From Hobbes to Rousseau: Inequality, Development and Democratization Matteo Cervellati

Piergiuseppe Fortunato

Uwe Sunde∗

February 13, 2004

Abstract This paper analyzes the interdependent processes of economic and political development in a dynamic perspective. In contrast to the ‘state of nature’ characterizing oligarchies, democracies are based on the ‘rule of law’ and imply a more favorable environment for economic activity. The model provides a characterization of the endogenous transition from oligarchic to democratic institutions under different potential scenarios. Democratization can be initiated by a strong elite aiming for economic benefits of democracy, or can be forced by a politically disenfranchised majority under the shadow of conflict. Also revolutions can arise. The implications are discussed in light of historical evidence. JEL-classification: H10, O20, N10 Keywords: Democratization, Oligarchy, Conflict, Universal Suffrage, Revolution

∗

Matteo Cervellati: Universitat Pompeu Fabra and Universit` a di Bologna. Piergiuseppe Fortunato: EUREQua (Universit´e de Paris I) and Universit` a di Bologna. Sunde: IZA, Bonn and University of Bonn. Contact: Uwe Sunde, P.O. Box 7240, D-53072 Bonn, Germany. Phone: +49-228-3894-221, Fax: +49-228-3894-510, Email: [email protected]. Financial support from IZA is gratefully acknowledged.

1

Introduction

The economics profession has always been concerned with the analysis of the interdependencies between institutional environment and economic performance. Recent contributions have revived this field by explicitly investigating political institutions, and the effects of these institutions, such as democratic rules and constitutional details on economic outcomes. While a lot is known nowadays about these interdependencies, research on the dynamics of these interdependencies, and hence the endogeneity of institutions and economic performance in a perspective of dynamic development, is still scarce. Few systematic studies exist on why we observe the political structures that are in place in different countries, and why they differ across countries. Also, little is known about the transition from midieval oligarchic structures, where the aristocracy had all decision power, even power over life and death, to modern democracies. This transition, and the associated endogenous changes of political institutions, however, is at the center of the structures that shape today’s societies. In a series of articles and book chapters, Acemoglu and Robinson (2000, 2001, 2003, 2004), have put forward theories that argue that most of this transition was actually driven by those that had oligarchic power, but faced substantial opposition, facilitated by economic conditions or technological advances in warfare. Essentially, the threat of revolution and expropriation induced the elites to involve larger and larger groups of the population in the political process, thereby sharing the responsibilities and decreasing the opposition to their rule. On the other hand, Lizzeri and Persico (2004) argue that under certain circumstances it might have actually been in the elite’s interest to enlarge the suffrage, which would allow, for example, to increase efficiency by facilitating the provision of public goods, and setting limits to the rent seeking behavior on the side of the elite. They argue that this mechanism might have played a larger role in England’s history of development than the threat of revolutions. However, both lines of argument are inherently static and mutually exclusive. Moreover, it is not clear what the extension of the franchise actually means. But in fact, there is evidence that both mechanisms have played a role for the transition to democracy in different countries and different contexts. The goal of this paper is to provide a simple theoretical framework to study the endogenous transition to democracy as the result of economic and political forces with particular attention to the conditions under which the different forces toward democratization should be expected to prevail. In contrast to the aforementioned studies, this paper adopts a dynamic perspective of development, where the environment under which decisions are made evolves endogenously. We view economic and political development as dynamic processes that are closely intertwined, leading to bidirectional feedback effects. This approach corresponds very much in the spirit to re1

cent theories to the transition from economic stagnancy to long-run growth and its determinants, like Galor and Weil (2000) and Hansen and Prescott (2002). The major contribution of this paper is a full dynamic characterization of economic development and the political transition, taking interdependencies and dynamic feedback effects between economic environment and implemented institutions explicitly into account. This way, our theory is able to nest previous approaches, such as the democratization from above as suggested by Lizzeri and Persico (2004), and the forced democratization as described by Acemoglu and Robinson (2001), within a single framework. The model bases on a stylized representation of oligarchies and democracies by focusing attention to some key features.1 Oligarchies are characterized by a ‘state of nature’ where political power of the ruling elite is is based on the threat of violent repression and is virtually unrestricted (even including decisions over life and death). There is no legal frame that can solve conflicts of interests in a standardized way, and there is no enforcement of property rights or law and order by an executive branch. Hence, oligarchies reflect a Hobbesian view of society. Democracies, in contrast, are characterized by the ‘rule of law’ regulating both the relationship between the government and citizens and among citizens. The rule of law is perfectly enforced by the power of a superstructure, the state, and its executive and judiciary branches, and includes property rights protection. All citizens have political representation (universal suffrage). Hence, democracy reflects the social contract as advocated by Rousseau. These different features have important politico-economic implications. In democracies, the conflicts of interests are resolved by aggregating political preferences without the need to recur to violent conflict. Government actions mirror the preferences of (the majority of) the citizens and tend to implement reforms in the majority’s favor, including redistribution.2 Democracies also provide a better environment for economic activity for (at least) two reasons: the protection of property rights represents a strong incentive to economic activity, avoiding hold-up problems; and the government can provide public goods and services more efficiently by facilitating their financing through general taxation. We consider the following simple framework to reflect these thoughts. Under oligarchy and the state of nature, political power is allocated depending on the relative conflict power, represented by a simple guns model. The absence of property rights induces individuals to distract resources from productive activities to devote them to defensive or predatory activitivies.3 1

For the purpose of this paper, the terms oligarchy and dictatorship could be used interchangably. 2 Even with redistribution, the rule of law and the protection of property rights rule out full expropriation of the minority. 3 We use the metaphor of an arming game to model the role of property rights protection as incentive to economic activity.

2

Under democracy, political decisions, e.g. on the redistribution of incomes, are made according to the outcome of majority voting. Since property rights are enforced, no resources need to be spent for defensive or predatory uses. The choice of the political envirnment (democracy vs state of nature) results endogenously from the conflict game between the different groups. Initially, a rich elite controls the natural resources. In this initial situation the politico-economic equilibrium is characterized by the elite keeping the power under the threat of conflict with the production possibilities of the community not being efficiently utilized. Over the course of subsequent generations, technical change and investment in physical capital extends the production possibilities. This has implications both for the inequality of incomes and for importance of having control over natural resources (for both productive and conflict purposes). The analysis of the economic and political outcomes over the course of generations allows to characterize a politico-economic development path. In particular, we are able to characterize the conditions under which the elite finds it profitable to offer democracy or becomes unable to resist the pressure of keeping power in a state of nature. A third possible scenario is a revolution or coup leading to a change in the political landscape, potentially a new oligachic structure. The model implies that whatever the development path looks like, it eventually leads to democratization. The empirical relevance of the model implications can be assessed, since the model makes testable predictions concerning the required environment for a particular transition path. Our results are in line with recent empirical findings of causal effects of economic development on the probability of non-democracies to democratize, see Boix and Stokes (2003), as well as the stabilizing role of ecnomic development on democratic structures, and vice versa, of democracies providing development enhancing conditions such as infrastructure, see Boix (2001). The following section introduces the economic environment we study. Section 3 presents the different dimensions of political conflict individuals face, as well as the political game played. Section 4 analyses the interactions between economic and political environment, and presents a characterization of the process of development. Finally, Section 5 discusses the results in a historical perspective, and concludes.

