why did the west extend the franchise? democracy, inequality… · why did the west extend the...

WHY DID THE WEST EXTEND THE FRANCHISEDEMOCRACY INEQUALITY AND GROWTH IN

HISTORICAL PERSPECTIVE

DARON ACEMOGLU AND JAMES A ROBINSON

During the nineteenth century most Western societies extended voting rightsa decision that led to unprecedented redistributive programs We argue that thesepolitical reforms can be viewed as strategic decisions by the political elite toprevent widespread social unrest and revolution Political transition rather thanredistribution under existing political institutions occurs because current trans-fers do not ensure future transfers while the extension of the franchise changesfuture political equilibria and acts as a commitment to redistribution Our theoryalso offers a novel explanation for the Kuznets curve in many Western economiesduring this period with the fall in inequality following redistribution due todemocratization

I INTRODUCTION

The nineteenth century was a period of fundamental politicalreform and unprecedented changes in taxation and redistribu-tion Britain for example was transformed from an lsquolsquooligarchyrsquorsquorun by an elite to a democracy The franchise was extended in1832 and then again in 1867 and 1884 transferring voting rightsto portions of the society with no previous political representationThe decades after the political reforms witnessed radical socialreforms increased taxation and the extension of education to themasses Moreover as noted by Kuznets inequality which waspreviously increasing started to decline during this period theGini coefficient for income inequality in England and Wales hadrisen from 0400 in 1823 to 0627 in 1871 but fell to 0443 in 1901Two key factors in the reduction in inequality were the increase inthe proportion of skilled workers [Williamson 1985] and theredistribution of income toward the poorer segments of thesociety For example taxes rose from 812 percent of NationalProduct in 1867 to 188 percent by 1927 and the progressivity ofthe tax system increased substantially (see Lindert [1989])

We would like to thank two anonymous referees and Pranab BardhanRuth Collier Steven Durlauf Jeffry Frieden Edward Glaeser Claudia GoldinPeter Lindert Torsten Persson Dani Rodrik John Roemer Kenneth SokoloffPeter Temin Erik Thorbecke Jaume Ventura Michael Wallerstein and seminarparticipants at Boston University the University of Chicago Cornell UniversityHarvard University the National Bureau of Economic Research Universidad delos Andes Singapore National University and the World Bank for helpfulcomments and suggestions

r 2000 by the President and Fellows of Harvard College and the Massachusetts Institute ofTechnologyThe Quarterly Journal of Economics November 2000

1167

During the late nineteenth and early twentieth centuries thefranchise was also extended in most other Western societiesDemocratization was again followed by increased redistributionand in most cases by the downturn of the Kuznets curve

These events are hard to understand with our existingtheories If democratization is likely to lead to increased taxationand redistribution (eg Meltzer and Richard [1981]) why shouldthe elite extend the franchise Our answer is that the elite wereforced to extend the franchise because of the threat of revolutionWe argue that extending the franchise acted as a commitment tofuture redistribution and prevented social unrest In contrast todemocratization the promise by the elite to redistribute in thefuture while maintaining political power would not have beencredible

The second contribution of our paper is to point out thepotential link between democratization and the Kuznets curve1Rising inequality often associated with industrialization in-creases social unrest and induces democratization Democratiza-tion in turn opens the way for redistribution and mass educationand reduces inequality Interestingly in line with this approachin a number of Western economies the peak of the Kuznets curvecoincides with the extension of the franchise

The thesis that the elite extended the franchise in order toavoid a revolution or social unrest is at the heart of our paperAlthough we are aware of no other papers in the economicsliterature a nonformal literature in political science startingwith the seminal work of Lipset [1960] and Moore [1966] hasstudied the origins of democracy To the best of our knowledgehowever no previous work explained franchise extension as acommitment device to future redistribution or pointed out the linkbetween democratization and the Kuznets curve though Ther-born [1977] and Rueschemeyer Stephens and Stephens [1992]have stressed the role of social conict in democratization2 In

1 Other theories of the Kuznets curve include Kuznetsrsquo [1955] originalconjecture that the curve was generated by the dual economy dynamics generatedby the switch from the agricultural to the industrial sector Lindertrsquos [1986]hypothesis that the curve resulted from the falling importance of income from theland Williamsonrsquos [1985] argument that technological change increased wagesrelative to capital income and Aghion and Boltonrsquos [1997] model where accumula-tion by the rich eventually pushes down the interest rate sufficiently to allow thepoor to invest and catch up

2 In a different context North and Weingast [1989] have argued that theintroduction of the English Parliament in the seventeenth century was a commit-ment to low taxes in the future Our paper is also related to models of political

QUARTERLY JOURNAL OF ECONOMICS1168

Section IV below we outline three alternative theories anddiscuss whether any of these appear to give a better description ofthe historical events than our preferred story The rst which werefer to as the Enlightenment view is that the elite extended thefranchise because their social values changed The second theoryis that political competition within the elite led to the extension ofthe franchise when one of the factions in an attempt to increaseits support brought new groups into the political system (seeHimmelfarb [1966] and Collier [1999]) The third which is akin tothe famous thesis of Moore [1966] is that the middle class was thedriving force behind the extension of the franchise in part hopingto shift the future balance of power Although all three stories arelikely to have contributed to democratization in Europe in thenineteenth century especially to the extension of the franchise towomen our interpretation is that in Britain France Germanyand Sweden the threat of revolution was the major factor in theextension of the franchise to the poorer segments of the society

II A MODEL OF DEMOCRATIZATION

In this section we develop our main model of democratizationWe postpone a discussion of alternative models of franchiseextension until Section IV

We consider an innite horizon economy with a continuum 1of agents A proportion l of these agents are lsquolsquopoorrsquorsquo while theremaining 1 2 l form a rich lsquolsquoelitersquorsquo Throughout the papersuperscript p will denote a poor agent and r will denote a richagent (or member of the elite) We will treat all poor agents asidentical and all members of the elite will also be identicalInitially political power is concentrated in the hands of the elitebut l 12 so if there is full democracy the median voter will be apoor agent

There is a unique consumption good y with price normalizedto unity and a unique asset h (which should be thought of as acombination of human and physical capital and land) We beginour analysis of the economy at time t 5 0 where each poor agenthas capital h0

p and each member of the elite has h0r h0

p $ 1 Inthis section these stocks are exogenous so we drop time sub-scripts Accumulation is investigated in Section III

conict and revolution including Roemer [1985] Grossman [1991 1995] andAdesand Verdier [1993]

WHY DID THE WEST EXTEND THE FRANCHISE 1169

There are two methods of producing the nal good bothlinear in capital The rst is a market technology Y t

m 5 AH tm

where H tm is the amount of capital devoted to market production

The second is an lsquolsquoinformalrsquorsquo or home production technology Y th 5

BH th where H t

h is the amount of capital used in home productionNaturally we have H t

h 1 H tm 5 H e h i di We assume that A

B thus market production is always more productive The onlyrole of home production in our analysis is to ensure an equilibriumtax rate less than 100 percent because while taxes can be imposedon the market sector home production is not taxable A high valueof B implies that only limited taxes can be levied on income

All agents have identical preferences represented by a linearindirect utility function over net income and a discount factor b [(01) Posttax income is given by yt

i (1 2 t t)Ah i 1 Tt for i 5 p rwhere t t is the tax rate on income and Tt $ 0 is the transfer thatthe agent receives from the state We assume throughout thattaxes and transfers cannot be person specicmdashhence Tt and t t arenot indexed by i The government budget constraint thereforeimplies that Tt 5 t tAH m

The l poor agents though initially excluded from the politicalprocess can overthrow the existing government and take over thecapital stock in any period t $ 0 We assume that if a revolution isattempted it always succeeds Revolution provides a window ofopportunity for a large-scale redistribution of assets away fromthe rich to the poor so the poor take over control of the capitalstock of the economy but a fraction 1 2 microt of the capital stock getsdestroyed in the process3 Therefore if there is a revolution attime t each poor agent receives a per-period return of micro tAH l inall future periods total income in the economy is microtAH and isshared between l agents We assume that micro is stochastic andchanges between two values microh and microl 5 0 with Pr(microt 5 microh) 5 qirrespective of whether microt 2 1 5 microh or micro l The fact that micro uctuatescaptures the notion that some periods may be more conducive tosocial unrest than others It will also enable us to model the idea

3 The assumption that the elite receive nothing after a revolution is only forsimplicity The functional form assumption that the poor receive a fraction micro of theassets of the economy is also inessential Instead the important feature is thatrevolution is more attractive to the poor in a more unequal society which receivessupport from the positive association between inequality and instability docu-mented in Muller and Seligson [1987] and Alesina and Perotti [1996]

More generally one could imagine that certain assets such as human capitalare harder to expropriate than others Since we interpret h as a mixture of physicaland human capital and land one can imagine that part of it can be expropriatedwhile the rest will be wasted during a revolution


that a promise to redistribute today may not materialize due tochanges in circumstances tomorrow A low value of micro means thata revolution is very costly while a low value of q implies thatthe threat of revolution is rare perhaps because the poor areunorganized4

Finally in each period the elite have to decide whether or notto extend the franchise If it is extended the economy becomes ademocracy and the median voter a poor agent sets the tax rateWe assume that if voting rights are extended they cannot berescinded so the economy always remains a democracy5

The timing of events within a period can be summarized asfollows

1 The state micro is revealed2 The elite decide whether or not to extend the franchise If

they decide not to extend the franchise they set the taxrate

3 The poor decide whether or not to initiate a revolution Ifthere is a revolution they share the remaining output Ifthere is no revolution and the franchise has been ex-tended the tax rate is set by the median voter (a pooragent)

4 The capital stock is allocated between market and homeproduction and incomes are realized

The analysis can be simplied by exploiting two features ofthe model First the capital allocation decision takes a simpleform if t t tˆ (A 2 B)A then all agents allocate their capital tohome use thus H t

m 5 0 On the other hand if t t tˆ then H tm 5 Ht

is a best-response No voter would ever choose t t tˆ so we canrestrict attention to t t tˆ and H t

m 5 Ht which reduces the numberof actions to be considered Second all members of the elite haveidentical preferences so we can treat them as one player Also allpoor agents have the same preferences and when it comes towhether or not to participate in a revolution there is no lsquolsquofree-rider problemrsquorsquo because if an agent does not take part in the

