmentary to relate them to the extent of suffrage. 5 Theory and History 5.1 Econometric Evidence From the econometric point of view, the model implies that growth rates should first increase and then decline as the extent of suffrage becomes larger. Hence, the general form of the equation to be estimated is where 7 is the rate of growth of per capita income, s is the extent of suffrage, x are economic factors that affect growth and z are control variables. The data cover the longest possible period and all the political units in the world for which information is available. The growth rates are from Maddison (2003). The data concerning suffrage are based on an originally constructed data set, derived from multiple sources (Prze-worski et al. 2007). They were originally available only for years of elections and were linearly interpolated for the interim periods. Note that s is measured as the proportion enfranchised in the entire population. Altogether, the data cover, albeit unequally, 114 countries between 1821 and 2000, yielding 5,403 annual observations. The analysis is conditioned on legislatures being open. While suffrage rules are often left on paper by various kinds of dictators, they are irrelevant when the legislature is not elected. One might also want to exclude periods in which legislatures are "elected" without any competition, since under such conditions suffrage cannot be exercised as an instrument of choice. The variables I use as controls are per capita income (gdpcap) and the occurrence of an election during the particular year. Per capita income allows to control for convergence (if any). Election years tend to have lower growth. The results, based on a fixed effect model with ar(l) are, in Table 1. Models 1, 3, and 4 are conditioned only on the legislature being open (openl = 1), while Model 2 is also conditioned on the presence of opposition (opposition = 1). k I 18 Table 1: Suffrage and growth Variable Model 1 Model 2 Model 3 Model 4 s 0.0561** 0.0541** 0.0764*** 0.0465* (0.0231) (0.0232) (0.0249) (0.0267) s2 -0.0007** -0.0006** -0.0009** -0.0005 (0.0003) (0.0003) (0.0003) (0.0003) election -0.2280 -0.2704 -0.0630 0.0182 (0.1584) (0.1686) (0.1601) (0.1633) gdpcap 0.0785* 0.0580 0.0643 0.0774 (0.0416) (0.0406) (0.4107) (0.0461) pop > 20 -0.0643 0.0198 (0.4107) (0.4046) female 1.0385** (0.4449) age 0.2037*** (0.0614) constant 0.7466*** 0.9955*** 0.5255-4.0470*** (0.3506) (0.3476) (0.3959) (1.2591) max at 0.42 0.42 0.44 0.43 condition openl = 1 openl 1 en^ _ ^ Qpenl = 1 opposition = 1 N 5161 3973 3670 3536 Note: Fixed effects with ar(l) estimates. Models 3 and 4 present checks for robustness. Since s is measured to the base of the total population, the negative part of the slope may result from the presence of older, unproductive agents in the electorate. For a subset of data for which this information is available, Model 3 replicates the analysis of Model 1 controlling for for the proportion of population over the age of 20 [pop > 20). It turns out that while ageing has a slight negative effect on growth, this effect is almost zero, and suffrage still has the predicted effect. One might also think that the changes in growth rates are due to the inclusion of women or of younger voters. Model 4, therefore, replicates the analysis by including a dummy variable indicating that women could vote on the same basis as men (female) as well as the age at which people qualified for suffrage (age). Inclusion of women has a positive effect on growth rates, perhaps because of their relatively higher preference for public goods that 19 increase productivity (Lott and Kenny 1999 for the United States; Aidt, Dutta, and Loukoianova 2001 for Europe) but it does not change the effect of suffrage (not shown). In turn, lowering the voting age reduces growth and when voting age is introduced into the specification the term with square suffrage loses significance (t = —1.54;p = 0.123). But since younger people tend to be poorer, this only means that the effect of extensive suffrage is now split in two parts. The two coefficients are jointly significant with t = —3.33;p = 0.001 (the sign of age is inverted here). That the model fits to a cross-section time-series does not mean that it fits in all countries. But the overall pattern is not due to cross-sectional effects. As shown in Figure 8, among the twelve Western European countries for which a sufficient range of enfranchisement can be observed, an internal maximum of growth occurs in all but the United Kingdom; among the seven