2

The Economy

This section introduces the basic structure of the economy, while the following section describes the struggle for influence and income within that economy.

3

2.1

Individuals

Consider an economy, which is populated by an infinite series of subsequent generations of individuals. A given generation t consists of a continuum of individuals i of measure Lt . Each individual has a single parent and a single offspring, so the size of the population is constant over generations, Lt = L. In the following, we use i to interchangeably denote an individual and the family or dynasty to which the individual belongs. Individuals engage in producing a unique commodity, which can be consumed or bequeathed. We abstract from labor-leisure choices and assume that every individual inelastically supplies one unit of labor during his life. Aggregate labor input in production therefore equals the total population size L. For simplicity, individuals’ utility is logarithmic in consumption c and bequests b:4 uit = u(cit , bit ) = (1 − β) log cit + β log bit .

(1)

Consequentially, each individual devotes a constant fraction of his individual income yti to consumption and bequest, respectively, with cit = (1 − β)yti and bit = βyti . As is shown below, production involves physical capital K and natural resources N , which we call land. There is no possibility for individuals to invest resources in capital formation. Rather, capital can only be created through investing bequests. On the other hand, bequests can only be invested in capital creation, otherwise they are wasted. This implies that the capital stock available to an individual equals the individual parent’s bequests, kti = bit−1 , and likewise the aggregate capital stock in the economy
i equals the invested bequests of the previous generation, Kt = Bt−1 = bt−1 di. The capital stock of a generation fully depreciates when the generation dies. In contrast, land is ready to use for its owners, and does not depreciate, hence Nt = N . However, only a fraction γ < 1/2 of all families owns land, and land is equally distributed among land-owners. Land is passed-on from generation to generation, and there is no market for land.5 In the initial generation, capital is distributed equally among all members of the economy. For notational clarity, we denote per capita variables by lower case letters, and aggregate variables by upper case letters, i.e. yt = Yt /L, kt = Kt /L, and n = N/L. 4

This formulation of the utility function is not crucial for the main insights, even though, as will become clear below, the development dynamics of the economy depend on the distribution of factor endowments, and hence the decision on consumption and bequest. An OLG environment would be equally appropriate, but, apart from changing the dynamics, taking into account issues such as retirement and savings would complicate the analysis while leaving the essential insights of the model unchanged. Moreover, we think that, in the current context, a certain degree of myopic behavior is more realistic than an infinite horizon model. 5 This assumption is without loss of generality. In fact, as will become clear below, even allowing for land markets does not change the results, because selling or buying land using e.g. bequests is always a (weakly) dominated strategy.

4

2.2

Production

The unique final commodity in the economy, which is used both for consumption and investment in capital through bequests, is produced in two sectors, a traditional, land-intensive sector, and a modern, physical-capital-intensive sector. The former can be interpreted as agriculture, while the latter can be thought of as manufacturing. Aggregate output of the economy is then given by (2) Yt = YtT + YtM . Production in both sectors involves a Cobb-Douglas technology with constant returns to scale. Both sectors employ homogenous labor supplied by individuals. Let LTt = (1 − θt )Lt denote the amount of labor employed in the traditional sector, and let LM t = θt Lt denote the aggregate labor input in the modern sector, where θt represents the respective fraction of the population employed in the modern sector. For simplicity, the traditional sector uses only land and no physical capital, while the modern sector uses only physical capital, but no land as other input besides labor. With Dt and At as the level of (labor-augmenting) productivity in the traditional and modern sector, respectively, the sectoral production functions can be written as (3) YtT = F (N, LTt ) = N α (Dt LTt )1−α , and

α M 1−α . YtM = F (Kt , LM t ) = Kt (At Lt )

(4)

Technological progress favors the modern relative to the traditional production sector, in the sense that labor-augmenting technological change is faster in the former than in the latter. To keep the model simple, we assume that technological progress is exogenous and only affects the modern sector, hence we simplify the expressions by normalizing Dt = D = 1 ∀t, and assuming 6 At − At−1 = at = a ∀ t . A˙ t = At−1

2.3

(5)

Factor Prices

Factor prices are determined competitively on the aggregate level and equal marginal productivities.7 Consequentially, the equilibrium wages wtT , and wtM , read wtT and

wtM

 −α = (1 − α)Ntα LTt ,  M −α α 1−α = (1 − α)Kt At , Lt

6

(6) (7)

Endogenizing the rate of technical progress would not affect the main argument. Evidence supports this assumption that different sectors were competing for factors and that factor prices reflected productivities, even before or at early stages of the industrial revolution, see Magnac and Postel-Vinay (1997). 7

5

respectively. Labor is homogeneous and moves freely between the two sectors, such that in equilibrium wtT = wtM = wt .

(8)

Hence, substituting for the expressions for the respective labor inputs from equations (3) and (4), one can derive the equilibrium allocation of labor in the modern sector, 1−α

θt∗

At α kt

=

1−α

.

(9)

At α kt + n

For notational convenience, denote the effective capital stock available per member of generation t in the economy as k˜t , with 1−α K t k˜t = At α . Lt

(10)

Using this short-cut and substituting the equilibrium labor allocation back into the expression for wages, one obtains  α (11) wt = (1 − α) k˜t + n as expression for the equilibrium wage. Similarly, the equilibrium rental rates for physical capital, rt and land ρt , are 1−α

rt = and

ρt =

1−α 1 α At α , 1 − α k˜t + n 1 α , = wt 1 − α k˜t + n

αAt α 1−α ˜ (kt +n) α

(k˜t +n)

= wt

1−α

(12) (13)

respectively. From this it follows that the ratio of rents only depends on the level of technological development,

2.4

rt ρt

1−α

= At α .

Individual Incomes

As mentioned before, only a fraction γ of the population owns land. In the following, these individuals are called the land-lord elite and denoted by E. Individuals (or dynasties) owning no land are referred to as landless people, or proletariat, denoted by P . Individuals derive their income, which they can then either consume or bequeath (and hence invest in the next generation’s capital stock), from selling their endowments in terms of labor, invested bequests, and, in case they are landlords, land, hence yti = wt + rt kti + ρt nit 6

with i ∈ {E, P } .

(14)

Since incomes of landlords and landless differ, and since all individuals of a generation bequeath a constant share of income, bequests in terms of capital stocks differ as well between landlords and landless. For illustrative purposes, consider the following index of how much capital an individual owns in comparison to the average per capita capital stock in the economy, λit :=

kti . kt

(15)

Using the previous results for the equilibrium factor prices, individual income can then be expressed as    α ˜t α k α i i i (1 − α) + (16) yt = k˜t + n λt + n . k˜t + n k˜t + n The respective expressions for landlords and landless differ only with regard to rents accrueing from land ownership, the last term in brackets. A landlord i ∈ E therefore owns ni = nE = nγ land, while a landless individual i ∈ P owns ni = nP = 0. A landless individual’s imcome therefore lacks the last term in brackets. This completes the description of the technological environment.

3

Conflict and the Struggle for Power

Apart from the technological environment described before, the economy is characterized by an institutional-political environment, which affects all individuals’ decisions. This section describes the political environment, as well as the individuals’ decision problem implied by this environment.