4 Another determinant of the frequency of the threat of revolution may beurbanization and urban concentration (see eg DiPasquale and Glaeser [1998])

5 This is not to deny that coups happen Nevertheless once voting rights areextended and political parties are formed it is relatively costly for any group toexclude the rest from the political process We discuss coups in our more recentpaper [Acemoglu and Robinson 1999]

Also notice that democratization is all-or-nothing Extending the franchise toonly a segment of the citizens would not be useful in this model if the poor becomethe majority then the consequences are the same as in the all-or-nothing case andotherwise the reform has no commitment value


revolution he can be excluded from the resulting redistribution6

So we can treat all poor agents as one player This economy cantherefore be represented as a dynamic game between two playersthe elite and the poor

In the text we characterize the pure strategy Markov PerfectEquilibria of this game in which strategies only depend on thecurrent state of the world and not on the entire history of thegame Although the focus on Markovian equilibria is natural inthis setting for completeness we discuss non-Markovian equilib-ria in the Appendix and show that they do not change our generalresults The state of the system consists of the current opportunityfor revolution represented by either micro l or microh and the politicalstate (democracy or elite control) More formally let s r(microP) be theactions taken by the elite when the state is micro 5 microh or microl and P 5 E(elite in power) or D (democracy) This consists of a decision toextend the franchise f when P 5 E and a tax rate t r when f 5 0(ie when the franchise is not extended) Clearly if f 5 0 Premains at E and if f 5 1 P switches to D forever Similarlys p(microP f t r) are the actions of the poor which consist of a decisionto initiate a revolution r ( r 5 1 representing a revolution) andpossibly a tax rate t p when the political state is P 5 D Theseactions are conditioned on the current actions of the elite whomove before the poor agents according to the timing of eventsabove Then a (pure strategy Markov Perfect) equilibrium is astrategy combination s r(microP) s p(microP f t r) such that s p and s r

are best-responses to each other for all micro and PWe can characterize the equilibria of this game by writing the

appropriate Bellman equations Dene V p(R) as the return topoor agents if there is a revolution starting in state micro 5 microh Sinceonly the value of microh at the time of the revolution matters V p(R) 5microhAH l (1 2 b ) which is the per-period return from revolution forthe innite future discounted to the present Also because therich lose everything V r(R) 5 0 Finally since micro l 5 0 the poorwould never attempt a revolution when micro 5 micro l

In the state (micro lE) the elite are in power and there is nothreat of revolution so in any Markov Perfect Equilibrium f 5 0and t r 5 0 Therefore the values of poor and rich agents j 5 p or r

6 Although there is no free-rider problem there may be a coordination issuewhereby each poor agent expects others not to take part in the revolution andprefers not to do so himself We ignore this problem here


are given by

(1) V j(microlE) 5 Ahj 1 b [(1 2 q)V j(microlE) 1 qV j(microhE)]

Next consider the state (microhE) and suppose that the eliteplay f 5 0 and t r 5 0 that is they neither extend the franchisenor redistribute to the poor Then we would have V p(microhE) 5Ahp(1 2 b ) The revolution constraint is equivalent to V p(R) Vp(microhE) so that without any redistribution or franchise exten-sion the poor prefer to initiate a revolution when micro 5 microh Weassume

ASSUMPTION 1

hr

hp

l (1 2 microh)

(1 2 l )(microh 2 (1 2 b )((A 2 B)A)

This is a condition slightly stronger than the revolution con-straint V p(R) V p(microhE) Specically it ensures not only thatthe revolution constraint binds but also that redistribution onlyfor one period is not sufficient to prevent a revolution7 Thisfeature will simplify the discussion below

Since the revolution is the worst outcome for the elite theywill attempt to prevent it They can do this in two different waysFirst the elite can choose to maintain political power f 5 0 butredistribute through taxation In this case the poor obtainV p(microhE t r) where t r is the tax rate chosen by the rich With eitheraction by the elite the poor may still prefer a revolution Thus

V p(microhE) 5 max V p(R) f V p(D) 1 (1 2 f )V p(microhEt r)

where V p(D) is the return to the poor in democracy Instead thereturn to the poor when the elite choose the redistributionstrategy is

(2) V p(microhEt r) 5 (1 2 t r)Ahp 1 t rAH1 b [qV p(microhE t r) 1 (1 2 q)V p(micro lE)]

The rich redistribute to the poor taxing all income at the rate t rThe poor therefore receive net income (1 2 t r)Ahp from their owncapital and a transfer of T 5 t rAH If in the next period we are stillin state micro 5 microh redistribution continues But if the state switches

7 More explicitly this condition is derived from microhH l (1 2 b ) hp(1 2 b ) 1(A 2 B)(H 2 hp)A where the second term on the right is the one-period transferfrom the rich to the poor


to micro 5 micro l redistribution stops and the poor receive V p(micro lE) Thiscaptures the discussion in the introduction that the elite cannotcommit to future redistribution unless the future also poses aneffective revolution threat Notice that t r tˆ that is the elitecannot tax themselves at a rate higher than tˆ (A 2 B)A if t tˆthen each (rich) agent would privately prefer to use all their assetsin the home sector reducing aggregate tax revenues to zero

The second strategy to prevent revolution is to extend thefranchise f 5 1 Since l 12 in a democracy the median voter is apoor agent and wants as much redistribution as possible Redistri-bution has no allocative cost so long as t tˆ so the equilibriumtax rate is t t 5 tˆ (A 2 B)A and Tt 5 (A 2 B)H The returns topoor and rich agents in democracy are therefore

V p(D) 5Bhp 1 (A 2 B)H

1 2 band V r(D) 5

Bhr 1 (A 2 B)H

1 2 b

We simplify the exposition by restricting attention to the areaof the parameter space where democratization prevents a revolu-tion ie V p(D) V p(R) Thus we assume

ASSUMPTION 2

Bhp 1 (A 2 B)H microhAH l

To determine whether the elite can prevent a revolution withthe redistribution strategy let V p(microh E q) be the maximum utility(as a function of the parameter q) that can be given to the poorwithout extending the franchise This maximum utility is achievedby setting t r 5 tˆ in (2) Therefore combining (1) and (2) we obtain

(3) Vp(microhE q) 5 V p(microhE tˆ)

5Bhp 1 (A 2 B)H 2 b (1 2 q)(A 2 B)(H 2 hp)

1 2 b

If V p(microhE q) V p(R) then the maximum transfer that can bemade when micro 5 microh is not sufficient to prevent a revolution Noticethat V p(microhE q 5 1) 5 Vp(D) V p(R) by Assumption 2 andV p(microhE q 5 0) 5 Ahp(1 2 b ) 1 (A 2 B) (H 2 hp) V p(R) byAssumption 1 Moreover V p(microh E q) is monotonically and continu-ously increasing in q Therefore there exists a unique q [ (01)such that V p(microhE q) 5 Vp(R) Finally note that V r(microhEt r) isdecreasing in t r and for all t r it is greater than V r(D) The latter


fact follows because when there is a democracy t 5 tˆ in all periodswhereas with the power in the hands of the elite t [ (0 tˆ]whenever micro 5 microh but t 5 0 when micro 5 microl From this discussion thefollowing characterization of the equilibrium follows immediately

PROPOSITION 1 Suppose that Assumptions 1 and 2 hold Then forall q THORN q there exists a unique pure strategy Markov PerfectEquilibrium such that

1 If q q then the revolution threat will be met byfranchise extension More formally the equilibrium iss r(micro l E ) 5 ( f 5 0 t 5 0) s r (micro h E ) 5 ( f 5 1 )s p(micro h E f 5 0 t ) 5 ( r 5 1) s p (micro h E f 5 1 ) 5( r 5 0t 5 tˆ) and s p(microhD) 5 (t 5 tˆ)

2 If q q then the revolution threat will be met bytemporary redistribution More formally s r(micro l E) 5( f 5 0t 5 0) s r(microhE) 5 (f 5 0t r) where t r [ (0 tˆ) isdened by V p(R) 5 V p(microhEt r) and s p(microhE f 5 0t ) 5( r 5 0) for all t $ t r Also off the equilibrium paths p(microhE f 5 0t ) 5 (r 5 1) for all t t r s p(microhE f 5 1 ) 5( r 5 0t 5 tˆ) and s p(microhD) 5 ( t 5 tˆ)

Starting with the elite in power if q q then the rich set azero tax rate when micro 5 micro l and extend the franchise when the stateswitches to micro 5 microh The poor play the optimal strategy of initiatinga revolution if the state is micro 5 microh and the franchise has not beenextended After the franchise extension the median voter is a pooragent and sets the tax rate t 5 tˆ In contrast when q q the richcan prevent a revolution by redistributing So in the state micro 5 microlthey set t 5 0 and when micro 5 microh they set a tax rate t r just highenough to prevent a revolution This strategy combination is theunique pure strategy (Markov Perfect) Equilibrium of the game

In the Appendix we show that even without the restriction toMarkov Perfect Equilibria similar results obtain revolution canbe stopped with temporary redistribution when q q whereq q So franchise extension can be prevented for a larger setof parameter values but if q q the elite can prevent arevolution only by extending the franchise

There are two main conclusions to be drawn from thisanalysis

First even though the elite face a lower future tax burdenwith redistribution than under democracy they may prefer toextend the franchise This is because when q q redistribution


is not sufficient to prevent a revolution With q low the revolutionthreat is transitory so the poor realize that they will only receivetransfers for a short while Redistribution when micro 5 microh cantherefore be viewed as a noncredible promise of future redistribu-tion by the elite Unconvinced by this promise the poor wouldattempt a revolution The revolution is only prevented by fran-chise extension

Second perhaps paradoxically a high q makes franchiseextension less likely A high q corresponds to an economy in whichthe poor are well organized so they frequently pose a revolution-ary threat8 A naive intuition may have been that in this casefranchise extension would be more likely This is not the casehowever because with a frequent revolutionary threat futureredistribution becomes credible This result may explain why inthe nineteenth century Germany the country with the mostdeveloped socialist party at the time instituted the welfare statewithout franchise extension while Britain and France extendedthe franchise We return to this issue in Section IV