Latin American countries that satisfy this condition, growth peaks at intermediate levels of suffrage everywhere except in Chile and Uruguay. Suffrage and growth in some Western European countries Belgium Denmark France Germany 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 Proportion enfranchised Proportion enfranchised Proportion enfranchised Proportion enfranchised Italy Netherlands Norway Portugal 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 Proportion enfranchised Proportion enfranchised Proportion enfranchised Proportion enfranchised Sweden Switzerland Spain United Kingdom 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 Proportion enfranchised Proportion enfranchised Proportion enfranchised Proportion enfranchised Fractional polynomial fit. Excludes war years. Figure 8.1 20 Suffrage and growth in some Latin American countries Argentina n-1-1-1-r- 0 20 40 60 80 Proportion enfranchised Brazil ~i-1-1-1-r- 0 20 40 60 80 Proportion enfranchised Mexico n-1-1-1-r- 0 20 40 60 80 Proportion enfranchised ä to o ^> CM Peru 20 40 60 Proportion enfranchised 80 Venezuela ä to - O ^- -cnc\J -o°- C\J 4 - 20 40 60 Proportion enfranchised 80 ä to. o ^-. cnc\J ■ C\l Chile 20 40 60 Proportion enfranchised 80 fto o cnc\J o o 0) Uruguay 0 20 40 60 80 Proportion enfranchised Fractional polynomial fit. Excludes war years. Figure 8.2 5.2 Fetters (*** still incomplete) The main thesis developed above is that the acceleration of growth rates over the past two centuries was at least in part due to a gradual removal of politically induced barriers to development and that these fetters were removed as a consequence of extensions of political rights. The story told above generalizes a class of models in which, to defend their incomes, incumbent elites maintain barriers to entry, credit, technology adoption, or labor mobility (Parente and Prescott 2000, Djankov et al. 2002, Ra-jan and Zingales 2003, Acemoglu, Aghion, and Zilibotti 2004, Comin and Hobjin 2004, Llavador and Oxoby 2005, Acemoglu 2005, 2007, Ra-jan and Zeingales 2003). Perhaps the closest is Acemoglu (2007), who points out that where political power is in the hands of major producers, it protects their property rights while erecting barriers against the entry of new entrepreneurs. Acemoglu, Aghion, and Zilobotti (2004) analyze an eqilibrium in which capitalists bribe politicians to maintain anti-competitive policies. Djankov et al. (2002: 33) offers contemporary evidence that "countries with more limited government, governments more open to competition, and greater political rights have lighter reg- 21 ulation of entry even holding per capita income constant." The problem is that while econometric evidence shows that the first cause in the chain of causality the extent of suffrage, is related to the ultimate consequence, economic growth, it is next to impossible to systematize evidence about the intermediate causal chain. Ideally, one would find direct evidence, such as in this textbook of history of Portugal (De Oliveira Marques 1998: 95): In Portugal, as in revolutionary France, and as in entire Europe, extensive legislation had as its well-defined objective to liberate the rural worker and to permit an agricultural ressurgence... The tithe to the Church was abolished.... All the feudal privileges disappeared: forced labor, monopolies of furnaces and oil presses, banishment, payments to the lord or the king, etc. Other feudal traditions, such as game preserves and stud farms, were also abolished.... All this legislation was decreed between 1821 and 1823 by the parliaments elected in 1820 and in 1821...." Yet while it is obvious that many barriers were removed during the past two hundred years, the historical information is too fragmentary to relate it systematically to the extent of political rights. All I can do to support the argument is to point to some fetters that have been abolished or weakened. Not always, however, can I systematically demonstrate that they were abolished as a consequence of extended franchise. 