3.1

The State of Nature and its Implications

For simplicity, there are two possible institutional regimes. Under oligarchy, which reflects the initial state of the world, property rights concerning endowments are not ensured by any sort of executive authority, and individuals have no possibility to commit not to try to appropriate other individuals’ endowments. Moreover, political rights are not equally distributed. In particular, as was the case e.g. in medieval Europe, only landowners have decision rights for the community as a whole, for example on taxation or warfare, that is, in other words, political power. However, under oligarchy there is no way for the elite to commit to pay taxes or redistribute, so there is effectively no state with its typical duties and responsibilities, such as public good provision and redistribution. Rather, the environment implies a individualistic race for maximizing well-being, possibly on the others’ expenses. In contrast, under democracy, property rights are perfectly protected by the

7

state by means of efficient executive branch and judiciary. Moreover, democracy is associated with universal and equitable suffrage. Under democracy, collective decisions, in particular on taxation and redistribution, are made by the electorate, that is, essentially, the median voter. In this model, since agents are only heterogeneous with respect to whether or not they own land, the median voter is landless as γ < 1/2. Assume that the only decision to be made under democracy is about the size of the proportional tax rate on income. All tax revenues are then evenly and equally distributed as a lump-sum transfer to all individuals. Without specifying the voting game, we take it as given that democracy then implies full redistribution, so that, after redistribution, all members of the economy, regardless of what group they belong to, earn the same income. This extreme implication helps to keep things simple and drastically illustrates the difference to the oligarchic regime. The particular living conditions under the possible regimes imply potential for conflict on the individual as well as on the aggregate level. We discuss both in turn.

3.2

Protection in a Predatory World: The Arming-Game

In the model, as in the real world, individuals can expropriate others of their endowments if they are ‘stronger’, that is if they are armed, when there is no state protecting property by means of efficient police and judiciary. Hence, individuals might want to engage in expropriation instead of production activities, and consequently invest resources in arms.8 On the other hand, individuals may want to invest resources in arms in order to protect their property, that is their endowments, or, equivalently, their income (opportunities). In terms of costs, we do not distinguish between offensive or defensive arming. Resources spent on arms or protection are wasted in terms productive use. For example, individuals may build safer houses with lockable doors, and thick walls; or they may arm themselves spending resources on weapons like swords, instead of production tools, like plows. Alternatively, one could imagine that protection from privation or expropriation requires a degree of alertness that leads to lower productivity, just as does the alertness required to spot an appropriate victim for expropriation. In terms of the model, we assume that, in order to arm himself effectively for either offensive or defensive purpose, an individual foregoes a fraction ψ ∈ (0, 1) of his inheritance or capital.9 Note that effective arming is possible for landowners as well as landless. Whether an individual owns 8

For an overview over conflict models see e.g Hirshleifer (2001). The assumption of a unique level of resource spending on arms is without loss of generality, since spending any lower fraction on arming than needed to effectively protect oneself, or, equivalently, than maximally possible, is a strictly dominated strategy, as is spending more resources on arms than necessary. Hence, ψ can be interpreted as the individually optimal level of armament in case arming is the desired choice in the first 9

8

land does not affect his chances of expropriating or being expropriated, the only relevant criterion is whether he is armed or not. Moreover, the impossibility of commitment to property rights rules out any sort of group formation, making the arms game essentially an individualistic problem of ‘everyone against everyone’. Figure 1: The Arming-Game under Oligarchy Arms

No Arms

  yti kti (1 − ψ), nit ;

  yti kti (1 − ψ), nit + ytj (ktj , njt );

j i Arms ytj No Arms ytj

  j j kt (1 − ψ), nt 0;  i i kt , nt +  j kt (1 − ψ), njt

i yt

yti

0  i i kt , nt ;

  ytj ktj , njt

Note: i, j ∈ {E, P }.

The strategic form of the arming game played in the predatory world of oligarchy, that is without property rights enforcement, is depicted in Figure 1 with the respective payoffs for two arbitrary individuals i and j, who are landowners or landless, i, j ∈ {E, P }. The unique Nash equilibrium of the game is ‘Arm/Arm’. In the case of two players from the same respective group, not being armed is a strictly dominated strategy for every player. In case j does not arm himself, i has an incentive to deviate from not arming himself since he can then profitably expropriate j. In case j is armed, the only way for i not to be fully expropriated by j is to arm. Hence, arming is the best response of i to any action of player j. In the case of one player from the elite and one from the proletariat, e.g. i ∈ E and j ∈ P , the richer player i might actually have a trade-off between arming and not-arming himself since the potential gain from expropriating a landless might be less than the costs for arming. However, for the poorer player j ‘No Arms’ is a strictly dominated strategy, in turn ruling out ‘No Arms’ as a possible equilibrium strategy for i as well, leaving ‘Arm/Arm’ as the only Nash-Equilibrium of the game. A historical remark is in order. During the middle ages, the ruling elite did indeed discretionary grant and enforce property rights for certain groups. However, the mere possibility of discretionary actions on the side of the elite, and the fact that not all citizens were treated equally, place. Moreover, assuming that ψ is a fraction is inconsequential for the results. A fixed absolute cost of arming, which is unrelated to income, would lead to the same results, but would require making an additional assumption about the size of this cost in terms of affordability of individual protection for all members of the economy.

9

implied a conflictual state of the world incentivating individuals to at least take some precautionary measures. Moreover, the elites established their political power mainly by mere military strength. The arming-equilibrium under oligarchy seems therefore warranted in historical terms, as political power and property rights enforcement were closely intertwined indeed. The payoffs in the arming game are somewhat different under democracy, that is when property rights are perfectly enforced, as is displayed in Figure 2. In this state of the world, ‘No Arms/No Arms’ is the unique Nash equilibrium: Given j does not arm, property rights enforcement prevents i from expropriating j, so arming means merely wasting resources and i does not arm himself, too. Arming is no profitable deviation.10 Likewise, if j were to arm himself, i would still do better by not arming, since his property is enforced costlessly, while arming would mean wasting resources. Figure 2: The Arming-Game under Democracy Arms

No Arms

  yti kti (1 − ψ), nit ;

  yti kti (1 − ψ), nit ;

  ytj ktj (1 − ψ), njt   yti kti , nit ;

ytj (ktj , njt )   yti kti , nit ;

  ytj ktj (1 − ψ), njt

  ytj ktj , njt

j i Arms

No Arms

Note: i, j ∈ {E, P }.

Note that the equilibrium pay-offs are higher under democracy than under oligarchy. The difference reflects the efficiency gain by adopting democracy, with property rights protection and universal suffrage, compared to oligarchy. Clearly, part of the efficiency gain is necessary to finance effective executive and judiciary authorities, but as long as the taxes necessary to finance the state authorities do not exceed the waste implied by arming, equilibrium income is higher under democracy.11 10 Note that this would be even more so, if arming were forbidden, with the possession of arms implying the incurrance of a costly penalty, or if the attempt of expropriation using arms (robbery) were sure to lead to punishment with a penalty in terms of foregone income. 11 For simplicity, we refrain from studying the costs of police and judiciary, and the taxation required to finance these costs. We instead assume that taxes are levied in a non-distortionary way and that the individual tax liability caused by the existence of enforcement authorities is strictly lower than the income waste associated with arming.