The comparative statics with respect to inequality microh and Balso deserve a brief discussion As Assumption 1 shows a certainlevel of inequality is necessary for the revolution constraint tobind So a very equal society may never democratize or democra-tize with considerable delay because there is relatively little socialunrest More interestingly in a more unequal society it is alsomore difficult to stave off a revolution without democratization9Specically there exists a threshold level of inequality hr hp forgiven q such that if hrhp hr hp then the revolution can beprevented by temporary redistribution but if hrhp hr hp theonly way to prevent a revolution is democratization

An increase in microh also tightens the revolution constraintWhen q q this has no effect at rst since the threat ofrevolution already ensured democratization However if microh in-creases further Assumption 2 would be violated so that revolu-tion could not be prevented even with democratization In con-

8 Alternatively if microl is sufficiently greater than zero then even in this statethe elite have to redistribute to the poor In this case a high value of microl would alsolead to the same result

9 To see this take the equation Vp (microhE q) 5 Vp(R) divide both sides byH and totally differentiate This gives dqd(hpH ) 0 so when inequalityincreases q also increases Equivalently x q and this equation denes a criticallevel of inequality hp h r Since Vp(R) increases faster in the level of inequalitythan Vp(microhE q) with inequality greater than hp hr revolution cannot beprevented by temporary redistribution


trast if q q a higher level of microh affects scal policy by forcingthe elite to choose a more redistributive policy to prevent revolu-tion If it increases further it may induce democratization Anincrease in B limits the taxes that can be set on the rich So if Bincreases it becomes harder to prevent a revolution with tempo-rary redistribution because the level of taxes that can be setduring this period are lower Moreover if B increases furtherAssumption 2 would be violated once again leading to a revolu-tion along the equilibrium path The general principle that followsfrom this discussion is that revolutions are harder to prevent insocieties where democracy creates only limited gains for the poor

III A MODEL OF GROWTH AND INEQUALITY DYNAMICS

The previous section established that the elite may be forcedto extend the franchise We now explore the implications ofpolitical reform for growth and inequality

Now each agent (or forward-looking dynasty) can decide toinvest in order to increase their capital further In particular thecapital of agent i at time t 1 1 given his stock h t

i at t is

(4) h t1 1i 5 d et

ih t

i

where d 1 and eti is an indicator that takes the value 1 if there is

investment Investment costs an amount Z and is indivisible Thisnonconvexity in the investment technology implies that very pooragents will be unable to accumulate as in Galor and Zeira [1993]From the budget constraint we have that ct

i 1 etiZ yt

i Anindividual earns income Ah t

i in market activity or Bhti in home

production as in the previous section with B A We also assumethat investment in capital is always protable (but b d 1) andfurthermore Bh0

r 1 (A 2 B) ( l h0p 1 (1 2 l )h0

r ) Z which impliesthat the rich have enough income to invest even if they weretaxed at the maximum tax rate tˆ (A 2 B)A The revolutionthreat is parameterized as before the poor receive a fraction microt ofthe total resources Ht 5 l h t

p 1 (1 2 l )htr and share it between

themselves microt takes the value microh with probability q and micro l 5 0with probability 1 2 q

We continue to limit our attention to pure strategy MarkovPerfect Equilibria Moreover because equilibria are now moreinvolved instead of giving a comprehensive taxonomy we focus ondemocratic development paths (see Acemoglu and Robinson [1997]for nondemocratic development paths)


Consider the case where Z Ah0p so that the poor cannot

accumulate unless they receive transfers When there is no threatof a revolution there is no taxation As a result the richaccumulate capital and their income grows at the rate d 2 1while the poor are unable to accumulate Therefore incomeinequality increases steadily This continues until the threat ofrevolution occurs At this point there are two possibilities

First the rich could extend the franchise After democratiza-tion the median voter is a poor agent and sets the maximum taxrate tˆ Then the dynamics of inequality and income depend onwhether this transfer from the rich to the poor is sufficient toenable the poor to accumulate capital Suppose that the threat ofrevolution occurs for the rst time at t Then the crucial conditionis whether

(5) Bh0p 1 (A 2 B)Ht $ Z

where Ht 5 l h0p 1 (1 2 l )ht

r is the total capital of the economy attime t This expression follows from the fact that a poor agentkeeps (1 2 tˆ) of his income Bh0

p and receives a fraction tˆ of totalincome in the economy (A 2 B)Ht If condition (5) holds then thetransfer is sufficient to push the income of the poor above theaccumulation threshold and the poor start accumulating andinequality drops10 This will be the mechanism leading to theKuznets curve in our economy We assume that this conditionholds at time t 5 0 and since Ht grows continuously it will hold atall future dates too11

To determine whether the rich will be forced to extend thefranchise we once again compare the return from a revolutionwith the return from receiving temporary redistribution undercontinued elite control If the poor undertake a revolution theyreceive

(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b

10 Notice that this condition does not guarantee that the poor will be able toaccumulate without transfers Also because we have a linear economy after thepoor start accumulating inequality drops discretely and remains constant In aprevious version of the paper [Acemoglu and Robinson 1997] we analyzed the casewith a convex production function where inequality falls steadily after franchiseextension

11 Even if inequality (5) does not hold at time t it will eventually hold sincethe rich will continue to accumulate so ht

r will grow steadily At the point wherethis condition holds say time t the poor will start accumulating Then inequalitywould drop after franchise extension but increase again after that until time t isreached when it will level off


where Ht 5 l h0p 1 (1 2 l )h t

r and h t1 1r 5 d h t

r and we used the factthat after a revolution they take over the productive resources ofthe rich losing a fraction 1 2 microh of all the capital in the processWe have also assumed that the transfer of resources during arevolution is sufficient for the poor to accumulate12

The rich can try to defuse the threat of revolution usingtemporary redistribution instead of franchise extension To deter-mine whether this is possible we write the value functions for thepoor under elite control When there is no threat of revolution

V p(micro lE htph t

r) 5 Ah tp 2 et

pZ

1 b [qV p(microhE h t 1 1p ht 1 1

r ) 1 (1 2 q)V p(microlE h t1 1p h t 1 1

r )]

where ht 1 1r 5 d h t

r and h t 1 1p 5 d e t

p h tp Whether et

p 5 0 or 1 depends onwhether the poor are above the accumulation threshold iewhether Z Ah t

pThe value function for the poor when the state switches to microh

depends on the elitersquos strategy We are interested in the maximumutility that they can give to the poor without extending thefranchise Vp so we suppose that they set the maximum tax rate tˆthen

Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]

where T p tÂ(Ht 2 htp) (A 2 B)(Ht 2 h t

p) is the net transferthey receive and h t 1 1

r 5 d htr and this time h t1 1

p 5 d htp also because

the poor are receiving the transfer T pAn analysis similar to the previous section immediately

implies that there exists qt such that if q qt then the elitecannot prevent a revolution by redistributing temporarily13 More-over it is straightforward to verify that V p(micro lE h t

phtr) increases

in h tr faster than V p(R h t

ph tr) intuitively as ht

r increases revolu-tion becomes more attractive for the poor Therefore qt is

12 A sufficient condition for this is microh( l h0p 1 (1 2 l )h0

r ) l Z If this holds att 5 0 then it also holds at all subsequent dates We are also assuming that theequivalent of Assumption 2 Bht

p 1 (A 2 B)Ht microhAHt l holds at all t that isfranchise extension is sufficient to prevent the revolution In the absence of thisassumption when inequality increases substantially democratization would beinsufficientto prevent the revolution so the elite may initiate redistribution beforethis point in order to prevent a future revolution

13 Notice that Vp is continuously increasing in q and as q 0 the poor willnever receive any more transfers so prefer a revolution Therefore there exists acritical value of q such that when q is below this threshold temporary redistribu-tion is not attractive enough to prevent a revolution


decreasing in t as inequality increases the threat of revolutionbecomes harder to prevent by redistribution alone Let us denethe rst time when qt q as t

When turbulence arises at some time after t the outcome isdemocratization and the Kuznets curve Inequality has increasedconsiderably by this time so the elite cannot prevent social unrestby temporary measures alone and are forced to extend thefranchise14 In the resulting democracy the median voter is a pooragent and votes for redistributive taxation at the rate tˆ With thistransfer from the rich the poor start accumulating as well andinequality which had been increasing since t 5 0 drops15 In ourview this case gives a stylized description of the experiences ofBritain France and Sweden In these instances the threat ofrevolution forced democratization and inequality which waspreviously increasing started to decline in large part due tomajor redistributive efforts including increased taxation invest-ment in the education of the poor and labor market reform (seethe discussion in Section IV)

Alternatively if the revolution threat occurs before t8 t theelite can prevent it with temporary measures because inequalityis limited If Z d Ah0

p the one-period temporary redistribution isnot sufficient to enrich the poor sufficiently that they can accumu-late without transfers As a result when the revolution threatgoes away transfers stop and inequality grows again A furtherperiod of turbulence may then lead to democratization and to aKuznets curve type behavior We believe that this type of lsquolsquode-layedrsquorsquo Kuznets curve may capture the sequence of events in theGerman case In Germany social unrest was initially met withredistribution but eventually the shock of the First World Warcreated further unrest and induced democratization (in 1919)Redistribution increased and inequality fell after this date

14 It might sometimes be optimal for the elite to collectively reduce accumu-lation in order to avoid the future threat of revolution (and hence democratization)Nevertheless because each member of the elite chooses their optimal savingindependently such an equilibrium never arises in our model

15 The implications for growth are also straightforward before franchiseextension only the rich accumulate so the overall growth rate of the economy islow but after democratization all agents accumulate so the economy grows at therate d 2 1 The result that democratization increases growth is not general Itdepends on whether the gains from relaxing the credit constraints emphasizedfor example by Galor and Zeira [1993] and Benabou [2000] exceed the costs ofdistortionary taxation emphasized by among others Alesina and Rodrik [1994]and Persson and Tabellini [1994] Interestingly in their empirical work usinghistorical data Persson and Tabellini nd that the growth rate increases when agreater fraction of the population are enfranchised


