Elite Representation and Local Conflict: Evidence from Ancien R´egimeFrance

Anne Degrave∗

April, 2021

Abstract

Representative institutions survive in authoritarian regimes to mitigate the conflict between state and elites. What is their impact on the conflict between local elites and non-elite populations? Local representative institutions may lead to policies that fit better local preferences than centralized gov- ernance. However, limits on central authority can make it easier for local elites to exert power at the expense of non-elite populations. I analyze the case of late Ancien R´egimeFrance, taking advantage of geographic heterogeneity in the presence of representative institutions consenting to taxation. Using data on 1789 grievances and popular rebellions during the 1700s, I show that consent led to a reduced tax burden for the provinces, but exacerbated the local elite-peasant conflict. I find no evidence of a positive effect on living standards or economic development.

1 Introduction

Accounts of representative institutions in authoritarian regimes usually focus on their impact on the conflict between state and elites. Here I analyze how they impact non-elite populations, through their effect on governance and local conflict. On the one side, local representative institutions may lead to policies that fit better local preferences than centralized governance. On the other side, limits on central authority can make it easier for local elites to exert power at the expense of non-elite populations, exacerbating local conflict. To shed light on these questions, I analyze the case of representative institutions consenting to taxation in late Ancien R´egimeFrance. Since late medieval times, several provinces of the French territory had the right to consent to the level of royal taxation and manage tax collection through institutions called the provincial estates (´etats provinciaux). Similar representative institutions arose throughout Europe during the early modern era and an extensive literature analyzes how these institutions impacted economic growth and democratization (North, 1993; Acemoglu, Johnson, and Robinson, 2005; Van Zanden, Buringh, and Bosker, 2012; Angelucci, Meraglia, and Voigtl¨ander, 2017). The trajectory of Ancien R´egimeFrance is typically considered as a failure of representative institutions, where democratization ultimately happened

∗PhD Candidate at the Wilf Family Department of Politics, New York University. Contact: [email protected]. I thank Ali T. Ahmed, Mariana Alvarado, Rafael Ch, Dongil Lee, Tiberiu Dragu, Dimitri Landa, Melissa M. Lee, Julia Payson, Pablo Querub´ın,Peter Rosendorff, Cyrus Samii, Melissa Schwartzberg, David Stasavage and Hye Young You, participants at the NYU Graduate Political Economy Seminar and the Politics & History conference for their helpful comments. I thank C´edric Chambru and Victor Gay for sharing the data on rebellions.

1 only through violent revolution in 1789. The provincial estates are thus dismissed as obsolete, or at best a missed opportunity for representation (Tocqueville, 1856). However, there is little direct evidence of the policy implications of consent to taxation. Given that popular grievances about taxation and elites played a major role in precipitating the revolution, understanding whether existing representation mitigated or exacerbated them is all the more important. The provincial estates represented elites of the Clergy (0.5% of the population), Nobility (1.5%), and Third Estate (the commoners, 98%). In 1789, they existed only in provinces called the pays d’´etat, representing about one third of the French territory. Provinces without provincial estates were taxed without consent, and were called pays d’´election (Figure1). The provincial estates were the broadest- based representative institutions under the absolute monarchy, but they were dominated by local noble, clergy or urban bourgeois elites, reflecting the unequal social structure of the Ancien R´egime.

Figure 1: Consent to taxation in French provinces (1789)

Pays d'élection: no consent Pays d'état: consent to taxation. Pays d'imposition: late acquisitions with no consent to taxation (excluded from analyses). Artois 9 10

Brittany

Burgundy

Provence 1 Languedoc 2 3 4 5 6 8

Notes: 1. Labourd 2. Basse-Navarre 3. Soule 4. B´earn5. Bigorre 6. Quatre-Vall´ees. 7. Couserans 8. Comt´ede Foix, 9. Cambr´esis,10. Hainaut.

The estates were created at the end of the Middle Ages at a time of central weakness: the monarchy granted the right to consent to taxation to some provinces as a concession to local elites in order to fund wars and to secure loyalty from recent territorial additions to the realm. Assuming that these strategic considerations are conditionally exogenous to popular grievances, we can analyze geographic variation in the presence of consent to taxation, using both OLS and a geographic regression discontinuity approach. I use data on social conflict from two sources. First, I use data on popular rebellions between 1661 and 1789 from Nicolas(2008). Second, I analyze social groups’ grievance lists drafted for the 1789 Estates- General, drawing from a content analysis by historians John Markoff and Gilbert Shapiro (Shapiro et al.,

2 1998).1 For both sources, I analyze two topics of grievances: taxation and local elites, particularly exploita- tion of peasants through the seigneurial system. I complement data on dissatisfaction with measures of objective conditions from various historical sources: soldiers’ heights during and after the Ancien R´egime, mortality, local taxation and urban growth. My first set of results show that the provincial estates decreased overall taxation, which translated into lower rates of grievances and rebellions about taxation. There is also suggestive evidence that rural populations benefited from it, in comparison to provinces with no consent. At the same time, elite grievances about taxation are similarly lower, indicating that any gains peasants might have were at the expense the central state rather than local elites. In a second set of analyses, I show that consent is not associated with higher welfare or economic growth in either the short or long run. Grievances and rebellions motivated by the seigneurial system are more frequent in consent provinces. For peasants, lower pressure from the central state was counterbalanced by increased pressure from local elites, exacerbating local conflict.