10

3.3

The Struggle for Decision Power: Political Conflict

While the level of property rights enforcement mostly affects conflict among single individuals, the distribution of political power fuels conflict among groups with different political influence. In the current model, all individuals are equal, except for the possession of land: a group of individuals owns land, while others do not. Moreover, as was mostly the case in all countries most of the times, land ownership was the prerequesite for political influence. For example, in medieval England the noblety, which exclusively was represented in parliament, mainly distinguished itself from the ordinary people by the fact that they owned land. Likewise, in Germany suffrage during medieval times was limited to land-owners. Clearly, such an unequal distribution of political power, combined with the possibility to ‘legally’ expropriate those without political influence through taxation, compulsory military service for ‘ordinary men’ etc., facilitates the formation of groups or classes. To illustrate the argument, we explicitly model the potential conflict for influence in terms of political rights (suffrage) between the elite and the proletariat. Consider the following ‘guns model’:12 In an oligarchic world, during any generation, both groups, elite and proletariat, have the possibility to go to an open conflict to change the political system to their own favor. The winner of this open conflict is determined by which group has more fighting power, which can be interpreted terms of which group has more ‘guns’ and ‘gunners’. Conflict power in terms of guns is derived from resources available to produce them, physical capital and also land. In terms of the model,  the E fighting power of the elite of generation t is then given by γ N + Kt , while generation t’s proletariat is able to mobilize (1 − γ)KtP of fighting power, since the proletariat has no land to its disposal. Note that these conflict powers merely reflect the fighting potential of the groups. Unlike in the arms-game, no investments are required for credibly and effectively realizing fighting power. Rather, one could think of the fighting power as potential that can instantaneously and costlessly be mobilized for the conflict (or civil war), implying full reversibility of production inputs (land, capital, people) into fighting potential. Consequentially, the conflict outcome depends on the sign of the condition γ

kP nE + ktE ≥ (1 − γ) t . < kt kt

(17)

The elite of generation t can enforce a political structure to its taste if it has larger military power, that is, if the left hand side is larger than (or equal to) the right hand side, while the proletariate can do the same if the inequality holds in the reverse direction.13 12 13

Acemoglu and Robinson (2004) use a similar model. By now, it should be clear that even if we would allow for markets for land, they would

11

3.4

Timing of Events and Decision Problem

In the economy with the economic environment described previously, and the conflict potential just laid down, each generation faces the following order of events and choices. When the generation is born, all individuals with political decision power for the economy as a whole can decide whether to keep the status quo as inherited from their parents, or whether to change the political structure of the economy. Since there are only two possible regimes, this implies that the ruling elite can propose to switch from oligarchy to democracy.14 Given this decision, the politically excluded can decide whether to accept the decision and stay calm, or whether to go to open conflict in order to enforce a political system of their taste.15 Given these decisions, open conflict, in case it is inevitable, takes place, and the resulting political environment for the generation is implemented. In this environment, individuals then make their decisions whether to arm or not according to the arming-game described before. Finally, given the political environment and the arms investments, production occurs, income is realized, individuals make their consumption and bequest choices, and then die, making way for the next generation. In order to illustrate the decision problem for individuals, consider the initial situation of oligarchy of the land-owning elite. Figure 3 displays the extensive form of the game played between those with political decision power, the land-owning elite E, and the politically deprived, the landless proletariat P . Upon birth, the elite can choose to stay put in oligarchy, or, alternatively, to offer the people democracy. Note that offering democracy means giving up political influence, since the median voter is member of the landless as γ < 1/2. Given the political proposal by the elite, the people can challenge the elite (and its decision) by triggering open conflict. Otherwise, they agree with the elite’s decision and do not go to conflict. If they decide for the latter, the elite’s proposed system is implemented, and all individuals not play a role. Under oligarchy of the elite, the missing enforcement of property rights would prevent any land-owner from selling land, since the buyer cannot commit to actually pay the price, and since, even if the price is paid, the proceeds could be expropriated in the state of nature. Moreover, the very fact of owning land grants political power, the loss of which in case of selling land prevents the elite from selling. Under democracy, on the other hand, there will be full redistribution of income, since the median voter is landless, so that there is no incentive to buy or sell land whatsoever, because individual landownership does not determine individual income in that case. 14 Clearly, the game structure remains the same under democracy, where all individuals decide upon status quo or going back to an oligarchy. The oligarchy would then favor the median voter, or, respecively, the particular group he belongs to, i.e. the landless. 15 Under democracy, there are no politically excluded who could challenge the majority’s decision, so the possibility to conflict is available to an empty set of individuals. Alternatively, one could also assume that the politically inferior group, facing the threat of becoming the political loser under an oligarchic regime, has the option to disagree with the majority’s decision and by doing that to trigger conflict. However, as will become clear later, this assumption would not change the results.

12

play the arming-game. Thereafter, all individuals allocate their remaining resources to production, earn their income, make their consumption-bequest choice, and die. If people decide for conflict, the guns speak, and the winner group imposes its most preferred political system: oligarchy or democracy. Given the respective outcome, all individuals play the arming-game, produce, consume and bequeath, and eventually die. The respective payoffs for the individuals as a function of the moves made are implied by the economic environment described before, as well as the outcome of the arming-game under the respective political environment chosen, as illustrated above. Also note, that all agents have perfect information about the sequence of events, about their choice sets, and about the implied payoffs. This completes the description of the model, and in particular of the game played between the two interest groups in the society, elite and people. We now turn to the characterization of the equilibrium choices of a given generation, and the associated economic outcomes.

4

Economic Development and the Process of Democratization

This section starts with a presentation of the possible subgame perfect equilibria of the game played among members of a given generation, e.g. as displayed in Figure 3. We then analyze the dynamic evolution of the economy, which illustrates the interrelation between economic and political development. The section closes by an interpretation of the findings in a historical perspective.

4.1

Taxonomy of Intragenerational Equilibria

The game played among the members of any given generation described in the previous section has a unique subgame perfect equilibrium, as we now show. However, the equilibrium strategies, and hence the associated allocative outcomes, can differ among generations, depending on the respective (parametric) environment. We start by stating the four possible types of intragenerational equilibria and the conditions under which they arise, and postpone the description of how and when these equilibria arise endogenously, to the next subsection. Status Quo. The status quo equilibrium is characterized by the (respectively ruling) elite proposing no change of the political system, persisting in the status quo, and by the proletariat approving the status quo by not going to open conflict.

13

Two conditions are necessary and sufficient for this equilibrium:   ytE ktE (1 − ψ), nE ≥ yt (kt ) ; γ

nE

+ kt

ktE

> (1 − γ)

ktP kt

.