If in contrast Ah0p Z d Ah0

p and the threat of revolutionoccurs at t8 t then the outcome is a nondemocratic developmentpath In this case the temporary redistribution at time t8 issufficient to enable the poor to accumulate steadily and inequalityremains constant thereafter After the poor start accumulatingthe return to revolution at time t in the state microh is

V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b

where Ht 1 1 5 d Ht because both the rich and poor accumulate Onthe other hand the return to remaining in a nondemocraticregime at time t in the state microh is

Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d

where T p tÂ(Ht 2 h tp) is the maximum net transfer to the poor

Intuitively the poor accumulate irrespective of whether theyreceive transfers or not Overall they receive a net transfer T p

today and expect to receive it in the future with probability q buttake into account that it will be larger in the future because ofincome growth Therefore both V p(microhE ht

ph tr) and V p(R h t

ph tr)

(net of Z(1 2 b )) grow at the rate d 2 1 and the revolutionconstraint does not change over time Since the threat of revolu-tion at time t8 could be prevented without democratization futurerevolution threats can also be prevented by redistribution There-fore in this case because inequality stops growing and the gainsfrom it are shared between the rich and the poor social unrest isweak and democratization is avoided forever or at least delayedconsiderably This nondemocratic development path may be rele-vant to South Korea and Taiwan Both countries used landredistribution early on in response to the threat of revolutionfueled by the communist regime in China16 They were subse-quently relatively equal and did not democratize until much laterA related case is Indonesia after 1965 where the rural strength ofthe communists induced large-scale scal redistribution and

16 Chrsquoen Chrsquoeng the governor of Taiwan at the time of the reforms forexample explains this as follows lsquolsquo the situation on the Chinese mainland wasbecoming critical and the villages on the island were showing marked signs ofunrest and instability It was feared that the Communists might take advantage ofthe rapidly deteriorating situationrsquorsquo [Haggard 1990 p 82] The extent of the landredistribution in both countries was quite large For example in the 1949 reformSouth Korea redistributed 50 percent of the agricultural land while the Taiwaneseland reforms of 1949ndash1953 redistributed 246 percent of the land [Ho 1978]


major educational reforms These reforms led to rapidly fallinginequality in the 1970s and 1980s and a long-lived nondemocraticregime17

IV HISTORICAL PERSPECTIVE

In this section we discuss the historical evidence related tothe extension of the franchise Our objective is to provide apreliminary assessment of whether our theory gives a gooddescription of the salient features of Western democratization Weoutline the major events of the period and document the impor-tance of the threat of revolution We then discuss three alternativehypotheses that could explain why the franchise was extendedand the evidence in favor of and against these approaches In thenal two subsections we discuss other empirical implications ofour model In all cases our most detailed evidence is from Britainbut we also discuss evidence from France Sweden and Germany18

A The Threat of Revolution and Franchise Extension

Britain In Britain the franchise was extended in 1832 andthen again in 1867 and 1884 (and later in 1919 and 1928 when allwomen were nally allowed to vote)19 When introducing theelectoral reform to the British parliament in 1831 the primeminister Earl Grey said lsquolsquoThere is no-one more decided againstannual parliaments universal suffrage and the ballot than am I The Principal of my reform is to prevent the necessity ofrevolution I am reforming to preserve not to overthrowrsquorsquo(quoted in [Evans 1983]) This view of political reform is shared bymodern historians such as Briggs [1959] and Lee [1994] Forexample Darvall [1934] writes lsquolsquothe major change of the rstthree decades of the nineteenth century was the reform ofParliament by the 1832 Reform Act introduced by the Whigs as a measure to stave off any further threat of revolution byextending the franchise to the middle classesrsquorsquo In fact the yearspreceding the electoral reform were characterized by unprece-

17 The Gini coefficient in Indonesia was 041 during the 1970s and fell to030 during the 1980s [Campos and Root 1996]

18 The evidence from the U S democratization experience is more mixed Ina previous version of this paper [Acemoglu and Robinson 1997] we discussed whythe U S case is different

19 In our formal model only a one-time franchise extension was possibleThe model can be extended by introducing more heterogeneity (such as a middleclass) to incorporate multiple gradual extensions


dented political unrest including the Luddite Riots from 1811ndash1816 the Spa Fields Riots of 1816 the Peterloo Massacre in 1819and the Swing Riots of 1830 (see Stevenson [1979] for anoverview) The reforms that extended political power from anarrow elite to larger sections of the society were immediatelyviewed as a success not because of some ideal of enlightenment ordemocracy but because the threat of revolution and furtherunrest were avoided (see Lee [1994])

Although the 1832 Reform Act reduced property and wealthrestrictions on voting and increased the total electorate to 813000the majority of British people (the remaining 23 million) could notvote and the elite still had considerable scope for patronage since123 constituencies the lsquolsquorotten-boroughsrsquorsquo contained less than1000 voters There is also evidence of continued corruption andintimidation of voters until the Ballot Act of 1872 and the Corruptand Illegal Practices Act of 1883 These practices were reversedlater in the process of increased representation which gainedmomentum with the Chartist movement during the 1830s and1840s (see Briggs [1959]) The response of the elite to the Chartistmovement was again one of preventing further unrest Forexample during the 1850s Lord John Russell made severalattempts to introduce reform arguing that it was necessary toextend the franchise to the upper levels of the working classes as ameans of preventing the revival of political radicalism But as Lee[1994 p 137] notes lsquolsquoThe House of Commons was largely hostileto reform because at this stage it saw no need for itrsquorsquo

This had changed by 1867 largely due to a sharp businesscycle downturn that caused signicant economic hardship andincreased the threat of violence Lee writes lsquolsquoas with the rstReform Act the threat of violence has been seen as a signicantfactor in forcing the pace (of the 1867 Reform Act) history wasrepeating itselfrsquorsquo This interpretation is supported by many otherhistoriansmdashfor example Trevelyan [1937] and Harrison [1965]The Act was preceded by the founding of the National ReformUnion in 1864 and the Reform League in 1865 and the Hyde Parkriots of July 1866 provided the most immediate catalyst Searle[1993 p 225] argues that lsquolsquoreform agitation in the country clearlydid much to persuade the Derby ministry that a Reform Bill anyReform Bill should be placed on the statute book with a minimumof delayrsquorsquo As a result of these reforms the total electorate wasexpanded from 14 million to 252 million and working-classvoters became the majority in all urban constituencies The


electorate was doubled again by the Reform Act of 1884 and theRedistribution Act of 1885 removed many remaining inequalitiesin the distribution of seats (see Wright [1970]) Once again socialdisorder appears to have been an important factor behind the1884 act as put by Hayes [1982] lsquolsquoAt the bottom the course ofevents in mid-November reected the importance of the battle outof doorsrsquorsquo

Other Countries In France the 1830 revolution led to a highlyrestricted democratic regime where property restriction limitedthe electorate to about 075 percent of the population (see Cole andCampbell [1989]) The collapse of the Orleanist monarchy in the1848 revolution led to the Second Republic with the introductionof universal male suffrage in 1849 (see Collier [1999] Chapter 2p 10) The effect of this was cut short however rst by restric-tions on voting rights introduced in 1850 disenfranchising 28million men and then by the coup of Louis Napoleon in 1851Historians split this period into two phases the lsquolsquoauthoritarianrsquorsquophase from 1852 to 1860 and the lsquolsquoliberalrsquorsquo phase from 1860 untilthe defeat of the French armies in the Franco-Prussian War of1870 The defeat in the war led to further unrest (in particular theParis Commune) and to the collapse of the regime making way forthe Third Republic (see Zeldin [1958] Plessis [1985] and Price[1995])

The history of modern democracy in Germany starts with the1848 revolution when nearly all German states signicantlyincreased popular participation in government again in the faceof revolutionary pressures (see Blackbourn [1998] Chapter 3)The effects of this democratization were strongly mitigated byinstitutional restrictions however This regime featured a three-class voting system and was controlled initially by Junker land-lords and after the 1870s by the coalition of lsquolsquoiron and ryersquorsquo theparliament could not appoint ministers or discuss foreign policyand voting was oral Although after 1870 all adult males over theage of 25 had the right to vote voting was controlled in rural areasby the landlords (see Gosnell [1930] and Goldstein [1983]) AsAbrams [1995 p 10] puts it during this period lsquolsquothe GermanEmpire was in theory a constitutional monarchy yet in practiceit was governed by a Prussian oligarchyrsquorsquo The nal emergence ofGerman democracy the Weimar Republic in 1919 was in re-sponse to the very severe threat of social disorder and revolutiontriggered by the collapse of the German armies on the Western


Front in August 1918 (see for example the classic account inGerschenkron [1943] and also Mommsen [1981])

It is interesting to note at this point that although democrati-zation in Germany did not occur during the nineteenth centurysocial unrest was certainly as strong there as it was in Britain andFrance While there were no strong socialist parties in Britain andFrance and trade unions were of little importance the SocialDemocratic Party in Germany was by far the largest left-wingparty in Europe at that time and labor movementwas strong20 Atrst sight one might expect franchise extension in Germanyrather than in Britain and France Our model in contrastpredicts that the German elite should have had more exibility indealing with social unrest by promising future redistributionwhich was the pattern in practice21

In Sweden democracy arrived via a series of gradual fran-chise extensions starting in 1866 with the creation of a bicameralparliament with First and Second Chambers Universal malesuffrage was introduced in 1909 in the First Chamber but trueparliamentary government arrived only in 1918 when the politi-cal power of the Conservative Party and the monarchy werelimited once again an outcome of unusual turbulence spurred bythe end of the First World War and by the severe economic crisis(see Rustow [1955] and Verney [1957]) Tilton [1974 p 567]argues that lsquolsquoneither [of the rst two reform acts] passed withoutstrong popular pressure in 1866 crowds thronged around thechamber while the nal vote was taken and the 1909 reform wasstimulated by a broad suffrage movement [and] a demonstrationstrikersquorsquo The reform in 1909 had been preceded by strikes anddemonstrations and even though Sweden was not a participant inthe First World War the revolution in Russia and the situation inGermany forced the concession of democratic rights In 1917 theLiberals and Social Democrats formed a coalition government andproposed full male suffrage but this was defeated by the Conser-vative-dominated Second Chamber Collier [1999 Chapter 3 p 9]explains that lsquolsquoit was only after the economic crisis of 1918 and