2 Literature and contribution

This article contributes to the literature on the impact of representative institutions in Europe. Schol- ars have argued that European early representative institutions paved the way for democratization and economic development by protecting property rights from extractive central monarchies (Bates and Lien, 1985; North, 1993; Acemoglu, Johnson, and Robinson, 2005; Van Zanden, Buringh, and Bosker, 2012; Cox, 2017; Angelucci, Meraglia, and Voigtl¨ander, 2017). This literature identifies a divergence from the early modern period between England and the Netherlands, where representation succeeded in effectively constraining central states at the national level, and the rest of Europe, where representative institutions became obsolete and dominated by absolutist regimes. Unfortunately, there are few empirical studies of representative institutions in absolutist regimes.2 Scholars have identified the state’s dependence on the nobility as a major factor of the divergence between France and England (Lachmann, 1989; Root, 1994; Hoffman and Norberg, 2002), but the consequences of the divergence itself are not quantified. This paper contributes to filling this gap by analyzing the impact of representative institutions on non-elite populations in late Ancien R´egimeFrance. In the case of France, the failure of consent to taxation is all the more salient since the 1789 revolution was in large part motivated by dissatisfaction with taxation (Shapiro et al., 1998). Substantively, my argument is similar to Stasavage(2014), who shows that medieval and early mod- ern autonomous cities initially promoted urban growth but later became captured by private interests. My article complements this work by highlighting similar extractive logics in larger scale representative institutions: the narrow elite that was represented in the provincial estates defended its interests at the expense of the majority of the population.

1The grievances of the Clergy were not coded by Markoff and Shapiro. 2Cross-country studies typically include only the national-level but less active Estates-General as the relevant represen- tative institution in early modern France (e.g. Acemoglu, Johnson, and Robinson(2005); Van Zanden, Buringh, and Bosker (2012); Cox(2017)). To my knowledge, the only reference to the French provincial estates in an empirical political economy paper is in Stasavage(2010). In a sociology thesis, Weitman(1968) compares pays d’´etatand pays d’´electionusing an earlier content analysis the 1789 nobility grievance lists. Møller(2017) analyzes the case of the Spanish cortes, but only at their beginnings.

3 Because the provincial estates consented to taxation and were controlled by elites, this article relates to the literature on the role of domestic elites in the development of fiscal capacity, which focuses on the state-elite conflict (Garfias, 2019; Mares and Queralt, 2015; Garfias, 2018; Boucoyannis, 2015). I contribute to these works by analyzing the impact of a state-elite fiscal institution on the elite-peasant conflict. More broadly, the question of the consequences of empowering local elites relates to longstanding debates in the political economy of development literature. A good example is the literature on traditional institutions in Africa. On the one side, scholars argue that these institutions are bad for state-building and democracy as they promote clientelism and possibly “capture” civil society organizations (e.g. Acemoglu, Reed, and Robinson(2014), de Kadt and Larreguy(2015)). On the other side, Baldwin(2015) emphasizes that traditional leaders may provide public goods and are informally accountable to their co-villagers in ways the central state is frequently not. Similar questions can be found in the literature on the pros and cons of indirect rule in colonial settings (Mamdani, 1996; Lechler and McNamee, 2018; Naseemullah and Staniland, 2016).

3 Analytical framework

We consider an authoritarian state, where there is no election. There are three relevant groups of actors: the state, (non-state) elites, and the non-elite populations (i.e. the majority). Both the state and elites are susceptible to engage in rent-seeking at the expense of non-elite populations. We want to compare the welfare of non-elite populations under two distinct institutional arrangements. In the first one, “no consent”, the state decides on all relevant policies. In the second one, “consent”, power over some policies is formally shared with elites through local representative institutions, which I will call “consent institutions”. Specifically, under the second arrangement, the central state cannot collect taxation without the consent of elites.

3.1 Direct effect: local informational advantage

We can distinguish direct and indirect effects of the consent institutions. The direct effect is that they are local institutions: provincial elites can be expected to have better information about local preferences than centrally appointed officials and are therefore able to manage taxation in a more effective way. The literature has shown that this informational advantage is the source of many early representative institutions (Angelucci, Meraglia, and Voigtl¨ander, 2017; Mayshar, Moav, and Neeman, 2017; Ahmed and Stasavage, 2020). This leads to the following hypothesis:

• Hypothesis 1: Taxation is better accepted by the population under consent institutions.

This hypothesis is not mechanistic. Some historians argue that provincial estates has lost all power by the 1700s and remained in place only for alternative services they provided to the monarchy, such as borrowing funds for the king (Potter and Rosenthal, 1997) or paying for military stations (Blaufarb, 2010). Local elites’ informational advantage could also have diminished as the central state build a centralized bureaucracy in no consent regions.

3.2 Indirect effect: more or less accountable?

The indirect effect of consent institutions is that they give power to local elites. In the absence of consent, elites might be powerful in other respects but they do not control taxation.

4 This effect is more ambiguous on the welfare of the non-elite population. On the one hand, local elites, even they are not representative of the commoners, may be more accountable than central officials, which would support the previous hypothesis and make citizens better-off. On the other hand, it may be easier for local elites to extract rents under consent. Brennan and Hamlin(1994) and Persson, Roland, and Tabellini(1997) show that separation of power by itself is not sufficient to improve outcomes for the majority. Bardhan and Mookherjee(2006) suggests that the impact of decentralization on government accountability is a first-order concern in developing countries (a setting similar to Ancien R´egimeFrance), and that it typically depend on the relative potential for capture of the central and local government. Stasavage(2014)’ s account of the development of autonomous cities also suggests that representative institution may easily devolve into oligarchic institutions over time. This leads to the following hypothesis:

• Hypothesis 2: Elite rent-seeking is more prevalent under consent institutions.

3.3 Total effect

The informational and accountability mechanisms go in opposite directions in terms of their impact on the welfare of the governed, but they are not in contradiction with each other. Elites may successfully bargain for lower taxation and engage in rent-seeking through other avenues than taxation. The empirical analysis cannot capture all relevant dimensions of elite (or state) rent-seeking. Its goal is rather to assess the terms of the trade-off peasants face between extraction by the state or local elites. It is therefore outside of the scope of this article to estimate a net total effect of consent the impact on the welfare for non-elite populations. I will however approximate such an exercise by analyzing data on living standards.