(18) (19)

Condition (18) implies that the status quo of having exclusive political power without ensuring property rights results in a higher income for individuals belonging to the elite than a democracy, which entails property rights protection and thus higher efficiency, but also full redistribution in the sense of full equalization of incomes. The respective individual income under democracy therefore equals the average per capita income in the economy, which, by inserting the average per capita endowments into equation (16), can be simplified to16  α wt = k˜t + n yt (kt ) = . (20) 1−α Moreover, as indicated by condition (19), the elite is sufficiently strong to prevail in an open conflict. Consider again the extensive form game implied by Figure 3. Note first that, by (19), any open conflict entails the political system favored by the elite, which, by condition (18), is the status quo of oligarchy. Hence, the dominant strategy for the elite is offering the status quo, while the proletariat’s best response will always be to agree whatever the offer.17 Moreover, there is no profitable deviation for any group. Hence, the only subgame perfect equilibrium of the game under the given environment is the status quo equilibrium. Democratization From Above. Under certain circumstances, however, the elite might voluntarily choose to give up their political predominance and propose democratization, even though the elite would be able to defend its position in case the political struggle would escalate to open conflict. Such a situation, which is at the heart of the model proposed by Lizzeri and Persico (2004) to describe the democratization process in England during the ‘Age of Reforms’, is characterized by the following conditions:   (21) ytE ktE (1 − ψ), nE < yt (kt ) ; γ 16

nE + ktE kt

> (1 − γ)

ktP . kt

(22)

As indicated before, we refrain from analyzing the political process determining the tax rate. The processes of taxation and redistribution do not involve any inefficiencies or loss of aggregate income. 17 Strictly speaking, the proletariat is indifferent between going to conflict or not, since the outcome for both choices, ‘agree’ or ‘disagree’, is the same. Introducing some costs for conflict removes this indifference and gives strict dominance for the ‘agree’-strategy.

14

The unique subgame perfect equilibrium in this situation is an offer to democratize by the elite, which is accepted by the proletariat. To see this, note that, as before, any conflict would see the elite as winner, and thus their most preferred political system. However, condition (21) indicates, that members of the elite enjoy strictly higher incomes under democracy than under oligarchy due to the efficiency gains induced by property rights enforcement, despite the taxation and redistribution that democracy implies. Moreover, since the proletariat (including the median voter) strictly gain from appropriating some of the income resulting from land rents in the production process, the proletariat agrees to the offer to democratize. There is no profitable deviation, since disagreement and conflict would lead to the same outcomes in terms of political system choice, arming-game outcome and implemented allocation. Hence, conditions (21) and (22) imply ‘democratize’ and ‘agree’ as the respective equilibrium strategies for elite and proletariat. Democratization From Below. Of course, democratization from above is not the only transition one could think of. Rather, there might be situations in which the elite would actually prefer to keep exclusive political power in order to secure the benefits from not having to redistribute part of their income. But, at the same time, the elite realizes that it would lose power if it came to open conflict, and therefore anticipates any (potentially costly and anyway useless) conflict by proposing democratization. This is essentially the scenario of democratization under the threat of revolution described by Acemoglu and Robinson (2000, 2001, 2004), and requires that the following conditions hold:   (23) ytE ktE (1 − ψ), nE ≥ yt (kt ) ;   (24) yt (kt ) ≥ ytP ktP (1 − ψ), nP ; γ

nE + ktE kt

< (1 − γ)

ktP . kt

(25)

Condition (23) implies that the elite prefers the status quo in terms of income. But, as indicated by condition (25), the proletariat’s fighting power suffices to win an open conflict and implement a new system. However, unlike in the two previously described situations, in this situation the proletariat needs to know which system to implement. For democratization to be a preferred choice, the proletariat must be willing to implement democratization as opposed, for example, to an oligarchy after a land-reform that redistributes all the land uniformly from the elite to the proletariat, with nP = n/(1 − γ). The consequences of the latter are displayed at the right hand side of condition (24).18 18

A harsher reform would also include redistribution of all capital from elite to prole-

15

These three conditions are sufficient for a unique subgame perfect equilibrium characterized by the elite’s strategy to ‘democratize’, and the people’s strategy to ‘agree’ to democratization. To see this, note that any conflict will lead to the proletariat imposing their favored regime, which is democracy as opposed to full redistribution of resources combined with lack of property rights protection. Hence, proposing the status quo is a dominated strategy for the elite.19 The proletariat’s equilibrium contingency plan is to ‘agree’ if democratization is proposed by the elite, but to ‘disagree’ if status quo is proposed.20 The equilibrium therefore implies democratization which is essentially enforced by the proletariat’s power, i.e. from below, but without open conflict.21 There are no profitable deviations for anyone. Revolution. However, not always will the political transition from the initial state of a landlord elite that makes all political decisions be a peaceful one. Rather, given that the proletariat is strong enough, they will want to oust the elite and reform the political system in their most preferred way. The outcome is ‘revolution’, which requires the following conditions to be satisfied:   (26) yt (kt ) < ytP ktP (1 − ψ), nP ; γ

nE + ktE kt

< (1 − γ)

ktP . kt

(27)

The revolution equilibrium is characterized by stronger fighting power of the proletariat, together with a particular income constellation, which makes the proletariat go to conflict regardless of whether the elite offers democracy or not. As already indicated, one could think of several alternative scenarios implemented by the (old) proletariat following an escalation of the political conflict, i.e. a revolution and determining the proletariat’s income after a revolution. The right hand side of inequality (26) presents one potential outcome that the proletariat strictly prefers to democracy. This alternative condition represents the state of the world after a proletarian revolution, tariat. Since this would clearly imply an even higher per capita income for proletarians, this case is included if the proletariat chooses to go for an oligarchy of the masses. However, we abstain from that possibility to keep the model symmetric. 19 To be precise, the status quo is a weakly dominated strategy in the current context. But assuming that conflict is costly, or that proposing democracy allows the elite to influence the democratic design in their favor in contrast to the case of a lost conflict, suffices to make it a strictly dominated strategy. 20 Of all the cases, this is the only equilibrium for which timing, i.e. the extensive form of the game, is relevant, since it avoids a coordination problem that does not arise in the other cases. 21 Note that the allocative outcome is the same regardless whether the elite proposes status quo or democratization, that is, essentially, regardless whether it comes to open conflict or not. Again, assuming that conflict is avoided by both parties where possible, open conflict is not an equilibrium strategy for any party.

16

which is characterized by oligarchy of the masses, implying redistribution of all land to the proletariat, i.e. nP = n/(1 − γ), no taxation, and no universal property rights protection. In this scenario, it is a strictly dominant strategy for the proletariat to ‘disagree’ to any offer the elite may make, with the consequence of open conflict between elite and proletariat, which the former lose due to condition (27). In this particular case, it is of no consequence what the elite proposes on the first stage. Assume for simplicity, that the elite refrains from an inconsequential proposition, implying a proposal to leave the status quo unquestioned. Hence, the unique subgame perfect equilibrium is ‘status quo’ followed by ‘disagreement’ by the proletariat and conflict. Two aspects of this equilibrium are noteworthy. First, the environment is such that the elite cannot possibly evade disempowerment and expropriation of their property. Democratization is not an acceptable offer to the people, so all the elite can do is wait for the end. Second, one might think of other alternative political regimes the people are inclined to impose by force rather than the ‘revolution’ alluded to before. One possibility could imply a full expropriation of the elite also concerning  their capital, leading to an income kt n P (1 − ψ), nP , where nP = (1−γ) . of a member of the ‘new elite’ of yt (1−γ) The (defeated) members of the old elite only keep the revenues from their labor supply. Another, even harsher alternative regime implemented by the proletariat is an equal society after eradication of the elites. Under this scenario, the proletariat not only redistributes all resources to themselves, but leaves no income whatsoever to the former elite. Essentially, the old elite of the proletariat with an income of   is killed, leaving the members kt P yt (1−γ) (1 − ψ), N, L(1 − γ) . Note that total production is lower in the last case since the labor input of the elite is wasted by killing them. To keep the symmetry of the model intact, however, we go on assuming that the most the proletariat can go for is to establish a new oligarchy in their favor. Summing up, we have shown the existence of a unique subgame perfect equilibrium of the game for the political structure of the economy. The equilibrium depends crucially on the particular economic environment. In particular, we have shown that there are four different types of equilibria that can arise under four mutually exclusive parametric (economic) environments. The type of equilibrium, however, in turn implies the economic allocation, and therefore production, incomes, and development.