20 See the essays in Katznelson and Zolberg [1986] on the relative strengthof the working classes in Germany in the nineteenth century For example Nolanbegins her chapter [p 354] by stating lsquolsquoAlthough Britain experienced the rstindustrial revolution and France developed the rst signicant socialist associa-tions Germany produced the largest and best-organized workersrsquo movement in thelate nineteenth centuryrsquorsquo

21 Williamson[1998] for example writes that lsquolsquothe main aim of [Bismarckrsquos]welfare program was to avoid revolution through timely social reform and toreconcile the working classes to the authority of the statersquorsquo


ensuing worker protests for democracy led by the Social Demo-crats that the Reform Act was passed Indeed in November 1918labor protests reached such a point as to be perceived as arevolutionary threat by Swedenrsquos Conservative party and upperclassesrsquorsquo Tilton [1974 p 568] summarizes the consensus view ofhistorians succinctly lsquolsquoSwedish democracy had triumphed with-out a revolutionmdashbut not without the threat of a revolutionrsquorsquo(italics in original)

B Alternative Theories of Democratization

We have documented that the threat of revolution oftenspurred by wars and depressions was important in leading up tothe political reforms of the period Although this account supportsour main thesis it is useful to review alternative theories and toinvestigate whether they receive support from the events surround-ing the reform process

A1 The Enlightenment This theory is that the elite extendedthe franchise because their social values changed In particularthe Enlightenment movement may have made it harder for theelite to view a society in which a large fraction of the populationhad no representation as fair and just

It is undoubtedly true that changes in social values have areal effect on political reform For example the enfranchisementof women in Britain in 1919 and 1928 appears mostly due tochanges in the societyrsquos approach to women22 Although thesuffragette movement in Britain organized demonstrations andsocial unrest23 these did not constitute a serious threat ofrevolution This is not inconsistent with our approach Sinceextending voting rights to women does not have major conse-quences for redistribution from the rich to the poor social valuesrather than redistributive motives should be more important

In contrast to the enfranchisement of women it is hard toargue that changing social values were the major driving forcebehind the extensions of the franchise to the poorer segments ofthe society The main argument against the importance of social

22 In France women were allowed to vote in 1945 while in Germany andSweden mass enfranchisement included women from the beginning

23 These included destroying letter boxes cutting telegraph lines andbombing railway stations and even Lloyd Georgersquos house Emmeline Pankhurst aleader of the suffragette movement understood the importance of social unrest inforcing reform and argued lsquolsquoParliament never grants reform unless it is terrorised Terrorism is in fact the only argument that parliament understandsrsquorsquo (quotedin Jorgensen-Earp [1997 p 48])


values is the timing of democratization Since the ideas of theEnlightenment and the writings of Rousseau and Paine werewidely disseminated in Europe by the early decades of thenineteenth century it is hard to understand the timing ofdemocratization and especially why these ideas inuenced thepolitical elite in Sweden and Germany long after they hadpersuaded the elite in France and Britain Rather the timepattern of democratization is tied much more closely to that ofindustrialization inequality and political unrest Perhaps moreimportant the evidence discussed in this and the previous sectionsuggests strongly that franchise extension was forced on the eliteand not willingly given as this theory would suggest Overall wetherefore conclude that changes in social values are unlikely tohave been the major factor behind the extension of the franchiseto the poorer segments of the society

A2 Political Party Competition This theory is inspired inpart by the British experience where the competition betweenDisraeli and Gladstone was a major factor in political reformAccording to this theory politicians have a strong preference tostay in power and may extend the franchise with the expectationthat the newly enfranchised will return the favor by voting fortheir party (see for example Himmelfarb [1966] who argues thisfor the British case and the general discussion in Collier [1999])

In 1866 Russellrsquos Liberal government proposed a relaxation ofthe property restrictions on voting This measure was defeated bya coalition of Conservatives led by Disraeli and right-wing Liber-als the lsquolsquoAdullamitesrsquorsquo who thought the extension too generousThe Liberal government then collapsed and Disraeli formed aminority administration (290 to 360) Disraelirsquos rst move was tointroduce a less generous franchise extension but he realized thatthis would not gain majority support He then switched to aproposal even more radical than the initial Liberal measurewhich passed by gaining the support of a heterogeneous group ofLiberals Disraeli can then be argued to have extended thefranchise that he rst opposed to encourage the newly enfran-chised to vote Conservative

In our view however the evidence does not support thisinterpretation The Conservatives lost the 1868 election immedi-ately after having passed the franchise extension (and the Liberalparty lost the election of 1885) So if the strategy was aimed atwinning elections it was clearly a failure Although the fact thatthe Conservatives lost the election does not prove that franchise


extension was not aimed at winning elections other aspects ofthis reform also appear inconsistent with a strategy of maximiz-ing Conservative votes In particular as the result of the split overthe Corn Laws support for the Conservative party was essentiallyconcentrated in rural areas with Tory landowners exertingsubstantial control over the electorate in the absence of a secretballot The reform measure passed under Disraeli increased thevoting population by only 45 percent in counties compared with145 percent in the boroughs effectively ensuring a Conservativedefeat in the following elections24

Overall the most plausible interpretation of the interpartyrivalry in Britain during the 1860s and 1870s was that while bothparties regarded the extension of voting rights as inevitable due tomounting social pressure they clearly saw that it could bestructured in ways which were more or less advantageous tothemselves This created a complicated lsquolsquoendgamersquorsquoCowling [1967p 89] argues that the Conservative party supported Disraeli in1867 because if the Act failed lsquolsquothe Liberals might then doprecisely what Derby and Disraeli had striven in 1866 to preventtheir doingmdashcarry Reform on their own linesrsquorsquo The one triumph ofthe 1867 reform for Disraeli was the fact that it limited theredistribution of seats away from the counties to the boroughswhich would have been even more substantial otherwise Thisstrategy reduced the impact of the franchise extension for theConservative party and its constituency Smith [1967 p 97] alsoagrees and argues that lsquolsquoDerby and Disraeli in 1867 did notdetermine to trust the people or put their faith in a Conservativedemocracy They did what they felt they had to do to satisfy thepopular agitation and reconcile the upper strata of the workingclasses to the established political systemrsquorsquo

The party competition theory does not t well with any of ourother cases With army units in revolt and the economy collapsingin Germany in 1918ndash1919 the former political elite attempted toprevent revolution by generating a transition that would causeminimal damage to it interests In France there were moredistinct subsets within the elite Orleanists and Legitimistsformed separate factions within the Monarchist camp and the

24 The notion that the Liberals and Conservatives were prepared to extendthe franchise simply to keep their party in power is also not completely persuasiveBetween 1859 and 1865 the Liberal prime minister Palmerston who was opposedto franchise extension and the Conservative leader Lord Derby colluded so thatthe issue of suffrage would never be raised in parliament [Lee 1994 p 138]


Republicans though democratic were basically middle class andwere not in favor of universal male suffrage in 1848 When theMonarchy collapsed in 1848 these groups had to agree to thedemands of the revolutionaries In support of this view Aminzade[1993 p 35] argues lsquolsquoFrench workers mainly artisans consti-tuted the revolutionary force that put the Republican party inpower in February 1848 and working class pressure from thestreets of Paris forced liberal Republican leaders to reluctantlyconcede universal male suffragersquorsquo The same is true for the periodafter 1870 The conict at the time particularly the Communeforced democracy along the lines of 1849 Although no groupwithin the elite was committed to universal male suffrage theywere forced to reintroduce it

The Swedish case is perhaps the most similar to Britain In1906 the Liberal partyrsquos rst ever government fell after failing topass a law introducing universal male suffrage The reformmeasure of 1909 was then passed (in 1907) by the Conservativegovernment under Lindman As with Disraeli in 1867 lsquolsquoLindmanand his Conservative ministry that took office a year after theLiberalsrsquo 1906 failure saw an opportunity to pass a political reformon its own termsrsquorsquo [Collier 1999 Chapter 3 p 9] Although malesuffrage was conceded in one house the Conservatives keptcontrol over the other through the maintenance of multiple votingand taxpayer suffrage As with the British case this pattern ofevents was not the result of attempts by the Conservatives to gainvotes but rather a damage limitation exercise in the face ofmounting social pressure for a full democracy

A3 Middle Class Drive This story is similar to the previousone except that economic incentives rather than political compe-tition are the driving force The main point can be illustrated witha very simple model related to our analysis of Section II Supposethat there are three groups lower middle and upper class withrespective fractions l L l M and l U and human capital hL hM andhU Let h 5 l L hL 1 l M hM 1 l U hU be the average human capitalSuppose that when the lower classes are excluded from thepolitical process the middle class are in power with probability y and when the political process includes the lower classes thelower classes are in power with probability y L and the middleclasses are in power with probability y M As before only lineartaxes and universal subsidies are allowed and the group in powercan also decide whether to extend the franchise (if it was notextended before) It is now straightforward to see that if h hM


the middle-classes prefer taxation to no taxation and also realizethat this is also what lower classes want So if y L 1 y M y that isif franchise extension increases the likelihood of taxation themiddle classes will extend the franchise when they control powerduring the prefranchise era In other words in this theory themiddle classes extend the franchise because they hold power onlytemporarily and realize that by including the lower classes in thepolitical processes they will shift the balance of power in theirfavor

There seems little evidence in Britain that the middle classesother than a few radical MPs such as Bright elected in the largeindustrial cities wanted to allow the working classes to vote Itwas well understood that this would lead to redistribution at theirexpense As Lord Elcho a leading Adullamite put it democracymeant lsquolsquohanding the country over to the Trade Unions and the ruleof numbers enabling the poor to tax the richrsquorsquo (quoted in Cowling[1967] p 51) Cowling further notes [p 54] lsquolsquoDisraeli andGladstone were attempting to push members of parliament intodoing what they had no desire to do far from wishing to extendthe franchise [they were] intensely suspicious of any attempt to dosorsquorsquo (italics in original) Viscount Cranborne a leading Conserva-tive saw the reform struggle as we do as lsquolsquoa battle not of partiesbut of classesrsquorsquo and lsquolsquoa portion of the great political struggle of ourcenturymdashthe struggle between property and mere numbersrsquorsquo(quoted in Smith [1967] pp 27ndash28)