4 Historical background

This section describes the functioning of provincial institutions in pays d’´etat,their origin, as well as the broader context of taxation in Ancien R´egimeFrance.

4.1 Functioning of the provincial estates

The main role of the provincial estates was to vote on the level of central taxation. They managed direct tax collection and therefore had the power to delay it if the central state excessively increased the fiscal burden. They were convened by the king on a regular basis (e.g. every year in Languedoc, every three years in Burgundy). The three medieval groups Clergy, Nobility and Third Estate were represented in the estates, but unequally. Most estates voted by order, which disadvantaged the Third Estate, but some voted by head. For instance, in Languedoc, the provincial estates were composed of 112 members, corresponding to 88 votes: 44 for the Third Estate, 22 for the Nobility, 22 for the Clergy. Such a rule still gave the Third Estate only half of the vote, but it did increase its bargaining power relative to the privileged orders, as compared to other provincial estates. Representation was also imbalanced within orders. The Third Estate delegates were municipal un- elected officials of the main cities of the province, which means that the interests of peasants were not

5 represented.3 Clergy and Noble representation was aristocratic. In Brittany, all nobles could legally par- ticipate in the estates’ discussions (which occasionally raised attendance up to 1,000). More commonly, access was restricted to the main noble families (e.g.: only 22 high barons could attend the estates in Languedoc; access was conditioned by possession of a century-old title in Burgundy). Similarly, Clergy representation was composed of bishops and abbots, not local priests. A clear limitation of the power of the estates’ power was that they lacked control over expenditures. During the 1700s, some provincial estates did develop important roles in agricultural improvement, roads and canals (Tocqueville, 1856; Swann, 2004; Durand, Jouanna, and P´elaquier, 2014), this was not the rule. Top-down appointed officials (the intendants) controlled most of provincial public good provision even in pays d’´etat. The estates had no role in local repression or justice.

Table 1: Characteristics of pays d’´etatand pays d’´elections

Consent (pays d’´etat) No consent (pays d’´election)

Taxation: shared power with royal adminis- Intendants allocate and oversee the collection trators (intendants). of royal taxation.

Provincial Estates: assemblies of represen- No provincial estates. tatives of the three orders (Clergy, Nobility & Third Estate). Role: Consent to, bargain over, and distribute No consent to taxation. royal taxation. Regular meetings. Representation: C: bishops of main dioceses; N: main noble families or restrictions; TE: un- elected mayors of main cities. Voting: by order, except in Languedoc and Provence; TE has at most half of the votes.

Note: PE: provincial estates; TE: Third Estate; N: Nobility; C: Clergy.

4.2 Origins of the provincial estates

We can distinguish two origins of the provincial estates.4 A first group emerged as institutions consenting to taxation during the Hundred Years’ War opposing France to England (1337-1453). At that time, the king had no right to levy general taxation to fund the war and thus needed the consent of various assemblies representing the towns, the lords and the clergy. Since national-level assemblies typically resulted only in vague agreements, bargaining with local assemblies was often the only way for the king to collect taxation (Mousnier, 1974; Major, 1980; Barbiche, 2015; Henneman, 2015). The most powerful of these

3While these offices used to be elective in the beginnings of the provincial estates, they became non-elective following 17th century reforms reinforcing the control of the central state over municipalities (Petit-Dutaillis, 1947). 4This section is based on the following general historical works: Major(1980), Marion(1923), Barbiche(2015) and Mousnier(1974). On individual provincial estates: Beik(1985) and Durand, Jouanna, and P´elaquier(2014) on Languedoc; Legay(2001) on the Northern estates; Swann(2003) on Burgundy.

6 Figure 2: Evolution of the area share of pays d’´etatin the French territory (1500-1789)

Area share of provinces with consent to taxation in French territory

Valois House Henry IV Louis XIII Louis XIV Louis XV Louis XVI Area share of pays d'etat Area share of pays 0.0 0.2 0.4 0.6 0.8 1.0

1500 1550 1600 1650 1700 1750 1800

local assemblies were the provincial estates, and most preserved their self-governing privilege after the war. A second group is found in territories which were acquired by the French monarchy through alliance or conquest (Burgundy, Brittany and Provence were acquired between 1477 and 1534). Later additions include a number of small provinces in the Southwest (1607-1620) as well as in the North (1659). All these territories had pre-existing forms of representative estates consenting to taxation before integration into France, which the monarchy preserved in order to insure their loyalty. As the monarchy became stronger, consent to taxation receded. Starting with the reign of Louis XIII, many provincial estates of the first group were suppressed as their military importance diminished (Blaufarb, 2010). These provincial estates were the weakest ones, as they consented to taxation but did not control the process of tax collection. Under Louis XIV, suppression of the weakest provincial estates continued and new territorial additions such as Franche-Comt´e,Alsace or Lorraine no longer benefited from consent to taxation. The share of territory with consent to taxation fell from almost 60% in 1600 to a third in 1700 (see Figure2). The provincial estates survived for reasons that are largely contingent on short-term dynamics and foreign policy, which is important for the validity of the empirical strategy of comparing regions with provincial estates to regions without. For instance, Minister Marillac intended to suppress the Burgundian provincial estates but was disgraced in 1630 on matters of foreign policy (Swann, 2003, p.42). Controller- General of Finances Jean-Baptiste Colbert was also adamant about suppressing the Southwestern estates but died in 1683 before reaching his goal (Major, 1980, p.652).