4.2

A Characterization of Development Dynamics

We are now in a position to derive the central result of the model, a characterization of the dynamics of economic and political development. The key state variables of the model are the level of development of the economy,

17

represented by effective capital k˜t , and the relative inequality in terms of λE kE wealth in physical capital, λt = λtP = ktP . The key parameters are the init t tial inequality in terms of land resources and thus political power, expressed by γ, the rate of technological progress, a, and the size (and therefore the importance, of natural resources or land, n. Since we are interested in a characterization of development from a dynamic perspective, and since the dynamics are measured in terms of generations rather than time, all arguments will be made in terms of generations that face some sense crucial environmental conditions. We will refer to these generations as threshold generations. Consider the initial conditions of the dynamic process of development at the beginning of time. The endowments of the members of the economy are as follows: there is no physical capital in the economy, so k0E = k0P = 0. Land is owned exclusively by members of the landlord elite, so nE 0 = n/γ while nP0 = 0. Moreover, all the political power is restricted to the ownership of land and thus in the hands of the landowners. As becomes clear from the structure of individual incomes as displayed in equation (16), already in the initial period landowners have higher incomes than landless people, simply by the fact that they own the same labor and capital endowments, but addionally land. This implies that the subsequent generation of landowners inherits more capital from their ancestors than the same generation’s landless members, i.e. k1E > k1P . Consequentially, landlords are initially relatively more capitalist than the landless, that is λP1 < 1 < λP1 .22 However, as shown in the following lemma, the degree of capitalism in both groups converges over time. Lemma 1. The individual capital stocks owned by a member of a landless family and a landlord family, respectively, converge over the course of generations. Compared to the average family, the relative capital stock owned by a landless family increases over the course of generations, while the opposite eventually holds for families that own land. Proof. First note that average income is given by yt = kt = βyt and kti = βyti , one can write

wt (1−α) .

i yt−1 wt−1 + λit−1 kt−1 rt−1 + ni ρt−1 = yt−1 wt−1 + kt−1 rt−1 + nρt−1 αk˜t−1 i αni + λt−1 , = (1 − α) + k˜t−1 + n k˜t−1 + n

Then, since

λit =

(28)

using conditions (11), (12), (13) and (16). The capital intensity index of a dynasty i therefore converges over the course of generations to a steady 22

Note that this is true also when accounting for the outcome of the arming-game, which implies wasting a share ψ of all individuals’ capital resources.

18

state value λi∗ , which can be computed to be   (1 − α) k˜ + n + αni ˜ = , λi∗ (k) ˜ − α) + n k(1

(29)

˜ Due to unbounded which in turn depends on the steady state value of k. technical progress, however, incomes and capital stocks rise over generations, implying that also effective capital grows unboundedly, limt→∞ k˜t = ∞, while land is a fixed factor. Consequently, using de l’Hˆ opital’s rule, = 1 ∀i ∈ {E, P }, which proves convergence. Morelimt→∞ λi∗ = (1−α) (1−α) ˜

λP , with over, note that condition (28) implies that λPt = (1 − α) + k˜αkt−1 +n t−1 ∂λP t ˜t−1 ∂k

t−1

=

αn P 2λ (k˜t−1 +n) t−1

> 0, which implies an increasing capital stock owned

by landless compared to average individuals as generations pass and k˜ inP creases. By the same token, however, since γλE t + (1 − γ)λt = 1 ∀t, this E ∂λ implies that ∂ k˜ t < 0, implying the last claim. This concludes the proof of t−1 the dynamics of relative capital ownership. Some comparative statics results concerning the capital ownership indices are also useful for later reference. Lemma 2. Everything else equal, the ratio of the individual capital endowment of a landlord relative to that of the average individual, λE t is (i) decreasing in the rate of technological progress, a, (ii) decreasing in the size of the elite, γ, and (iii) increasing in the stock (and hence the importance) of natural resources, i.e. land, n. Proof. Consider λE t as given by condition (28). Then 0, and (ii) follows from ∂λE t ∂n

=

1/γ−λE t−1 ˜ 2 αkt−1 (k˜t−1 +n)

∂λE t ∂γ

=

−α n ˜t−1 +n γ 2 k

∂λE t ∂a

=

˜t−1 α(n−nE ) ∂ k 2 ˜ +n k ( t−1 ) ∂a


0, which follows from the fact that 1/γ − λE 1 =

(1 − α)(1/γ − 1) > 0 and from the fact that λE decreases as generations evolve. This information is useful for the analysis of both the conflict and the income conditions determining the respectively appropriate type of equilibrium. We now analyze these conditions in turn. As shown above, the prospects of winning an open conflict if it arises, illustrated by the guns condition displayed in (17), are crucial for determining the equilibrium strategies of elitist and proletarian individuals. A straightforward implication of Lemma 1 in the light of this condition constitutes the first result: Proposition 1. For any distribution of land resources, and for any initial level of productivity, there is a unique generation t, during whose life the 19

proletariat becomes more powerful in terms of conflict potential than the n P + γλE elite. Formally: For any γ, A0 , ∃ a unique t: kt+τ t < (1 − γ)λt+τ ∀τ ≥ 0. Proof. Note that initially the average per capita endowment in terms of physical capital is small compared to the average endowment in natural n resources, and that λP1 < 1 < λE 1 , which implies for condition (17) that k1 + P γλE 1 > (1 − γ)λ1 , that is, the elite has initially a higher conflict power than the proletariat. However, Lemma 1, implies that, after sufficiently many generations, the tides turn, because, due to limt→∞ λPt = limt→∞ λE t = 1, limt→∞ (n/kt ) = 0, and γ < 1/2, condition (17) reduces to γ < (1 − γ). Hence there exists a threshold generation t, in which the proletariat becomes stronger than the elite in terms of conflict power, while the parents of this generations still faced a conflic predominance on the side of the elite. To see that there can only be one such threshold generation t, note that, whatever the political regime, from Lemma 1 we know that λPt is strictly increasing in k˜t−1 , which is strictly increasing over generations due to technological progress. Consequently, from generation t onwards, the proletariat will be able to impose its most preferred political system through its shere conflict power. Hence, the relations of power change in the process of development in favor of the proletariat as a direct consequence of decreasing inequality in physical capital endowments, or, equivalently, income. A direct corollary of this result concerns the determinants of the timing of this transition of conflict power. Corollary 1. Everything else equal, the transition of preponderance in terms of conflict power from elite to proletariat happens earlier during the process of development, (i) the faster the technological progress, ∂t/∂a < 0; (ii) the smaller the elite, ∂t/∂γ < 0; and (iii) the smaller the stock of (and hence the importance of ) natural resources, ∂t/∂n > 0. Proof. (i) follows from the fact that faster productivity growth implies faster convergence, (ii) and (iii) from inspection of the conflict outcome condition (17). While intuitive, these results have important implications for the development process in general, but also for development policies. For example, technological spillovers from internal R&D, or alternatively international technological spillovers, tend to strenthen the position of those people in the economy that have a comparative advantage in the production processes using these technological innovations. In the current model this even implies a stronger tendency towards equalization of incomes, and, as will become clear below, towards democratization. On the other hand, it is also obvious that 20