This theory does not seem to explain the other instances ofdemocratization either In Germany in 1918ndash1919 the middleclasses were either part of the coalition supporting the Monarchyor moderates within the Social Democratic party Even the SocialDemocrats saw suffrage extension as likely to help the moreleft-wing parties like the Spartacists In France the middle classcould best be associated with the Republican party which opposeduniversal male suffrage In Sweden the Liberal party partiallyrepresented the middle classes and entered into a tacticalcoalition with the Social Democrats to force full democracy on theintransigent Conservatives and the Monarchy Nevertheless theLiberals in Sweden were very different from the Liberals inBritain and included more left-wing groups and workers (seeVerney [1957] p 138) The struggle for democracy in Swedenshould therefore be seen as a battle between the Conservativesand two left-wing parties where the victory of the working-class


parties was secured by the large social unrest following WorldWar I

C Democratization and Redistribution

Our approach also predicts that after franchise extensionthere should be increased redistribution We now briey discussthe evidence related to the relationship between democratizationand redistribution in the four cases of franchise extension dis-cussed above

In Britain the Reform Acts of 1867ndash1884 were a turning pointin the history of the British state In 1871 Gladstone reformed thecivil service opening it to public examination thus making itmeritocratic Liberal and Conservative governments introduced aconsiderable amount of labor market legislation fundamentallychanging the nature of industrial relations in favor of workersDuring 1906ndash1914 the Liberal Party under the leadership ofAsquith and Lloyd George introduced the modern redistributivestate into Britain including health and unemployment insurancegovernment-nanced pensions minimum wages and a commit-ment to redistributive taxation As a result of the scal changestaxes as a proportion of National Product more than doubled inthe 30 years following 1870 and then doubled again In themeantime the progressivity of the tax system also increased[Lindert 1989]

Meanwhile the education system which was either primar-ily for the elite or run by religious denominations during most ofthe nineteenth century was opened up to the masses the Educa-tion Act of 1870 committed the government to the systematicprovision of universal education for the rst time and this wasmade free in 1891 The school-leaving age was set at eleven in1893 then in 1899 it increased to twelve and special provisionsfor the children of needy families were introduced [Mitch 1993]As a result of these changes the proportion of ten-year oldsenrolled in school that stood at a disappointing 40 percent in 1870increased to 100 percent in 1900 [Ringer 1979 p 207] Finally thereform act of 1902 led to a large expansion in the resources forschools and introduced the grammar schools which subsequentlybecame the foundation of secondary education in Britain

In France as in Britain democratization coincided withimportant educational reforms During the Second Empire therewas a signicant expansion of government support for educationilliteracy fell from 39 percent to 29 percent of adults and the


primary school enrollment rate increased from 51 percent to 68percent [Plessis 1985 Table 14 p 100] In 1881 the governmentabolished fees in public primary schools and in 1882 it introducedseven years of compulsory education for children The primaryenrollment rate increased from 66 percent in 1863 to 82 percent in1886 The lsquolsquoliberalrsquorsquo phase of the Second Empire saw signicantlabor market legislation with strikes legalized in 1863 and unionswere nally officially tolerated in 1868 Moreover central govern-ment expenditure as a percentage of GDP increased by one-thirdfrom 94 percent in 1872 (a gure itself inated by the war) to 124percent in 1880 [Flora 1983]

In Germany the large increase in redistribution in the 1920swas initiated by the Weimar state [Flora 1983] Also in Swedenmajor redistribution appears to have started only after democrati-zation Lindertrsquos [1994] data show that before 1920 there was noredistribution at all in Sweden while after this date it jumped upsharply

Overall we can summarize our discussion especially the partabout the relationship between democratization and educationalreforms by quoting Easterlin [1981] lsquolsquoto judge from the historicalexperience of the worldrsquos 25 largest nations the establishmentand expansion of formal schooling has depended in large part onpolitical conditions and ideological inuencesrsquorsquo and lsquolsquoa majorcommitment to mass education is frequently symptomatic of amajor shift in political power and associated ideology in a direc-tion conducive to greater upward mobility for a wider segment ofthe populationrsquorsquo25

D The Kuznets Curve

Our theory also suggests that the Kuznets curve for incomeinequality should be caused by franchise extension so we mayexpect the peak of the Kuznets curve to coincide with majordemocratic reforms To get an idea of whether this might be a validexplanation for the Kuznets curve we looked at the behavior ofincome inequality for Britain France Germany and Sweden

Data on income inequality for the nineteenth century are notextremely reliable Nevertheless a number of studies usingdifferent data sources on Britain reach the same conclusioninequality increased substantially during the rst half of the

25 See Engerman Mariscal and Sokoloff [1998] for historical evidence fromthe Americas that democratization tends to induce educational expansion


nineteenth century then started falling in the second half Theturning point appears to be sometime after 1870 [Williamson1985 Table 42] This picture is also consistent with the ndings ofCrafts [1989] and of Lindert [1986] on wealth inequality but isnot completely uncontroversial [Feinstein 1988] A similar pat-tern also emerges from earnings inequality data reported inWilliamson [1985 Table 42] where the Gini coefficient increasesfrom 04 in 1823 to 0627 in 1871 and falls to 055 in 1881 and0443 in 1901 It appears therefore that inequality peaked approxi-mately at the time of the major political reforms and fell sharplyafter the extension of the franchise

Data for other countries are even scarcer Morrisson [1999]surveys the existing evidence and argues that Germany Franceand Sweden all went through a Kuznets curve In Germanyinequality rose during the nineteenth century and most research-ers place the peak around 1900 For example Kuznets [1963]nds that the income share of the top 5 percent went from 28percent in 1873ndash1880 to 32 percent in 1891ndash1900 stayed at 32percent during 1901ndash1910 declining to 31 percent in 1911ndash1913Dumke [1991] nds the same income share to be 284 percent in1880 rising to 326 percent in 1900 and falling to 306 percent in1913 During the Weimar Republic inequality fell rapidly Kraus[1981] records that by 1926 the income share of the top 5 percenthad fallen by 62 percent Overall Morrisson [1997] argues thatthe Kuznets curve in Germany peaked in 1900 went at andstarted to fall in the 1920s This date corresponds closely to themajor democratization of 1918ndash1919

For France Morrisson [1991 1997] argues that inequalityrose until 1870 with the income share of the top 10 percentpeaking at around 50 percent Inequality started to fall howeverin the 1870s and in 1890 the income share of the top 10 percentwas down to 45 percent falling further to 36 percent by 1929 Themajor political reforms of 1860ndash1877 in France are thereforeapproximately around the peak of the Kuznets curve FinallySoderberg [1987 1991] records that income inequality grew inSweden peaking just before the First World War leveling off orfalling slightly during the 1920s and then falling rapidly thereaf-ter Once again there is a close correspondence between thedecline in inequality and the extension of the franchise Overalltherefore in Britain France Germany and Sweden the peak ofthe Kuznets curve appears to have followed democratizationwhich is in line with the mechanism proposed in this paper


V CONCLUDING REMARKS

This paper has offered a simple model of democratization andpolitical reform and investigated the implications for the dynam-ics of growth and inequality The two main contributions of thispaper are (1) it explains why the rich elite may want to extend thefranchise even though this implies higher taxation in the futureand (2) it offers a new explanation for the presence of a Kuznetscurve in the development experience of Western societies

Our emphasis on political reform as a way of changing futurepolitical equilibria may have a number of other applicationsOther reforms as well as constitutional restrictions also affectthe future balance of political power It appears for example thatin the Federalist papers Madison and Hamilton introduced theseparation of powers as a way of restricting future policies (in factMadison was very worried about the security of property rightsunder the Articles of Confederation see Madison [1961]) Simi-larly relations with international institutions can act as commit-ment devices to future policies Also redistributive programsoften differ in how easily they can be reversed In most countriesfor example social security entitlement programs appear to bemore difficult to cut than other redistributive programs Thisraises the question of whether there was a commitment motive inplay when these programs were instituted

Finally as already noted there are also major differences inthe form of redistribution across countries In Britain educationincreased substantially after the franchise due to increasedgovernment support In contrast in Germany early redistributionwas via the welfare state It is important to understand whatmight cause these differences and whether the same forces arealso important in shaping the differences in the extent and form ofredistribution we observe today

APPENDIX NON-MARKOVIAN EQUILIBRIA

IN THE MODEL OF SECTION II

We now analyze the model of Section II without the restric-tion to Markovian strategies More specically we look forsubgame perfect equilibria We will nd that there exists a cutoffprobability of state microh q q such that when q q there willbe redistribution without democratization which prevents a revo-lution In contrast when q q the only equilibrium will featurethe extension of the franchise when microt 5 microh


First note that if the poor initiate a revolution at time t theneffectively the game ends with V p(R) 5 microtAH l (1 2 b ) Thereforein any subgame perfect equilibrium s t

p(microhE ) 5 (r 5 1) only ifV p(R) V t

p where V tp is the payoff of the poor in the continuation

game at time t without a revolution26 This immediately impliesthat s t

p (microlE ) 5 ( r 5 0) Next note that after f t 5 1 andignoring revolution the elite are down to their minimum payoffsince t 5 tˆ in all future periods Therefore s t

p (microhE f 5 1 ) 5( r 5 1) only if V p(R) V p(D) Assumption 2 ensures that thisinequality never holds Thus in any subgame perfect equilibriums t

p(microhE f 5 1 ) 5 ( r 5 0) irrespective of the history of the gameup to this point So we have pinned down all of the strategies bythe poor other than s t

p(microhE f 5 0t r) Now consider thisLet V t

p (microhE f 5 0 t r) be the continuation payoff of the poorconditional on f 5 0 and t r when they play r t 5 0 Then in anysubgame perfect equilibrium s t

p(microhE f 5 0 t r) will only putpositive probability on r 5 1 if V p(R) $ V t

p (microhE f 5 0 t r) and willplay r 5 1 with probability 1 if V p(R) V t

p (microhE f 5 0 t r)Suppose that q q Recall from Proposition 2 that in this

case there were no Markov Perfect Equilibria with redistributionand no democratization Let t r(microt) be the tax rate chosen by theelite in state microt at time t Consider the following candidateequilibrium strategy combination For the elite s t

r(microhE) 5( f t 5 0 t t

r 5 tˆ) and s tr(microlE) 5 ( f t 5 0t t

r 5 t ) where t tˆ For thepoor s t

p(micro jE f t 5 0 t tr) 5 ( r 5 0) if t r(micros) $ tˆ if micros 5 microh and t r(micros) $

t if micros 5 micro l for all s t and ( r 5 1) otherwise Then the payoffs inthis candidate equilibrium are given by