4.3 The state under the absolute monarchy

In provinces without provincial estates, the monarchy taxed without consent. War dominated expendi- tures5, requiring large short-term increases in revenue (Bonney, 1999). Louis XIV (1661-1715) created a bureaucracy based on appointed royal officials called the intendants. The intendants had full powers over administration, justice and taxation at the g´en´eralit´elevel (a tax district roughly similar to the provinces

5In 1759, war expenditures represent almost half of all expenditures (109 millions livres out of 237 millions).

7 shown on Figure1), but shared powers with provincial estates in pays d’´etat. The main direct tax of the Ancien R´egimewas the taille, a land tax. It became a permanent tax in 1439, but Clergy and Nobility were soon exempted from it. Urban Third Estate elites and office-holders were typically exempted as well. As a result, not paying the taille became a marker of social status and it fell almost entirely on peasants. From the end of the 1600s the monarchy tried more and more to tax privileged elites. New taxes without exemptions were created: the capitation (created in 1695, suppressed and revived in 1710), the dixi`eme (1710), the vingti`eme (1749, 1756). While these taxes were less unequal than the taille, they still added to the burden of the peasantry and elites often managed to negotiate lower rates (Kwass, 2006).

5 Empirical strategy

5.1 Data

5.1.1 Rebellions (1700-1788)

As the primary outcome variable, I use a dataset of commune-level local “rebellions” constructed by Nicolas (2008). The author defines these rebellions as follows: “the threshold of collective violence is met when a group of at least four individuals, not belonging to the same family, directly perpetuate violence against one or more representatives of a political, religious, economic power, etc., or attacks property, buildings, furniture, papers, various signs symbolizing these powers.” The dataset provides information on 8,300 of such events between 1661 and March 1789. The left panel of Figure3 shows the geographic distribution of these rebellions. Clusters of fiscal rebellions are visible along tax zone boundaries. Nicolas(2008) coded the cause of these rebellions: 41% are fiscal rebellions, 15% are food riots and 9.6% are against local elites6. The remaining causes of rebellions are hostility to the repressive and judiciary apparatus (15%), work and religious conflict. In the appendix, Table A.2 provides further information on the typology of rebellions. I use the share of rebellions about central taxation and local elites as a proxy for peasants’ dissatisfaction with these respective topics. Since repression was not a function of the provincial estates, discrepancies in rebellions between consent and no consent provinces should reflect actual differences on the ground.

5.1.2 Grievance lists from the 1789 Estates-General

As a secondary outcome variable, I use data on grievances lists drafter for the 1789 Estates-General. In January 1789, after several decades of failed attempts to reform its fiscal system, the monarchy convened the Estates-General for the first time since 1614. 384 districts (bailliages) elected delegates to communicate their grievances to the King. The election of Third Estate delegates happened in two or three levels: villages, intermediary villages or towns (secondary districts), and finally district capital cities (primary districts). At each of these levels, grievance lists were drafted in order to provide guidance to the delegates. Grievances lists from rural villages were aggregated into general lists at the district level. We can thus distinguish two types of grievance lists of the Third Estate. First, the village-level lists (also called “parish lists” because voting was done at the parish level). Second, the district-level lists (or “general lists”), which resulted from the aggregation of lower-level lists.

6I created this category as including rebellions against: noblemen, the church, municipalities or notables.

8 Figure 3: Outcome variables

Rebellions (1700−1788) Grievances lists from the Estates General (1789)

No consent No consent Consent Consent

● ● ● ● ● ● ● Late additions Late additions ● ● Rebellions against taxation Bailliage district capital Rebellions against local elites ● Villages

●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●●●● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ●●● ● ● ● ●● ●● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

●

●● ● ● ● ●

● ●

●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

●

● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●

● ● ●

● ●

Notes: Left: data from Nicolas(2008). Conflict against local elites include conflict against: local lords, nobles, church, notables, municipality. Right: data from Shapiro et al.(1998).

I use the grievance data from historians Shapiro et al.(1998). The authors developed a coding scheme to analyze the grievance lists, summarizing each distinct grievance with a limited number of variables. Each grievance is coded using “subject” and “action” fields. The subject code consists of a four-level hierarchy (e.g. “Government > Taxation > Direct taxation > Taille”). Subject level 4 has 1,193 categories. There are 92 action codes (e.g.:“abolish”, “equalize”, “reform”, etc), as well as optional “object” and “remarks” codes.7 For more detail, see Tables A.3, A.4 and A.5 in the appendix, which show the main subjects levels that I will use in the empirical analyses. I use three grievance datasets: Third Estate grievances lists (154 primary districts, representing 384 secondary districts, amounting to 47,711 grievances); Nobility grievance lists (127 districts, 27,064 grievances); Third Estates village grievance lists (504 villages, corresponding to 19 primary districts and 28,322 grievances). I georeferenced all district capitals and villages from the Markoff-Shapiro dataset (see the right panel of Figure3). As with the rebellion data, I measure dissatisfaction with taxation and local elites as the share of grievances about these respective topics out of every grievances in the list.

5.1.3 Grievances, rebellions and objective conditions

Both outcome variables are proxies of dissatisfaction with a) taxation and b) local elites. The grievance data is certainly a more indirect proxy of non-elite attitudes. It is well known that the Third Estate general lists were influenced by the urban bourgeoisie and landowners. At the same time, it is unlikely that the Nobility and the Clergy directly influenced these lists (Shapiro et al., 1998). I complement data on grievances and rebellions with partial data on objective conditions during and

7An earlier content analysis of the 1789 grievance lists was created by Hyslop(1968) for of a study of nationalist sentiment. I do not use this dataset because it relies on subjective scales of “national sentiment”, “religiosity”, or“democracy”, with only a limited number of concrete topics.

9 after the Ancien R´egime.This includes measures of taxation, living standards (soldiers’ height, mortality) and economic development (urban growth). For the grievance data, the comparison between general Third Estate and rural village lists will help us address potential ambiguities.