ruling elites, that have comparative advantages in traditional technologies, benefit less, in fact lose, in terms of political power, the faster development, and therefore have a natural tendency to oppose progress. Next, turn to the income conditions. The relevant comparison that determines the elite’s decision is between incomes the individual landlords can appropriate under oligarchy and democracy, respectively. During the early stages of development, when land still plays an important role as production factor, and physical capital is still scarce, inequality in capital ownership and incomes between landlords and landless is reasonably large, as has been shown before. Under these circumstances, sharing, as is done under democracy, is likely not very desirable for the elite, making them prefer oligarchy, despite the inefficiencies implied by missing property rights protection. However, the larger the level of development, the more crucial become the income losses implied by these inefficiencies. At a certain point, the respective generation of the elite actually benefits more from removing oligarchy and the implied inefficiencies, than it loses from redistributing income to the landless. From this generation onwards, the members of the elite prefer democracy to oligarchy. This change of attitude towards prefering democracy is monotonous, that is, once a generation of landlords decides to prefer democracy, all their offspring do so as well. This transition is shown in the next result: Proposition 2. For any distribution of land resources and initial level of development, there is a unique generation of landowners, t˜, for whom democracy becomes more profitable compared to oligarchy, while their parents still preferred an oligarchic system. Moreover, their offspring will also always prefer democracy.   Formally: For any γ, A0 , ∃ a unique t˜: yt˜E+τ k˜t˜+τ (1 − ψ), nE < yt˜+τ (kt˜+τ , n) ∀τ ≥ 0. Proof. Consider once more the individual income under oligarchy, which, using equation 16, can be restated as α       ˜t , γ) − 1 (30) , k = k˜t (1 − ψ) + n 1 + α g(λE ytE k˜t (1 − ψ), nE t with

E ˜ ˜t , γ) := kt (1 − ψ)λt + n/γ . , k g(λE t k˜t (1 − ψ)λE + n

(31)

t

˜ Note that nE = nγ , that g(λE t , kt , γ) > 0 ∀t. On the other hand, average per capita income is given by condition (20). The relevant comparison thus simplifies to α   ≥  k˜t (1 − ψ) + n  ˜t , γ) − 1 α  , k 1, (32) 1 + α g(λE t < k˜t + n 21

with oligarchy being more desirable if the left hand side is larger than (or equal to) 1, while democracy is the optimal choice if the left hand side is α (k˜t (1−ψ)+n) ≤ 1 ∀t, with equality in the initial smaller than 1. Note that α (k˜t +n) period, i.e. when k˜0 = 0. Hence, at low levels of development (e.g. the initial period), the left hand side is larger than 1, implying a strict preference for oligarchy. But, for high levels of development, the elite stricly perfers democ˜ racy, since by Lemma 1 and opital’s rule, limt→∞ g(λE t , kt , γ) = 1, and  de l’Hˆ α ˜ (kt (1−ψ)+n) = (1 − ψ)α < 1. This implies the existence also limt→∞ α (k˜t +n) of an intermediate generation of landowners t˜ that strictly perfers democracy, while its parent generation still preferred oligarchy. The uniqueness of this generation t˜, and hence the monotonicity of this change in attitude
follows, since and

∂ E ˜ ˜t−1 g(λt , kt , γ) ∂k

α ˜ ∂ (kt (1−ψ)+n) α ˜ ˜ ∂ kt−1 (kt +n)

=

˜t (1−ψ)2 k

=

−ψn 2 ˜ k ( t +n)



∂λE t +n(1−ψ) ˜ ∂k t−1

˜t k

˜ ˜ ∂λE t − kt ∂ kt ˜ ˜ γ ∂k ∂k t−1 t−1 2

(k˜t (1−ψ)λEt +n) α ˜ ˜t (kt (1−ψ)+n) ∂k ˜t−1 < 0. ∂k (k˜t +n)

< 0,

An important corollary of this result is an analogous result for the landless: Even in the case where the proletariat comes to power through conflict and implements an oligarchy to its own favor as described in the “revolution” equilibrium in the previous section, the preference for a people’s oligarchy is not permanent. Rather, also this oligarchy will eventually be replaced by the desire to reap the benefits of universal suffrage and democracy. Thus, irrespective of their conflict potential, the proletariat eventually, i.e. at high levels of development, strictly prefers democracy, even to an oligarchy to their favor. Corollary 2. For any distribution of land resources for any initial level of development, there is a unique generation of the landless proletariat, t¯˜, for whom democracy becomes more profitable compared to any oligarchy, while their parents still preferred an oligarchic system with themselves being the oligarchs. Moreover, their offspring will also  always preferdemocracy. For¯ P ˜ k˜˜¯ (1 − ψ), nP < y˜¯ (kt˜+τ , n) mally: For any γ, A0 , ∃ a unique t: y ¯ t+τ t+τ t˜+τ ∀τ ≥ 0. Proof. The relevant comparison for the landless proletariat to decide between an oligarchy, in which they are in power, and democracy, is the same as ˜ (1−ψ)λP +n/(1−γ) k t . in condition (32), but with g(·) replaced by g¯(λPt , k˜t , γ) := t k˜ (1−ψ)λ P +n t t The proof is then analogous to the proof of Proposition 2. Proposition 3. Everything else equal and at a given level of development, democracy becomes more attractive for the ruling class in an oligarchy, (i) the faster the rate of technological progress, that is ∂ t˜/∂a < 0 and ∂ t¯˜/∂a < 0; 22

(ii) the larger the elite ∂ t˜/∂γ < 0 and ∂ t¯˜/∂γ < 0; (iii) the smaller the stock of resources, ∂ t˜/∂n > 0 and ∂ t¯˜/∂n > 0. Proof. The proof follows from partial derivation of the respectively appropriate (left hand side of) condition (32) and the results of Lemma 2. Note that Corollary 1 and Proposition 3 imply the same qualitative effects of the key parameters on the generation thresholds governing conflict and income conditions, t and t˜ (t¯˜), respectively. We are now in a position to characterize the development process in economic and political terms of an initially oligarchic economy under the rule of the landowning elite. Development is reflected by the (ascending) sequence of generations indexed t = 0, 1, 2, .., ∞. The results so far determine crucial generations, during whose life the political environment changes significantly, while the economic environment for the generations additionally changes according to the processes described in section 2. Initially, all generations t are stuck in the status quo equilibrium of landlord elite, namely as long as t < min{t, t˜} without any reference to t¯˜. The reason for this is given in Proposition 1, which essentially determines the maximum duration of the elitist rule under oligarchy, together with Proposition 2 implying that the elite still strictly perfers oligarchy. Hence, both conditions (18) and (19) are fulfilled: inequality in resources is high enough for the elite to defend their power, and the economy is not sufficiently developed for them to prefer democracy. However, a generation t experiences a democratization initiated by the elite if it is the first generation for which t˜ ≤ t < t with any t¯˜. As can be seen from the results of Propositions 1 and 2, this condition fulfills the necessary and sufficient conditions for democratization from above, conditions (21) and (22). In other words, while sufficient inequality still allows the elite to defent their position, it is not worth for them anymore, since they benefit more from democratization than defending oligarchy. Likewise, a generation t experiences a democratization from below if it is the first generation for which t¯˜ ≤ t ≤ t < t˜, since, for the same results as before, these conditions are equivalent to conditions (23) and (25) being met. Despite still preferring oligarchy, inequality is not sufficiently high anymore to allow the elite to defend power. However, the proletariat prefers to democratize, as demanded by condition (24). Finally, a generation t faces a revolution followed by an oligarchy of the former proletariat if it is the first generation for which t ≤ t < min{t˜, t¯˜}, which is equivalent to conditions (26) (27). However, after such a revolution, the economy democratizes under some generation t, and on initiation of the new elite (the formerly landless) of this generation, once t¯˜ ≤ t. This is essentially equivalent to an equilibrium of democratization from above under an oligarchy of the proletariat after they have appropriated all land resources from the elite in a preceeding revolution. 23