5.2.1 Barriers to factor mobility Coercive attachment to land: Slavery, corvee, debt peonage, long term contracts, seasonal compulsory labor, anti-vagrancy laws.7 Slavery persisted in several Latin American countries, Brazil being the last one to abolish it in 1889. Freemen were attached to land through a variety of mechanisms. According to Bulner-Thomas (2003: 86-90), in Latin America "Employers were extremely reluctant to tempt workers with higher wages. Even when they did pay higher nominal wages there were often able to reduce the real cost through the operation of company stores at which workers had to redeem their wages on goods sold at inflated prices. The advance was not necessarily cancelled with the death of the debtor; instead, it could pass to his children, so the system has often been described as debt bondage. Thus coercion ... was still found in many parts of Latin America on the eve of 7Note, however, that Bardhan (1986: 71) points out that not all forms of attachment to land are necessarily inefficient. 22 the First World War. " In several Latin American countries, Indians had to provide free labor on the haciendas during harvests. Even specialized cattle-breeding workers " were victims of legislative measures that obliged them to be constantly employed, under pain of imprisonment, forced labour or recruitment into the army..." (Halperin-Donghi 1985: 315). Beginning with Denmark in 1788, France in 1791, and Spain in 1812, all these forms of coercive attachment to land were progressively abolished. The process, however, was gradual, to the point that residual forms of attachment to land persisted in Sweden until 1925. Inter-sectoral efficiency One piece of more direct evidence about the effect of suffrage is that the variance of output per worker declined as suffrage became more extensive. Output per worker differs across sectors when there are some barriers to the mobility of factors (data are from Asadurian 2007). Hence, the variance of productivity across sectors indicates the strength of such barriers. Figure 9 shows that these barriers were sharply reduced when modern parliaments were first established, under highly restricted suffrage. Yet they also continued to decline as suffrage was further extended. Suffrage and inter-sectoral efficiency co _ 0 20 40 60 80 Proportion enfranchised Fractional polynomial fit. Variance is weighted by labor force. Both variables interpolated. Figure 9 23 Protectionism 5.2.2 Barriers to entry Monopolies and monopsonies Salt was the most widespread royal monopoly, but there were also royal monopolies on sugar, tobacco, and alcohol, as well as a variety of local monopolies on mills, oil and wine presses, or furnaces. According to Morgan (1989: 23), in eighteenth century England even hunting was legislated to be an aristocratic privilege: "no pheasants for peasants." Both the British and the Dutch East India companies were protected by legal monopolies. In Peru, guano was a monopoly of a group of concessionaires. Cartels were encouraged and forced by the state in the late nineteenth century Austria and Germany. Legal monopolies were abolished in most countries as barriers to trade. In England, the Statute of Monopolies curtailed in 1624 the royal prerogative to create private monopolies in domestic trade. In the United States they became illegal after 1890 and in Britain they were neither enforceable nor illegal (Mokyr 1990: 268). In Japan, the abolition of domainal monopolies in the late eighteenth century " led to an enormous boom in small-scale production of exports such as high-quality paper" (Pomeranz 2000: 251). Guilds Commercial and craft guilds were widespread in late mediaeval period. They died early in China under the pressure of competition from peasants and were officially abolished in 1645. They were made illegal in France during the revolution, in Prussia in the 1840s, in the United Kingdom by parliamentary acts of 1814 and 1835. My impression, however, is that these changes occurred before major extensions of suffrage and that they guilds lost their monopoly power because of competition from rural industries rather than because of legislation. Access to professions and occupations Access to public service was reserved for the aristocracy. By the end of the eighteenth century, " democracy" was a slogan directed against legal recognition of inherited distinctions of social status. One of the main points of the revolutionary tract of Sieyes (1979 [1789]) was that noble birth need not be an indication of talent. "Aristocracy" was attacked as a system that promoted incompetence and corruption. "Could any further proof be required of the republican complexion of this system," wrote Madison in The Federalist # 39, "the most decisive might be found in its absolute prohibition of titles of nobility." In France, the Constituent Assembly decided that aristocratic privilege was in conflict with the very principle of popular sovereignty (Fontana 1993: 119). The Batavian (Dutch) Republic established in 1796 required voters