(7) V j(microhE) 5 (1 2 tˆ)Ahj 1 tÂH 1 b [qV j(microhE) 1 (1 2 q)Vj(microlE)]

(8) V j(microlE) 5 (1 2 tˆ)Ahj 1 tÂH 1 b [qV j(microhE) 1 (1 2 q)V j(microlE)]

for j 5 p and r Now dene t such that V p (microhE) 5 V p(R) t têxists by Assumption 2 Therefore the above strategies are thebest response for the poor in all subgames Next we need to checkwhether they are the best-response for the elite Clearly if theelite reduce the tax rate in state (microhE) this will immediatelycause a revolution Thus s t

r(microhE) 5 ( f t 5 0t tr 5 tˆ) is optimal

conditional on the history up to time t characterized by t r(micros) $ tˆ ifmicros 5 microh and t r(micros) $ t if micros 5 micro l for all s t In contrast if the elite

26 We are now using s t instead of s which stands for s conditional on thepublic history of the game up to time t The public history includes all past actions(but not mixing probabilities when these are used)


deviate from s tr(microlE) 5 ( f t 5 0t t

r 5 tˆ) to s tr(micro lE) 5 ( f t 5 0

t tr 5 0) this will not cause a revolution immediately It will only do

so when the state changes to (microhE) But in this case the elite canplay s t

r(microhE) 5 ( f t 5 1) and as we saw above the best-response ofthe masses is always s t

p(microhE f 5 1 ) 5 ( r 5 0) irrespective of thehistory of the game up to this point The payoff to the elite fromfollowing this deviant strategy starting in the state (micro lE) is

Vdr (micro lE) 5 Ahr 1 b [qV r(D) 1 (1 2 q)Vd

r (micro lE)]

Therefore the above candidate equilibrium strategy combinationis a subgame perfect equilibrium if and only if V r(micro lE) given by(7) and (8) is greater than or equal to

V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)

It is straightforward that if q 5 q V r (micro lE) V dr (microlE) and at

q 5 0 V r (microlE) V dr (micro lE) Also V r (micro lE) falls faster in q than

V dr (microlE) So there exists q such that for all q q Vr (micro lE)

V dr (microlE) and there exists no equilibrium with redistribution and

democratizationFinally when q q the Markov Perfect Equilibrium with

redistribution and no democratization continues to be a subgameperfect equilibrium and with reasoning similar to the above wecan construct others that feature some redistribution both in statemicrot 5 microh and microt 5 micro l but all these equilibria have the samestructure of keeping the poor just indifferent between revolutionand no revolution in the state (microhE)mdashthus giving the samepayoffs to the elite and the poor

DEPARTMENT OF ECONOMICS MASSACHUSETTS INSTITUTE OF TECHNOLOGY

DEPARTMENT OF POLITICAL SCIENCE UNIVERSITY OF CALIFORNIA BERKELEY

REFERENCES

Abrams Lynn Bismarck and the German Empire 1871ndash1918 (New York Rout-ledge 1995)

Acemoglu Daron and James A Robinson lsquolsquoWhy Did the West Extend theFranchise Democracy Inequality and Growth in Historical PerspectiversquorsquoCEPR Discussion Paper No 1797 1997

Acemoglu Daron and James A Robinson lsquolsquoA Theory of Political TransitionsrsquorsquoMassachusetts Institute of Technology Working Paper 1999 American Eco-nomic Review forthcoming

Ades Alberto and Thierry Verdier lsquolsquoThe Rise and Fall of Elites EconomicDevelopment and Social Polarization in Rent-Seeking Societiesrsquorsquo unpublishedpaper DELTA 1993

Aghion Philippe and Patrick Bolton lsquolsquoA Trickle-Down Theory of Growth andDevelopmentrsquorsquo Review of Economic Studies LXIV (1997) 151ndash172


Alesina Alberto and Roberto Perotti lsquolsquoIncome Distribution Political Instabilityand Investmentrsquorsquo European Economic Review XLIX (1996) 1203ndash1225

Alesina Alberto and Dani Rodrik lsquolsquoDistributive Politics and Economic GrowthrsquorsquoQuarterly Journal of Economics CIX (1994) 465ndash490

Aminzade Ronald Ballots and Barricades Class Formation and RepublicanPolitics in France 1830ndash1871 (Princeton NJ Princeton University Press1993)

Benabou Roland lsquolsquoUnequal Societiesrsquorsquo American Economic Review XC (2000)96ndash129

Blackbourn David The Long Nineteenth Century A History of Germany 1780ndash1918 (New York Oxford University Press 1998)

Briggs Asa The Age of Improvement (London UK Penguin 1959)Campos J Edgar and Hilton L Root The Key to the Asian Miracle Making

Shared Growth Credible (Washington DC Brookings Institution 1996)Cole Alistair and Peter Campbell French Electoral Systems and Elections since

1789 (Aldershot Gower Press 1989)Collier Ruth B Paths Towards Democracy The Working Class and Elites in

Western Europe and South America (New York Cambridge University Press1999)

Cowling Maurice 1867 Disraeli Gladstone and Revolution (Cambridge UKCambridge University Press 1967)

Crafts Nicholas F R lsquolsquoReal Wages Inequality and Economic Growth in Britain1750ndash1850 A Review of Recent Researchrsquorsquo in P Scholliers ed Real Wages inNineteenth and Twentieth Century Europe (Oxford UK Berg 1989)

Darvall Frank O Popular Disturbances and Public Order in Regency England(London UK Macmillan 1934)

DiPasquale Denise and Edward Glaeser lsquolsquoThe Los Angeles Riot and the Econom-ics of Urban Unrestrsquorsquo Journal of Urban Economics XLIII (1998) 52ndash78

Dumke Rolf lsquolsquoIncome Inequality and Industrializationin Germany 1850ndash1913rsquorsquo inY Brenner et al eds Income Distribution in Historical Perspective (Cam-bridge UK Cambridge University Press 1991)

Easterlin Richard A lsquolsquoWhy Isnrsquot the Whole World Developedrsquorsquo Journal ofEconomic History XLI (1981) 1ndash19

Engerman Stanley L Elisa Mariscal and Kenneth L Sokoloff lsquolsquoSchoolingSuffrage and the Persistence of Inequality in the Americas 1800ndash1945rsquorsquounpublished working paper University of California at Los Angeles 1998

Evans Eric J The Forging of the Modern State Early Industrial Britain1783ndash1870 (New York Longman 1983)

Feinstein Charles H lsquolsquoThe Rise and Fall of the Williamson Curversquorsquo Journal ofEconomic History XLVIII (1988) 699ndash729

Flora Peter State Economy and Society 1815ndash1975 (Frankfurt Germany Cam-pus Verlag 1983)

Galor Oded and Joseph Zeira lsquolsquoIncome Distribution and MacroeconomicsrsquorsquoReview of Economic Studies LX (1993) 35ndash52

Gerschenkron Alexander Bread and Democracy in Germany (Berkeley Univer-sity of California Press 1943)

Goldstein Richard J Political Repression in Nineteenth Century Europe (LondonUK Croom Helm 1983)

Gosnell Harold F Why Europe Votes (Chicago IL University of Chicago Press1930)

Grossman Herschel I lsquolsquoA General Equilibrium Theory of Insurrectionsrsquorsquo Ameri-can Economic Review LXXXI (1991) 912ndash921 lsquolsquoRobin Hood and the Redistribution of Property Incomersquorsquo European Journalof Political Economy XI (1995) 399ndash410

Haggard Stephan Pathways from the Periphery (Ithaca NY Cornell UniversityPress 1990)

Harrison Royden Before the Socialists Studies in Labour and Politics 1861ndash1881(London UK Routledge amp Kegan Paul 1965)

Hayes WilliamA The Background and Passage of the Third Reform Act (OxfordUK Oxford University Press 1982)

Himmelfarb Gertrude lsquolsquoThe Politics of Democracy The English Reform Act of1867rsquorsquo Journal of British Studies VI (1966) 97ndash138


Ho Samuel Economic Development of Taiwan 1860ndash1970 (New Haven CT YaleUniversity Press 1978)

Jorgensen-Earp Cheryl R The Transguring Sword The Just War of the WomenrsquosSocial and Political Union (Tuscaloosa University of Alabama Press 1997)

Katznelson Ira and Aristide R Zolberg Working-Class Formation NineteenthCentury Patterns in Western Europe and the United States (Princeton NJPrinceton University Press 1986)

Kraus Franz lsquolsquoThe Historical Development in Income Inequality in WesternEurope and the United Statesrsquorsquo in Peter Flora and Arnold J Heidenheimereds The Development of the Welfare State in Europe and America (NewBrunswick NJ Transaction Books 1981)

Kuznets Simon lsquolsquoEconomic Growth and Income Inequalityrsquorsquo American EconomicReview XLV (1955) 1ndash28 lsquolsquoDistribution of Income by Sizersquorsquo Economic Development and CulturalChange XI (1963) 1ndash80

Lee Stephen J Aspects of British Political History 1815ndash1914 (London UKRoutledge 1994)

Lindert Peter H lsquolsquoUnequal English Wealth since 1670rsquorsquo Journal of PoliticalEconomy XCIV (1986) 1127ndash1162 lsquolsquoModern Fiscal Redistribution A Preliminary Essayrsquorsquo University of Califor-nia at Davis Agricultural History Center Working Paper No 55 1989 lsquolsquoThe Rise in Social Spending 1880ndash1930rsquorsquo Explorations in Economic HistoryXXXI (1994) 1ndash37