5.2 Identification strategy

Consent to taxation is the independent variable of interest. I use only locations where treatment is well defined: either pays d’´etat(consent) or pays d’´election(no consent). In order to focus on the long run consequences of consent, I exclude the more recent additions to the French territory that did not consent to taxation but preserved a degree of independence along other dimensions (the pays d’imposition). The RD is conducted along the boundary between pays d’´electionand pays d’´etatonly.

5.2.1 OLS regressions

I first conduct OLS regressions of the outcomes of interest on a “consent” indicator, estimating the following equation:

0 yd,g = α + β consentd,g + Xd,gγ + d,g. (1) yd,g is the outcome of interest, i.e. a function of the frequency of a grievance topic in district d in g´en´eralit´e g. Because royal taxation was implemented at this level, standard errors are clustered by g´en´eralit´elevel.

consentd,g is a variable equal to one when a district capital d is located in a pays d’´etat,and zero when located in a pays d’´election. Xd,g represents several geographic controls which could impact both the presence of provincial estates and the distribution of grievances: distance from Paris, terrestrial foreign border, sea, and major rivers; suitability to wheat agriculture; terrain ruggedness. I also include a control for the level of urbanization in 1400 (before the provincial estates became active), defined as the number of cities with a population of more than 10,000 within 50 km8. Including 18th century urbanization levels would lead to post-treatment bias if provincial institutions have an impact on economic growth and city development.

5.2.2 Regression discontinuity

Despite the controls, the OLS specification may not identify the true effect of having institutions consenting to taxation. Differences in geographic, economic and political conditions could explain both the emergence of the provincial estates and the pattern of 1789 grievances. To address this challenge, I use a geographic regression discontinuity design, comparing locations close to the pays d’´etat/pays d’´electionborder. The boundary forms a two-dimension discontinuity in the latitude-longitude space and is composed of five distinct segments: South, Southwest, West, North and East (Figure1). For a given district d in g´en´eralit´e g, located in a certain bandwidth of the boundary of interest, we estimate:

0 yb,g = α + β paysdetatb + f(geographic locationb,p) + δ segb,p + Xb,gγ + b,g. (2)

The controls are the same as the OLS specification. f(.) is a flexible function of geographic location. Following Dell, Lane, and Querubin(2018), I examine polynomials in latitude and longitude, as well as

8Results are robust to alternative definitions of this measure of urbanization. Population figures come from Bairoch, Batou, and Ch`evre(1988).

10 in distance to the boundary. The preferred specification is a linear polynomial in latitude and longitude, as it takes fully into account the geographic nature of the discontinuity of interest (Keele and Titiunik,

2015). segb,p is a border fixed effect based on the five previously mentioned boundary segments, insuring that we are comparing bailliages that are close to each other along the boundary.

5.2.3 Balance checks and selective sorting

The RD design requires that all relevant factors related to the outcome of interest vary smoothly at the boundary between pays d’´etat and pays d’´election. I run RD regressions with control variables as the dependent variable within a bandwidth of 10 to 50 km by 1 km increments. Figure4 summarizes the results. Balance is achieved on most geographical and administrative variables: wheat suitability, terrain ruggedness, distance to sea and navigable rivers. Provinces with consent to taxation also tended to be more urbanized prior to the creation of the estates. Conversely, noble elites had shorter lifespans. This suggests that the bargaining power of urban Third Estate relative to the nobility was relatively favorable in regions that gained consent to taxation. I include imbalanced variables as controls in all OLS and RD regressions. The RD design also requires that there should be no selective sorting of the units of analysis across the boundary, i.e. the emergence of provincial estates should not lead to out-migration from pays d’´election to pays d’´etat(or the opposite). Regarding rural populations, the demographic history literature paints a picture of largely sedentary rural societies. While mobility was relatively frequent, it rarely exceeded 10 to 15 km (Poussou, 1970; Hayhoe, 2016).

6 Results

6.1 Did the provincial estates reduce the fiscal burden?

First, I assess how consent impacted the level of taxation. Here by taxation I refer to indirect and direct taxation, collected by either the central state, provincial estates or tax farmers. This does not include taxation by non-state actors as part of the seigneurial system, which I will discuss later. There is unfortunately little fiscal data on the Ancien R´egime,because accounting documents were destroyed during a 1737 fire at the Court of Accounts ) and the revolutions of 1789 and 1848 (Bonney, 1999). I therefore rely on province level indicators as well as post-1789 data. In addition, a difficulty in estimating the fiscal burden in pays d’´etatis that a share of taxation remained under the control of the estates. While there is data on royal tax receipts from pays d’´etatand pays d’´electionduring the 1600s and some of the 1700s, similar information is unavailable for local taxation in pays d’´etat,making it difficult to assess the actual tax burden. Necker, who was Finance Minister between 1777 and 1781, provides provincial-level estimates of total tax burden in 1784, a few years before the revolution (Figure ??). A caveat is that these numbers are simply estimates and the exact breakdown between local and central taxation is unclear. The result is however informative: on average, provincial estates are associated with a 26% reduction in per capita tax burden, but there is a lot of variation across provinces. Information from aggregated reports by finance ministers on direct tax receipts during the 1600s and 1700s confirm the pattern of lower taxation in provinces with consent. Receipts from provinces without consent were always at least four times larger than in provinces

11 Figure 4: Balance checks 0.4 1.5

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 ● wheat ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

N grievances ● ● 0.0 ● ● ● ● ● ● ● ● ● ● −0.4

10 20 30 40 50 −1.5 10 20 30 40 50 xs

xs 2 0.1

● 1 rugged ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● N subjects 1 0 −0.2

10 20 30 40 50 −2

xs 10 20 30 40 50 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● xs dist. Paris 1.5