to swear an oath to the belief that 24 "all hereditary offices and dignities" were illegal (Palmer 1964: 195). In Chile, General O'Higgins, the first Director of the State, abolished in 1818 all outward and visible signs of aristocracy (Collier and Sater 1006: 42). The legal discrimination under which those of mixed race suffered during colonial times was abolished in most Latin American countries (Halperin-Donghi 1985: 322). Yet legal restrictions of access to occupations or professions are not the only mechanism that causes a misallocation of talents (Murphy, Shleifer, and Vishny 1995). All barriers to factor mobility, entry, or credit may result in mismatching skills and opportunities. Overall productivity may be low since some irrelevant personal attributes or personal connections will dominate the choice of professions. 5.2.3 Barriers to credit Financial development In personalistic credit markets, lending depends on the identity of the lender and the borrower, not on the merits of a project. Rajan and Zingales (2003) argue that incumbents oppose financial development because it breeds competition. Unfortunately, data on financial development are almost non-existent before World War I and still rare before World War II. Yet among the sixteen countries for which data on financial development (from Rajan and Zingales 2003) and on suffrage overlap in 1913, the relation between them seems to be positive. (*** Find stories of reforms) 25 Suffrage and financial development as of 1913 • unitedkingdom • belgium • f rance • Canada • nether •japan ved^n germany • denmark • brazil • chile »argentina# italy • norw ay ~i-1-1-1-1- 0 20 40 60 80 Proportion enfranchised Souce: Rajan and Zingales (2003). Australia is excluded as the only country w ith female suffrage Figure 10 Land as collateral If the poor are credit constrained when they have no collateral, ownership of land gives them access to credit. According to data collected by Thomas (2005), there were at least 175 land reforms entailing redistribution in the world between 1946 and 2000 and, at least during the earlier period, land reforms were often associated with franchise extensions. Since a long historical series on the proportion of farms that were owned and operated by family units ("family farms") is available from Vanhanen (1996), we can relate land ownership to the extent of suffrage. As Figure 10 shows, this relation is positive and very tight. 26 Suffrage and Family Farms 00 - Proportion enfranchised, lagged Fractional polynomial fit. Figure 11 5.2.4 Barriers to technology As Banerjee and Dumo (2004: 29) observe, firms may not choose the latest technology either because governments do not protect investors too much or because they protect them too much. Parente and Prescott (2000) offered the classical argument to the effect that technology adoption can be slowed down by the resistence of workers it would displace. Comin and Hobjin (2004) argue that those who operate the extant technology resist innovations that would make it obsolete and show that technology adoption is faster in countries that have a formal political structure (chief executives with formal titles) and slower in countries that have effective legislatures. Other evidence is largely anecdotal but vivid (See several examples in Mokyr 1990). 5.2.5 Summary These are just scattered illustrations, but the overall picture is manifest: various kinds of legally enabled forms of upward redistribution were abolished or mitigated as new groups became enfranchised. And since these mechanisms appear prima facie to have had nefarious consequences for development, growth accelerated. The effects of increasing 27 factor mobility, opening the access to professions and occupations, abolishing legally supported monopolies, opening access to credit and to new technologies must have contributed significantly to increase investment and productivity. Which of these reforms were a direct consequence of extended suffrage, however, remains not quite clear. 6 References Acemoglu, Daron. 2005. "Modeling Inefficient Institutions." Ms. Department of Economics, MIT. Acemoglu, Daron. 2007. "Oligarchic Versus Democratic Societies." Acemoglou, Daron, and James A. Robinson. 2000. " Why did the West Extend the Franchise? Democracy, Inequality and Growth in Historical Perspective." Quarterly Journal of Economics 115: 1167-1200. Acemoglu, Daron, Philippe Aghion, and Fabrizio Zilibotti. 2004. "Distance to Frontier, Selection, and Economic Growth." Ms. Department of Economics, MIT. Aidt, Toke, Jayasri Dutta, and Elena Loukoianova. 