Lipset Seymour M Political Man The Social Bases of Politics (Garden City NJAnchor Books 1960)

Madison James (with Alexander Hamilton and John Jay) The Federalist Papers(New York New American Library 1961)

Meltzer Allan H and Scott F Richard lsquolsquoA Rational Theory of the Size ofGovernmentrsquorsquo Journal of Political Economy LXXXIX (1981) 914ndash927

Mitch David lsquolsquoThe Role of Human Capital in the First Industrial Revolutionrsquorsquo inJoel Mokyr ed The British Industrial Revolution An Economic Perspective(San Francisco CA Westview Press 1983)

Mommsen Wolfgang J lsquolsquoThe German Revolution 1918ndash1920 Political Revolutionand Social Protestrsquorsquo in R Bessel and E J Feuchtwanger eds Social Changeand Political Development in Weimar Germany (London UK Croon Helm1981)

Moore Barrington The Social Origins of Dictatorship and Democracy (BostonMA Beacon Press 1966)

Morrisson Christian lsquolsquoLa Repartition des Revenues en France depuis 1880rsquorsquo inJ-C Casanova and M Levy-Leboyer eds Histoire economique de la France(Paris Gallimard 1991)

Morrisson Christian lsquolsquoHistorical Evolution of Income Distribution in WesternEuropersquorsquo chapter in Anthony B Atkinson and Francois Bourguignon edsHandbook of Income Distribution (Amsterdam North-Holland 1999)

Muller Edwin N and Mitchell A Seligson lsquolsquoInequality and InsurrectionsrsquorsquoAmerican Political Science Review LXXXI (1987) 425ndash451

North Douglass C and Barry R Weingast lsquolsquoConstitutions and CommitmentEvolution of the Institutions Governing Public Choice in 17th CenturyEnglandrsquorsquo Journal of Economic History XLIX (1989) 803ndash832

Persson Torsten and Guido Tabellini lsquolsquoIs Inequality Harmful for Growth Theoryand Evidencersquorsquo American Economic Review LXXXIV (1994) 600ndash621

Plessis Alain The Rise and Fall of the Second Empire 1852ndash1871 (CambridgeUK Cambridge University Press 1985)

Price Roger Napoleon III and the Second Empire (New York Routledge 1997)Ringer Fritz Education and Society in Modern Europe (Bloomington Indiana

University Press 1979)Roemer John E lsquolsquoRationalizing Revolutionary Ideology A Tale of Lenin and the

Tsarrsquorsquo Econometrica LIII (1985) 85ndash108Rueschemeyer Dietrich Evelyn H Stephens and John D Stephens Capitalist

Development and Democracy (Chicago IL University of Chicago Press 1992)Rustow Dankworth A The Politics of Compromise A Study of Parties and Cabinet

Government in Sweden (New York Greenwood Press 1955)


Searle Geoffrey R Entrepreneurial Politics in Mid-Victorian Britain (Oxford UKOxford University Press 1993)

Smith Paul Disraelian Conservatism and Social Reform (New York Routledge1967)

Soderberg Johan lsquolsquoTrends in Inequality in Sweden 1700ndash1914rsquorsquo Historical SocialResearch (1987) 58ndash78 lsquolsquoWage Differentials in Sweden 1725ndash1950rsquorsquo in Y Brenner et al eds IncomeDistribution in Historical Perspective (Cambridge UK Cambridge UniversityPress 1991)

Stevenson John Popular Disturbances in England 1700ndash1870 (New YorkLongman 1979)

Therborn Goran lsquolsquoThe Rule of Capital and the Rise of Democracyrsquorsquo New LeftReview CIII (1977) 3ndash41

Tilton Timothy A lsquolsquoThe Social Origins of Liberal Democracy The Swedish CasersquorsquoAmerican Political Science Review LXVIII (1974) 561ndash571

Trevelyan George M British History in the Nineteenth Century and After1782ndash1919 (London Longman Green and Co Publishers 1937)

Verney Douglas Parliamentary Reform in Sweden 1866ndash1921 (Oxford UKClarendon Press 1957)

Williamson D C Bismarck and Germany 1862ndash1890 2nd edition (New YorkLongman 1998)

Williamson Jeffrey G Did British Capitalism Breed Inequality (Boston Allenand Unwin 1985)

Wright D G Democracy and Reform 1815ndash1885 (London UK Longman 1970)Zeldin Theodore The Political System of Napoleon III (New York St Martinrsquos

Press 1958)


During the late nineteenth and early twentieth centuries thefranchise was also extended in most other Western societiesDemocratization was again followed by increased redistributionand in most cases by the downturn of the Kuznets curve

These events are hard to understand with our existingtheories If democratization is likely to lead to increased taxationand redistribution (eg Meltzer and Richard [1981]) why shouldthe elite extend the franchise Our answer is that the elite wereforced to extend the franchise because of the threat of revolutionWe argue that extending the franchise acted as a commitment tofuture redistribution and prevented social unrest In contrast todemocratization the promise by the elite to redistribute in thefuture while maintaining political power would not have beencredible

The second contribution of our paper is to point out thepotential link between democratization and the Kuznets curve1Rising inequality often associated with industrialization in-creases social unrest and induces democratization Democratiza-tion in turn opens the way for redistribution and mass educationand reduces inequality Interestingly in line with this approachin a number of Western economies the peak of the Kuznets curvecoincides with the extension of the franchise

The thesis that the elite extended the franchise in order toavoid a revolution or social unrest is at the heart of our paperAlthough we are aware of no other papers in the economicsliterature a nonformal literature in political science startingwith the seminal work of Lipset [1960] and Moore [1966] hasstudied the origins of democracy To the best of our knowledgehowever no previous work explained franchise extension as acommitment device to future redistribution or pointed out the linkbetween democratization and the Kuznets curve though Ther-born [1977] and Rueschemeyer Stephens and Stephens [1992]have stressed the role of social conict in democratization2 In

1 Other theories of the Kuznets curve include Kuznetsrsquo [1955] originalconjecture that the curve was generated by the dual economy dynamics generatedby the switch from the agricultural to the industrial sector Lindertrsquos [1986]hypothesis that the curve resulted from the falling importance of income from theland Williamsonrsquos [1985] argument that technological change increased wagesrelative to capital income and Aghion and Boltonrsquos [1997] model where accumula-tion by the rich eventually pushes down the interest rate sufficiently to allow thepoor to invest and catch up

2 In a different context North and Weingast [1989] have argued that theintroduction of the English Parliament in the seventeenth century was a commit-ment to low taxes in the future Our paper is also related to models of political












m 5 AH tm



BH th where H t

































are given by



ASSUMPTION 1

hr

hp

l (1 2 microh)

(1 2 l )(microh 2 (1 2 b )((A 2 B)A)












1 2 band V r(D) 5

Bhr 1 (A 2 B)H

1 2 b


ASSUMPTION 2




5Bhp 1 (A 2 B)H 2 b (1 2 q)(A 2 B)(H 2 hp)

1 2 b























i at t is

(4) h t1 1i 5 d et

ih t

i



i 1 etiZ yt




r 1 (A 2 B) ( l h0p 1 (1 2 l )h0









(5) Bh0p 1 (A 2 B)Ht $ Z





(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b









V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)












m 5 AH tm



BH th where H t

































are given by



ASSUMPTION 1

hr

hp

l (1 2 microh)

(1 2 l )(microh 2 (1 2 b )((A 2 B)A)












1 2 band V r(D) 5

Bhr 1 (A 2 B)H

1 2 b


ASSUMPTION 2




5Bhp 1 (A 2 B)H 2 b (1 2 q)(A 2 B)(H 2 hp)

1 2 b























i at t is

(4) h t1 1i 5 d et

ih t

i



i 1 etiZ yt




r 1 (A 2 B) ( l h0p 1 (1 2 l )h0









(5) Bh0p 1 (A 2 B)Ht $ Z





(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b









V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)

























are given by



ASSUMPTION 1

hr

hp

l (1 2 microh)

(1 2 l )(microh 2 (1 2 b )((A 2 B)A)












1 2 band V r(D) 5

Bhr 1 (A 2 B)H

1 2 b


ASSUMPTION 2




5Bhp 1 (A 2 B)H 2 b (1 2 q)(A 2 B)(H 2 hp)

1 2 b























i at t is

(4) h t1 1i 5 d et

ih t

i



i 1 etiZ yt




r 1 (A 2 B) ( l h0p 1 (1 2 l )h0









(5) Bh0p 1 (A 2 B)Ht $ Z





(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b









V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)





1 2 band V r(D) 5

Bhr 1 (A 2 B)H

1 2 b


ASSUMPTION 2




5Bhp 1 (A 2 B)H 2 b (1 2 q)(A 2 B)(H 2 hp)

1 2 b























i at t is

(4) h t1 1i 5 d et

ih t

i



i 1 etiZ yt




r 1 (A 2 B) ( l h0p 1 (1 2 l )h0









(5) Bh0p 1 (A 2 B)Ht $ Z





(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b









V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)






















i at t is

(4) h t1 1i 5 d et

ih t

i



i 1 etiZ yt




r 1 (A 2 B) ( l h0p 1 (1 2 l )h0









(5) Bh0p 1 (A 2 B)Ht $ Z





(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b









V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)





(5) Bh0p 1 (A 2 B)Ht $ Z





(6) V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b









V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)






V p(micro lE htph t

r) 5 Ah tp 2 et

pZ



r )]



p h tp Whether et




Vp(microhE h tpht

r) 5 Ahtp 2 Z 1 T p



r )]







phtr) increases


















V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)











V p(R h tpht

r) 5microhAHt

l (1 2 b d )2

Z

1 2 b


Vp(microhE h tph t

r) 5Ah t

p

1 2 b d2

Z

1 2 b1 Tp 1

q b Tp

1 2 b d





ph tr)
































































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)






























































D The Kuznets Curve

















































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)














































r (micro lE)]


V dr (micro lE) 5

Ahr 1 b qV r(D)

1 2 b (1 2 q)









REFERENCES








































































Press 1958)



































































Press 1958)


why did the west extend the franchise? democracy, inequality… · why did the west extend the...

Documents