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.2 N subjects 2 ● ● ● ● ● ● 0.0 ● ● ● ● ● ● 10 20 30 40 50

xs −1.5 10 20 30 40 50 0.1 dist. river ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

xs 1.0 −0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

N subjects 3 ● ● ● 0.0 10 20 30 40 50 ● ● ● ● ● ● ● ●

xs 0.2 −1.0

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 10 20 30 40 50 dist. sea ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0

xs 1.0 −0.2

● ● ● 10 20 30 40 50 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● N subjects 4 ● ● ● ● ● ● ● ● 0.0 xs −1.0 0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● dist. border ● ● ● ● ● ● ● ● ● ● ● ● 10 20 30 40 50 −0.4

xs 1.5 10 20 30 40 50 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● N venal ● ● ● ● xs offices 0.0

● ● ● 0.2 ● ● ● ● ● ● ● ● ● ● ● ● dist city ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1.5 10 20 30 40 50 −0.6

10 20 30 40 50 3

xs Value venal 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.5 ● ● ● ● ● ● ● ● ● ● ● offices −1 N cities −3

−1.0 10 20 30 40 50 10 20 30 40 50 1.0 xs

0.5 Noble age ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 dist. intendance ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● at death ● ● ● −0.5 −1.0 10 20 30 40 50 10 20 30 40 50 Notes: Each rows corresponds to a different dependent variable. Points represent the estimated effect of consent on the dependent variable for bandwidths between 10 and 50 km. 95% confidence intervals based on standard errors clustered at the g´en´eralit´elevel. N cities and distance to closest cities are respectively the number of large cities in a 50 km bandwidth and the distance to the closest large city. Large city is defined as having more than 10,000 inhabitants in 1400, population data from Bairoch, Batou, and Ch`evre(1988). Data on venal offices from Esteves(2015). Data on nobles’ lifespan from Cummins(2017). N=1,063 and 315 geographical locations.

12 with consent, although they started converging from the late 1600s as a result of Louix XIV’s centralizing policies (Figure ??). As a complement to the province-level data, I analyze arrondissement-level data on land tax per capita in 1802. After the revolution, Ancien R´egimetaxes were suppressed and replaced by a land tax, the contribution fonci`ere. Due to the lack of a proper cadaster, this tax was still largely based on pre-revolution evaluations, so this gives us a measure of local tax burden during the late Ancien R´egime.I find indeed that taxation per capita was much lower in former provinces with consent. Figure6 shows the result from a RD analysis close to the boundary between consent and no consent provinces: 1802 land tax per capita is significantly lower in provinces with consent.

Figure 5: Estimates of per capita taxation by province (1784)

Valenciennes Rennes Montpellier Lille Dijon Auch & Pau Aix Bordeaux & Bayonne Tours Soissons Rouen Riom pays Poitiers consent Paris no consent generalité Orleans Moulins Montauban Lyon Limoges La Rochelle Grenoble Caen Chalons Bourges Amiens Alencon

0 20 40 60 Per capita taxation (livres) Source: Necker(1785). Jacques Necker was finance minister between 1777 and 1781. Necker’s estimates include both central (royal) taxation and additional local taxation levied by the provincial estates.

13 Figure 6: RD: impact of consent on per capita land tax in 1802 5

consent 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● −5

20 25 30 35 40 45 50

6.2 Was taxation better accepted? impact on 1789 grievances and rebellions

We know that the provincial estates did lower taxation to a limited extent, but this does not tell us whether it benefited non-elite populations. Table2 shows OLS estimates of the impact of consent on 1789 grievances and rebellions. Third Estate grievances about taxation are clearly less prominent in provinces with consent, a difference that is statistically significant at the 1% level for the general lists (column 1) and 10% for the partial data on rural villages. Interestingly, we also observe lower rates of grievances about taxation in the Nobility lists (column 3). If we take into account the fact that the Nobility overall devotes less attention to the topic of taxation, the relative effect is even twice larger than the Third Estate (−30% versus −15%). This suggests that the decrease in taxation is not done at the expense of local elites since they benefit as well from it. Fiscal rebellions are also less frequent, a difference that is significant at the 1% level (column 4). Column 5 shows however that this is driven by enforcement of the salt tax, because the effect is positive (though not statistically significant at conventional levels) when we exclude this motivation from the count of rebellions. The RD analysis confirms that the OLS findings hold in a close bandwidth around the consent/no consent boundary (Figure7). In the appendix, Figure B.5 shows that the effect is not driven by one particular boundary segments and that dropping the provinces with a strong Third Estate, which are all located in the South, increases the size of the coefficients. Figure B.6 shows RD plots with distance from the boundary as the running variable, where the effect is estimated using a local linear polynomial within the optimal bandwidth (37 km), following Calonico, Cattaneo, and Titiunik(2014). The bottom-right panel show that non-salt tax fiscal rebellions were actually more frequent in consent provinces. The findings provide mixed evidence for the first hypothesis. The overall tax burden appears to have been lower, taxation appears relatively less in grievances and rebellions. At the same time, most of the variation in rebellions at least comes from the gabelle regime, which was not managed by the provincial estates.

6.3 Did consent improve living standards?

I have shown that provincial representation led to lower taxation, which translated in lower dissatisfaction with taxation from both elite and non-elite groups. Now I turn to the second question: did these gains on

14 Table 2: OLS: Consent and dissatisfaction with royal taxation

1789 grievances Rebellions Third Estate Villages Nobility % Fiscal % Fiscal no salt tax (1) (2) (3) (4) (5)

consent −0.636∗∗∗ −0.398∗ −0.737∗∗∗ −0.340∗∗∗ 0.176 (0.238) (0.229) (0.183) (0.095) (0.113)

mean dep var 0.304 0.325 0.145 0.431 0.096 N 137 504 122 3217 3217 N cluster 30 19 29 32 32

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors in parentheses, clustered at the g´en´eralit´elevel.