2001. "Democracy comes to Europe: Franchise Extensions and Fiscal Outcomes 1830-1938." European Economic Review. Asadurian, Tamar. 2007. Economic Turning Points and Major Political Events. Ph.D. Dissertation. Department of Politics, New York University. Bardhan, Pranab. 1986. "Marxist ideas in development economics: an evaluation." In John Roemer (ed.), Analytical Marxism. Cambridge: Cambridge University Press. Pages 64-78. Barro, Robert. J. 1990. "Government Spending in a Simple Model of Endogenous Growth." Journal of Political Economy 98: S103-S126. Bates, Robert H. 1984. Markets and States in Tropical Africa: The Political Basis of Agricultural Policies. Berkeley, CA: University of California Press. Bauer, Peter. 1987. "Marketing Boards." In John Eatwell, Murray Milgate, and Peter Newman (eds.), Economic Development. New York: W.W. Norton. Pages 232-237. Bendix, Reinhard, and Stein Rokkan. 1962. "The extension of national citizenship to the lower classes: a comparative perspective." Paper presented at the Fiftth World Congress of Sociology, Washington, D.C. Benhabib, Jess, and Adam Przeworski. 2006. "The political economy of redistribution under democracy." Economic Theory. Special Issue. Bertola, Giuseppe. 1993. "Factor Shares and Savings in Endogenous Growth." American Economic Review 83: 1184-1198. Bourguignon, Frangois, and Thierry Verdier. 2000. "Oligarchy, democracy, inequality, and growth." Journal of Development Economics 62: 28 285-313. Bourguignon, Frangois, and Christian Morrisson. 2002. "Inequality among World Citizens, 1820-1992." American Economic Review 92: 727-744. Bulner-Thomas, Victor. 2003. The Economic History of Latin America Since Independence. Second Edition. Cambridge: Cambridge University Press. de Lolme, J.L. 2002 [1776]. The Constitution of England. Delanco, N.J.: Legal Classics Library. de Meza, David, and J.R. Gould. 1992. "The Social Efficiency of Private Decisions to Enforce Property Rights." Journal of Political Economy 100: 561-580. Collier, Simon and William F. Sater. 1996. A History of Chile, 1808-1994- Cambridge: Cambridge University Press. Collini, Stefan, Donald Winch and John Burrow. 1983. That Noble Science of Politics. Cambridge: Cambridge University Press. Comin, D., and B. Hobjin. 2004. "Cross-country technology adoption: making the theories face the facts." Journal of Monetary Economics 51: 39-83. Conley, John P., and Akram Temini. 2001. "Endogenous Enfranchisement when Group Preferences Conflict." Journal of Political Economy 109: 79-102. Deininger, Klaus, and Lyn Squire. 1996. "A New Data Set Measuring Income Inequality." World Bank Economic Review 10: 565-591. Djankov, Simeon, Rafael de la Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2002. "The Regulation of Entry." Quarterly Journal of Economics 118: 1-37. Dunn, John. 2005. Democracy: A History. New York: Atlantic Monthly Press. Easterly, William, and Ross Levine. 2002. "Tropics, Germs, Crops: How Endowments Influence Economic Development." Working Paper 9106. Cambridge, MA: National Bureau of Economic Research. Fontana, Biancamaria. 1993. "Democracy and the French Revolution." In John Dunn (ed.), Democracy: The Unfinished Journey, 508 BC to AD 1993. Oxford: Oxford University Press. Pages 107-124. Freeman, John R., and Duncan Snidal. 1982. "Diffusion, Development, and Democratization: Enranchisement in Western Europe." Canadian Journal of Political Science 55: 299-329. Galor, Oded. 2004. "From Stagnaton to Growth: Unified Growth Theory." In P. Aghion and S. Durlauf (eds.), Handbook of Economic Growth. Amsterdam: North-Holland. Graham, Richard. 2003. "Ciudadam'a y jerarquia en el Brasil es- 29 clavista." In Hilda Sabato (ed.), Ciudadania politica y formacion de las naciones: Perspectivas histdricas de America Latina. Mexico: El Colegio de Mexico. Pages 345-370. Hafer, Catherine. 2006. "On the Origins of Property Rights: Conflict and Production in the State of Nature." Review of Economic Studies 73: 1-25. Hall, R.E., and C.L. Jones. 1999. "Why do some countries produce so much more output per worker than others?" Quarterly Journal of Economics 114: 83-116. Halperin-Donghi, Tulio. 1985. "Economy and society in post-Independence Spanish America." In Leslie Bethell (ed.), The Cambridge History of Latin America. Volume III: From Independence to c. 1870. Cambridge: Cambridge University Press. Pages 299-346. Justman, Moshe, and Mark Gradstein. 