Hence, the development path of an economy is fully characterized by the chronological sequence of the three threshold generations t, t˜ and t¯˜. Essentially, there are three possible scenarios of development of an economy: a smooth transition to democracy initiated by the elite, a less smooth transition initiated by the elite under the threat of revolution, and a revolutionary development, characterized by open conflict with violent expropriation of the elite, new oligarchic structures, but, eventually a transition to democracy. These results are summarized in the following theorem: Theorem 1. The dynamic pattern of political development of an economy can be characterized as follows: • a peaceful transition from an elitist oligarchy to democracy through an elite-driven extension of the franchise, if t˜ < t, regardless of t¯˜; • a peaceful transitition from an elitist oligarchy to democracy under the threat of revolution by the proletariat, if t¯˜ < t < t˜; • or an elitist oligarchy that ends in open conflict between elite and proletariat, and is succeeded by an oligarchy of the victorious proletariat, ˜}. Eventually, this oligarchy will give way to a peaceful if t < min{t˜; t¯ transition to democracy initiated by the oligarchs. Of course, the likelihood of each regime crucially depends on the structure of the economic environment, and its development over time. Lemmata 1 and 2, Corollary 1 and and Propositon 3 make clear predictions concerning the likelihood of the regimes. This needs to be discussed more!

5

Empirical Evidence and Historical Interpretation

This research argues that the process of democratization is primarily driven by the process of economic development. In particular, we argue that an increase in the rate of technological change increases the probability of a collapse of a non democratic regime and therefore of a transition to democracy. Moreover, we suggest a channel for the interdependency between development and democracy. As countries develop, the distribution of incomes becomes more equal, and the structure of the economy changes with the production becoming more concentrated in the industrial sector rather than in the agricultural sector. A more equal distribution of resources and the dramatic changes in the productive structure, in turn, change the relative power of different social groups in the political arena and directly affect the incentives that the oligarchic elite faces to defend a non-democratic regime. A richer proletariat represents a stronger threat to the oligarchic power, just as a more capitalist elite is more likely to favor a transition to democracy. 24

Empirical evidence presented by Boix and Stokes (2003) supports the idea that economic development increases the likelihood that poor countries undergo a transition to democracy via reduction in income inequality and industrialization. They show the process of economic development goes hand in hand with the availability of human capital, the portion of farmers in the population, and the degree of occupational diversification. The latter variable is indeed a measure of the degree of industrialization while the first two capture the reduction in income inequality. Moreover, using a dynamic probit model over a sample of observations which begins in the early 19th century, they find that both the reduction of inequality and the degree of industrialization substantially increases the probability of a transition to democracy. Similar ideas are put forward in the seminal work by Dahl (1971), who studies how the level of development and the changes in inequality affect the rights to participate in elections and public offices (“inclusiveness” of a system), and the openness to public contestation (“liberalization” of a system). From a historical point of view, there are a number of examples that suggest the existence of different transition paths towards democracy and that these different paths are linked to some basic characteristcs of the society. Many regimes have been overthrown by revolutions. Most of these revolutions, for example Russia, China, and Mexico, took place in primarily agrarian and very unequal societies where the rich elites were reluctant both to economic modernization and franchise extensions. Moreover, these revolutions took place almost half a century later than the democratization in most Western European countries. In the light of our model, this is because these countries owned a huge amount of natural resources and the ruling elite successfully tried to block the technological innovation; under such conditions, the democratization process was slowed down, and finally took place in economies, which were still not industrialized. Thus, the proletariat traded-off the economic advantage of democracy against the possibility of redistributing total wealth. In other cases, the old regimes have been transformed by evolutionary processes: the new regimes have been inaugurated by incumbent leaders, who succumbed peacefully (with or without an immanent shadow of conflict) to demands for democratic changes and participated in the inauguration of a democratization processes. Two different types of non-conflictual transitions to democracy emerged in the last two centuries: the extensions of franchise that took place amid significant social unrest and revolutionary threats (“from below”) and the ones that emerged in the absence of threats to the established order (“from above”). One can think to the differences between the so called first wave of democratization and the following two. The “first wave” include the march towards democracy in the second half of nineteen century in Western Europe while the second and the third focus mainly on Latin America’s experiences in the aftermath of the Second World 25

War and since the mid-1980s. The threat of revolution and social unrest played an essential role in the establishment of voting rights for common people in Latin America. For example, in Argentina male suffrage was istitutionalized in 1912 by Saenz Pena under an explicit threat of revolutions.23 The same argument can be made for the reinstatement of democracy in Venezuela (1958) or, more recently, in Uruguay (1985) and Chile (1988). All these cases fall under the category of “democratization from below” in our taxonomy. Finally, consider the first wave of democratization. Countries such as Britain, Germany, Sweden and Switzerland all made first steps towards democratization in the aftermath of the first industrial revolutions. Moreover, these early-industrialized countries were characterized by a reasonably diversified economy and their ruling elites had already begun to shift their resources in terms of capital from agriculture to the manufacturing sector. As a consequence, the ruling elites, rather than being forced to give up the power, willingly extended the franchise because elections with a broader franchise could guarantee the development of productive relationships that require cooperation from a broad basis of the society. This extension of co-operation was clearly constrained by the unequal distribution of political power under the status-quo at the time. Lizzeri and Persico (2004) describe the case of England in great detail, but similar arguments can be extended to the great majority of western European “first comers” in terms of democracy, which begun the process of democratization in the second half of the 19th century. All these countries fall under the category of “democratization from above” in our taxonomy.

23

Acemoglu and Robinson (2004) quote David Rock (1987): “Radicals, socialists and indirectly the anarchists helped fuel the movement for reform during the early years of the century. Progessives among the elite feared the growing popular support for the Radicals, wondering where their next revolt would come from.”

26

Figure 3: A Generation’s Decision Problem under Oligarchy of the Elite







Agree


{

- Oligarchy - Arming -

E≺P

P : ytP ktP (1 − ψ)

{

P



E  P
@ Disagree @
@ @ - Conflict

@
@ E≺P
Status Quo

E

A A A Democratize A A A A A A Agree A PA A A A Disagree E  PA A A - Conflict @ @

E: ytE ktE (1 − ψ), nE

Oligarchy - Arming -

i t

= yt

{i ∈ {E, P }: y

i t

= yt

{i ∈ {E, P }: y

i t

= yt

Democracy - Arming -

Democracy - Arming -

{

Oligarchy - Arming -

E: ytE ktE (1 − ψ)

E  P illustrates the case of inequality “≥”, while E ≺ P illustrates the case of inequality “