Figure 7: RD: Consent and dissatisfaction with royal taxation 1

% grievances ● 0 about taxation ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● (Third Estate) ● ● ● ● ● ● ● ● −1

20 25 30 35 40 45 50

xs

% grievances 1

about taxation 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● (Nobility) ● −1

20 25 30 35 40 45 50 1

% fiscal ● ● 0 rebellions ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1

20 25 30 35 40 45 50

xs 1

% fiscal ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● rebellions ● ● ● ● ● ● ● ● ● ● ● ● excluding 0 salt tax −1

20 25 30 35 40 45 50 taxation improved the welfare of the non-elite population? To answer this question, I first analyze the relationship between the provincial estates and a number

15 Table 3: Living standards in the short and long run

Before 1789 post-1789

soldiers’ age at age at % stunted mortality % urban % urban pop height death death conscripts growth (1) (2) (3) (4) (5) (6) (7) (8)

consent −0.046∗ −0.009 −0.261∗∗ 0.372∗∗ −0.170 0.188 0.065 −0.096 (0.027) (0.136) (0.109) (0.182) (0.270) (0.420) (0.506) (0.070)

Timeframe 1716-1784 1740-1790 1800-1829 1818-1830 1836 1800 1851 1800-1851 Bandwidth 25 50 50 25 25 50 50 50 Geographic commune village village canton arr. dep. dep. commune unit Locations 853 134 136 761 69 39 37 9,076 Observations 5,590 663 540 693 69 39 37 9,076 R2 0.025 0.210 0.175 0.237 0.247 0.648 0.632 0.057 Adj. R2 0.021 0.186 0.147 0.220 0.034 0.419 0.369 0.056

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. (1) Soldiers’ height (in mm) data from Komlos, Hau, and Bourguinat(2003). (2) and (3) Civil registry data on a 1/100 sample of rural communes between 1740 and 1829, amounting to 358 communes in the territory of interest, from S´eguy(2001). (4) Cantonal data on conscripts height from Demonet, Dumont, and Le Roy Ladurie (1976). (5-7) Census data. (8) commune-level population data from Motte and Vouloir(2007). Models (1-3) include decade fixed effects. Standards errors are clustered at the g´en´eralit´elevel.

of indicators of living standards and prosperity from various sources, including soldier’s height during and after the Ancien R´egime,mortality data and population growth. Table3 shows results from analyses for a small bandwidth (25 or 50 km depending on data sparseness close to the border). OLS analyses paint a similar picture and are shown in the appendix (Table A.8). Column (1) shows that the average height of soldiers between 1716 and 1784 is slightly lower under consent, suggesting lower levels of nutrition. The magnitude of the estimate is small (.05 mm) and the difference is only significant at the 10% level. Rural mortality, as proxied by average age at death in a sample of 134 villages, is similar during the Ancien R´egime(column 2) but becomes significantly higher between 1790 and 1830 (column 3). Stunting among 1818-1830 conscripts, which is a proxy of malnutrition during the 1798-1810 period, is more frequent in former provinces with consent to taxation (column 4). There is also no significant difference in 1836 mortality at the arrondissement level, urbanization in 1800 and 1851, and commune population growth (columns 5-8). Data quality is not ideal, but together the results paint a clear picture: the provincial estates were not associated with major improvements in welfare, either in the short or long run. All estimates point at either a negative effect or no effect of consent. This empirical exercise provides us with an upper bound of the benefits of consent institutions for non-elite populations.

6.4 Were local elites more extractive of peasants?

An intuitive explanation of the limited welfare benefits of the provincial estates is that the local elites who controlled them were more extractive of peasants. I provide evidence on this mechanism by analyzing

16 grievances and rebellions linked to the seigneurial system. The seigneurial system refers to taxes and rights (lucrative or honorific) that were collected by owners of noble land on peasants (a full description is available in the appendix on Table A.5). Peasants were required to make annual payments to the seigneur, either in cash (cens) or with produce (champart). The seigneur also charged taxes for using infrastructure like the flour mill, wine press and baker’s oven (banalit´es).Theseigneur could also demand a period of unpaid labour from his tenants, called the corv´ee. Many peasants were also subject to seigneurial courts, which were overseen by the seigneur. The seigneurial system derived from the medieval feudal system, keeping its economic dimension (Markoff, 2010).

Table 4: OLS: Dissatisfaction with local elites

Grievances seigneurial system Rebellions

Third Third Villages Villages % Seign. % Local % Food Estate Estate elites riots

(1) (2) (3) (4) (5) (6) (7)

consent 0.019 0.031∗∗ 0.045∗ 0.074∗∗ 0.062∗∗∗ 0.092∗∗∗ −0.003 (0.015) (0.014) (0.025) (0.034) (0.011) (0.016) (0.025)

mean dep var 0.136 0.142 0.078 0.08 0.06 0.107 0.154 N 140 110 504 442 3217 3217 3217 N cluster 30 25 19 16 32 32 32 Consent institutions All Weak All Weak All All All included

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors in parentheses, clustered at the g´en´eralit´elevel. The dependent variable for grievance data is the share of grievances about the seigneurial system out of all grievances. “Local elites” rebellions include rebellions against the seigneurial system, the Church, noble privilege, municipal authority and notables.