1999. "The Industrial Revolution, Political Transition, and the Subsequent Decline in Inequality in 19th Century Britain." Explorations in Economic History 36: 109-127. Ketcham, Ralph (ed.) 1986. The Anti-Federalist Papers and the Constitutional Convention Debates. New York: Mentor Books. Kaufmann, Daniel, Aaart Kraay, and Pablo Zoido-Lobaton 1999. "Governance Matters" World Bank Research Working Paper 2196. Lindert, Peter. 2004. Growing Public. Social Spending and Economic Growth Since the Eighteenth Century. Volume I. Cambride: Cambridge University Press. Lizzeri, Alessandro, and Nicola Persico. 2004. "Why did the elites extend the suffrage? Democracy and the scope of the government, with an application to Britain's 'Age of Reform'." Quarterly Journal of Economics 118: 707-765. Llavador, Humberto, and Robert Oxoby. 2005. "Partisan Competition, Growth and the Franchise." Quarterly Journal of Economics 119: 1155-1189. Lee, Woojin. 2003. "Is democracy more expropriative than dictatorship? Tocquevillinian wisdom revisited." Journal of Development Economics 71: 155-198. Lopez-Alves, Fernando. 2000. State Formation and Democracy in Latin America, 1810-1900. Durham: Duke University Press. Lott, John R. and Lawrence W. Kenny. 1999. "Did Women's Suffrage Change the Size and Scope of Government?" Journal of Political Economy 107: 1163-1198. Macaulay, Thomas B. 1900. Complete Writings, vol. 17. Boston and New York: Houghton-Mifflin. Madison, James. 1982 [1788]. The Federalist Papers by Alexander Hamilton, James Madison and John Jay. Edited by Gary Wills. New 30 York: Bantam Books. Machiavelli, Nicolö. 1970. The Discourses. Edited by Bernard Crick. London: Penguin Books. Maddison, Angus. 2003. The World Economy: Historical Statistics, OECD Development Centre. Manin, Bernard. 1997. A Theory of Representative Government. New York: Cambridge University Press. Mariscal, Elisa, and Kenneth L. SokolofT. 2000. "Schooling, Suffrage, and the Persistence of Inequality in the Americas, 1800-1945." In Stephen Haber (ed.), Political Institutions and Economic Growth in Latin America. Stanford: Hoover Institution. Pages 159-217. Marx, Karl. 1952 [1851]. Class Struggles in France, 1848 to 1850. Moscow: Progess Publishers. Meitzer, Allan H. and Scott F. Richard. 1981. "A Rational Theory of the Size of Government." Journal of Political Economy 89: 914 927. Mokyr, Joel. 1990. The Lever of Riches: Technological Creativity and Economic Progress. New York: Oxford University Press. Murphy, Kevin, Andrei Shleifer, and Roberty Vishny. 1995. "The allocation of talent: Implications for growth." In Robert Tollison and Roger Congleton (eds.), The Economic Analysis of Rent Seeking, vol. 49 of Elgar Reference Collection: International Library of Critical Writings in Economics. Ashgate: Elgar. Pages 301-328. North, Douglass C, and Robert Paul Thomas. 1973. The Rise of the Western World: A New Economic History. Cambridge, Cambridge University Press. North, Douglass C, and Barry W. Weingast. 1989. "The Evolution of Institutions Governing Public Choice in 17th Century England. " Journal of Economic History 49: 803-832. Palmer, R.R. 1964. The Age of the Democratic Revolution: vol. II. The Struggle. Princeton: Princeton University Press. Parente, Stephen L., and Edward C. Prescott. 2000. Barriers to Riches. Cambridge, MA: MIT Press. Peltzman, Sam. 1976. "Toward a More General Theory of Regulation." Journal of Law and Economics 19: 211 248. Piketty, Thomas, Gilles Postel-Vinay, and Jean-Laurent Rosenthal. 2006. "Wealth Concentration in a Developing Economy: Paris and France, 1807-1994." American Economic Review 96: 236-256. Pomeranz, Kenneth. 2000. The Great Divergence. Princeton: Princeton University Press. Przeworski, Adam. 2006. "Granted or Conquered: A History of Suffrage Extensions." Working Paper. Department of Politics, New York University. 31 Przeworski, Adam, and Fernando Cortés. 1971. "Sistemas partidis-tas, movilización electoral y la estabilidad de sociedades capitalistas." Revista Latinoameričana de Ciencia Politica 2: 220-241. Przeworski, Adam, and Michael Wallerstein. 1988. "Structural dependence of the state on capital." American Political Science Review 82: 11-29. Rajan, Raghuram G. and Luigi Zingales. 2003. Saving Capitalism from the Capitalists. New York: Crown Books. Rodrik, Dani, Arvind Subramanian, and Francesco Trebbi. 2002. "Institutions Rule: The Primacy of Institutions Over Geography and Integration in Economic Development." Ms. Saint Paul, Gilles, and Thierry Verdier. 1996. "Inequality, redistribution and growth: A challenge to the conventional political economy approach." European Economic Review 40: 719-728. Sieyes, Emmanuel. 1979 [1789]. Qu'est-ce Que le Tiers Etat? New York: Arno Press. Stasavage, David. 2007. "Partisan politics and public debt: The importance of the 'Whig Supremacy' for Britain's financial revolution." European Review of Economic History 11: 123-153. Stigler, George. 1970. "Director's Law of Public Income Redistribution." Journal of Law and Economics 13: 1-10. Stigler, George. 1975. The Citizen and the State: Essays on Regulation. Chicago: University of Chicago Press. Sussman, Nathan and Yishay Yafeh. 2005. "Constitutions and Commitment: Evidence on the Relation between Institutions and the Cost of Capital." Hebrew University of Jerusalem, http://bschool.huji.ac.il/facultye/yafeh/NW_CRITIQT Ticchi, Davide and Andrea Vindigni. 2006. "On Wars and Political Development. The Role of International Conflicts in the Democratization of the West." Ms. Vanhanen, Tatu. 1996. The Polyarchy Dataset. Norwegian University of Science and Technology, http://www.svt.ntnu.no/iss/data/vanhanen 7 Appendix: A Model 7.1 The Economy Consider a growth model in which the agent with the median income among the enfranchised chooses the tax rate and the revenue is distributed equally among the enfranchised. Since the economy assumed here is standard, I use several results without proofs (see Benhabib and Przeworski 2006). Note only that the tax rate is assumed to be constant for any fixed s. My strategy is to solve for tm (s) and then study comparative statics with regard to s. Thus the model is not fully dynamic, in 32 that changes of suffrage are not anticipated by the agents. There are N agents indexed by i, ordered by their assets k% that represent a combination of physical and human capital. These assets produce incomes according to yt = rkt,r > 1. (2) S among the agents, those with k% or y% above some threshold, have the right to vote, so that the proportion of eligible voters in the population is s = S/N. When S agents have the right to vote, the post-redistribution income of an agent endowed with kl is yl = (l-r)rkl + rrs^rkl (3) iSJV n_l_ s n where Ps indicates whether agent i is a member of S. Since ^ X^gjv rk% = X^gjvr^ = sr^t> where kt is the average value of assets, T yl = (l-r)rkl + Fs-rkt. (4) s The utility function is U(Ct) = ^-. (5) 1 — a The value of holding k% at t = 0 is then ^(fcS)=max^^(Q)1"g~1 (6) {ct} 1-0- Now, it can be shown that this economy grows at the rate ^ = (/3r(l-r))^, (7) and that the propensity to consume out of wealth is A = l-/3-ri^(l-T)i^, (8) the same for all agents. Each agent consumes a fraction A of his net capital income (1 — r) rk\ plus the value of transfers that he receives, discounted at r (1 — r) . Note that A is the propensity to consume out of wealth because r is one plus the rate of return, net of depreciation. To bound A away from zero, it must be true that r < r, where r satisfies 1 1 — a 1 — cr f3°r^~(l — t) a =1. Making all the substitutions, iterating backwards and solving for Cq yields 33 7.2 Optimal Rate of Redistribution Now, since all incomes grow at the same rate, for any fixed s the same agent has at all times the median income among agents who have suffrage. As s increases, however, the ratio of the endowment of the voter who is median in S to the average income, denoted by A (s) = °fco , declines. Let M 6 S be the agent with the median income among the S enfranchised agents. The first order condition dV(k^f) / dr\s = 0 implies 1 _ \(tm) tm A(t£) i - tm="a - ^M(-))- (io) This expression implicitly determines the optimal tax rate, r*(s), of the median voter in S. The second order condition is satisfied as long as a fZ 0.5, which is assumed throughout. The function r*(s) is described by the following proposition. Proposition 1 As s increases from 0, the taxrater*(s) declines, reaches ■ ■ , AM(s) , a minimum at s = — dAM(s)/ds' a then increases. Proof. To study note that d dV(r,s) _ dr d dV(r,s) _ d2V(r,s) dr ds dr dr ds dr dr2 ds ' Now, tedious algebra shows that Ts9-^ = %f(T,v) ~ - sAM(s)), where f(r,a) > 0 for all values of a that satisfy the second-order condition. Hence, %[f(r,a)-d^] = j-a(l-s^(s)). By second-order condition, the term in the square bracket is positive. In turn, £a(l - sAM(s)) = -a(AM(s) + s^sl), where ^ < 0. so that %[f(T,a)-^] = -a(A-(s) + s^), which implies that % = 0 as AM(s) + s^^- | 0. In turn, AM(s) + s^^ = 0 as s = -^7^L • Hence, ^ < 0 when ' v / 1 ds < > d/\m (s)/ds ' ds s is low, ^ = 0 at s = -dAM(lyds, and -£ > 0 when s is higher. ■ 34 Remark 1 Note that nothing guarantees that the minimum occurs in 0 < s < 1. Whether it does, depends on the distribution of assets. The value of the term sAM(s) depends on the distribution of income. To determine the value of the ratio AM(s), note first that for any distribution the percentile of the median value of assets among the enfranchised is F(AM(s)) = l-±s. (11) Assume that the distribution of assets is exponential, with the cdf given by F(Al) = 1 -pexp(-Al), (12) where A* = ^. F{0) = 1 — p is then the proportion of agents without any assets, and p is the proportion with. Substituting AM in (11), equalizing (10) and (11), and solving for AM(s) yields AM(s) = log2 + logp-logs. (13) We can now prove something that may be obvious: Proposition 2 If the distribution of assets is exponential, the optimal tax rate reaches the minimum when the median income among the enfranchised equals the average income in the population. Proof. The median value of assets among the enfranchised equals the average in the population when A(s) = log 2+logp—log s = 1, or s = ~ep. Now, by Proposition 1, dr*/ds = 0 when A(s) + sd^f^ = 0, which implies under exponential distribution that log 2 + logp — log s — 1 = 0, which is the same. ■ The intuition that tax rates reach the bottom when all and only the property owners are enfranchised, however, is not supported by the model. At the point when A(s) = 1 and dr*/ds = 0, logs — logp = log 2 — 1, so that s < p. (In fact, s = |p.) The political dividing line is the average income in the population, rather than being destitute: people with no property are just slightly poorer than people with very little of it. 35 7.3 Suffrage and Welfare The value function of agent with A* when s agents are enfranchised is given by: Proposition 3 An agent with A1 is enfranchised when s = l — F(Al) = sl. Hence, when the distribution of income is exponential, an agent who is enfranchised at sl has A1 = — log sl. If i is enfranchised before the median among the enfranchised is poorer than the average agent, sl < |; or A1 > 1 — log 2, the value function Vl(r*(s)) changes according to the following pattern: dVj ds if Comments %{-Al) > 0 s < sl %U-AO + ^U >0 5 = 5* A* =-logs* £(-Al + AM{s)) - ^ < 0 sl < s < 2s% A% < AM(s) -^iL^O s = 2s* A' = AM{s) %{-Al + AM{s)) - ^ < 0? 2sl < s < \ A1 > AM{s) -^L_2<0 s = l AM(s) = l %{-Al + AM{s)) - ^ < 0 f < s < 1 A1 > AM{s) If i is enfranchised when the median among the enfranchised is poorer than the average agent, sl > - and A1 < 1 — log 2, the value function V1{t*(s)) follows: dV j ds if Comments ^(-Al)>0 s < sl 0 s = l AM(s) = l %(-Al) < 0 s < sl %U(-Al) + > 0? I < s = sl A1 = - logs1 %{-Al + AM{s)) < 0? \ < ll < s < 1 A* < AM(s) Proof. After some algebra, it can be shown that at the maximum at which r = r*(s), ^=^i(-^)+^-(.))-j]}, (") where Aq = klQ/ka. The rest follows from Proposition 2. 36 Remark 2 The term Aq) is the loss or gain, depending on the sign of from the tax bill (1 - r)Al0; fs (1 - r)Al0 = -^Aj. T/ie term multiplying I, in turn, is the loss or gain resulting from the change of the value of transfers. These transfers are -, so that -rz = -^r — -^- But - = J J J s 7 as s s as s^ s -^-^ds = (log2 - logs), so that AM(s) = - j ±ds = - log s + log2. Hence, the term ^AM(s) should be interpreted as —j^ j ~ds: it is the part of the loss(gain) of transfers due to transfers changing as a result of changing s, while the part is the part due to fixed s. Hence, for i £ S, and for i £ N — S, dV* _ rk0dr r ^ ds a ds There are some situations in which the sign of dV/ds is ambiguous: (1) As long as the agent who entered at s = sl is wealthier than the median in S, who is in turn still wealthier than the average in the population, the term ( —A* + AM(s)) is negative and since dr/ds < 0, the entire first term is positive. Yet for Aq psj Am(s), this term is close to 0. In turn, in the vicinity of AM(s) = 1, dr/ds fx 0. I cannot prove that dV/ds < 0 in the entire interval 2sl < s < 2/e. All numerical simulations (see below), however, indicate that it is. (2) A similar situation occurs with agents who are poorer than the average. If their Aq psj 1, they enter when dr/ds 0, so that dV/ds > 0. If they are very poor and enter right before sl = 1, their Aq psj 0 and again they get a boost from being enfranchised. I cannot prove again what happens in the interval 1 > Aq > 0, but numerical example indicate again that enfranchisement increases the value of these agents. Finally, the same applies to the interval | < sl < s < 1, but simulations show that the slope of dV'/ds < 0. 37