Unlike royal taxation, a theme where elite and non-elite interests could plausibly be aligned at the provincial level, the seigneurial system raised a clear conflict of interest between elites and non-elite under the seigneurial system.9 Columns 1-4 on Table4 report OLS estimates of the impact of consent to taxation on Third Estate grievances about the seigneurial system. The coefficient is positive but not statistically significant at conventional levels if we consider all provinces with consent to taxation (column 1 and 3). However, when we restrict the analysis to provinces with consent and a weak Third Estate (i.e. excluding the Languedoc and Provence provinces), we find a positive effect that is statistically significant at the 5% level (column 2 and 4). Columns 5 reports results with the share of rebellions about the seigneurial system as the dependent

9For this topic, we can view the Third Estate general lists as more representative of both elite and non-elite preferences within this group, because Third Estate urban elites would have little reason to filter out grievances about noble elites’ extractive behavior. In addition, if urban elites were co-opted by the nobility in pays d’´etat,this would lead to downward bias in the estimates, making it a hard test for the hypothesis that consent to taxation results in a more extractive seigneurial system.

17 variable. Such rebellions are approximately twice more likely in consent provinces than in provinces without consent, an effect than is statistically significant at the 1% level. The effect is even larger with a dependent variable including other types of rebellion against local elites: municipalities, noble, Church and notables (column 6). As a placebo outcome I looked at the impact on food riots: we find no effect that is statistically significant at conventional levels.

Figure 8: RD: Dissatisfaction with local elites 1

● ● ● ● ● ● ● ● ● 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● % rebellions ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● % grievances 0 against seign 0 seign system system −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs xs 1

● ● ● ● ● ● ● 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● % rebellions ● ● ● ● ● ● ● ● ● ● ● ● ● % abolish 0 against local 0 seign. system elites −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs xs 1 1

% reform ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

0 ● ● 0 ● % food riots ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● seign. system ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

Figure8 shows that the same results hold in a small bandwidth along the pays d’´etat/pays d’´election boundary. Additional analyses show that consent to taxation is associated with more extreme views on the seigneurial system, as there are more grievances demanding the abolition of aspects of the seigneurial system and less demanding reform (i.e. compensation).

7 Conclusion

I investigated the consequences of provincial representation on non-elite populations. I showed that consent led to a reduced overall tax burden for the provinces which benefited from it but exacerbated the elite- peasant conflict. I found no evidence of an effect on living standards or economic development. This article illustrates the ambiguous long-term effects of early representation on later democratization trajectories. Early representative institutions can initially survive in an authoritarian setting, providing regime a degree of stability by satisfying local elites. At the same time, they also exacerbate local conflicts, in our case between privileged elites and peasants, undermining the long term stability of the regime. During the decades leading to the 1789 revolution, the provincial estates were at some point considered as an avenue for reforming the absolute monarchy: there were discussions of generalizing the provincial estate systems to the whole realm in the 1760s, but they ultimately failed because the state was unwilling to give too much power to noble elites (Legay, 2004). My results show that they were indeed unlikely to address broader sources of dissatisfaction with the regime, a failure that contributed to the violent destabilization of the regime in 1789.

18 References

Acemoglu, D., S. Johnson, and J. Robinson. 2005. “The rise of Europe: Atlantic trade, institutional change, and economic growth.” American economic review 95:546–579.

Acemoglu, D., T. Reed, and J.A. Robinson. 2014. “Chiefs: Economic development and elite control of civil society in Sierra Leone.” Journal of Political Economy 122:319–368.

Ahmed, A.T., and D. Stasavage. 2020. “Origins of early democracy.” American Political Science Review 114:502–518.

Angelucci, C., S. Meraglia, and N. Voigtl¨ander.2017. “The Medieval Roots of Inclusive Institutions: From the Norman Conquest of England to the Great Reform Act.” Working paper, National Bureau of Economic Research.

Bairoch, P., J. Batou, and P. Ch`evre. 1988. Population des villes europ´eennesde 800 `a1850: banque de donn´eeset analyse sommaire des r´esultats(la). Librairie Droz.

Baldwin, K. 2015. The paradox of traditional chiefs in democratic Africa. Cambridge University Press.

Barbiche, B. 2015. Les institutions de la monarchie fran¸caise`al’´epoque moderne. Presses universitaires de France.

Bardhan, P., and D. Mookherjee. 2006. “Decentralization, corruption and government accountability.” International handbook on the economics of corruption 6:161–188.

Bates, R.H., and D.D.H. Lien. 1985. “A note on taxation, development, and representative government.” Politics & Society 14:53–70.

Beik, W. 1985. Absolutism and society in seventeenth-century France: State power and provincial aristoc- racy in Languedoc. Cambridge University Press.

Blaufarb, R. 2010. “The Survival of the Pays d’Etats:´ The Example of Provence.” Past and Present 209:83–116.

Bonney, R. 1999. The rise of the fiscal state in Europe c. 1200-1815 . Clarendon Press.

Boucoyannis, D. 2015. “No taxation of elites, no representation: state capacity and the origins of repre- sentation.” Politics & Society 43:303–332.

Brennan, G., and A. Hamlin. 1994. “A revisionist view of the separation of powers.” Journal of Theoretical Politics 6:345–368.

Calonico, S., M.D. Cattaneo, and R. Titiunik. 2014. “Robust nonparametric confidence intervals for regression-discontinuity designs.” Econometrica 82:2295–2326.

Cox, G.W. 2017. “Political institutions, economic liberty, and the great divergence.” The Journal of Economic History 77:724–755.

Cummins, N. 2017. “Lifespans of the European elite, 800–1800.” The Journal of Economic History 77:406– 439.

19 de Kadt, D., and H. Larreguy. 2015. “Agents of the regime? Traditional leaders and electoral clientelism in South Africa.” Dept. Work. Pap., Dept of Polit. Sci., Massachusets Institute of Technology, pp. .

Dell, M., N. Lane, and P. Querubin. 2018. “The historical state, local collective action, and economic development in Vietnam.” Econometrica 86:2083–2121.

Demonet, M., P. Dumont, and E. Le Roy Ladurie. 1976. “Anthropologie de la jeunesse masculine en France au niveau d’une cartographie cantonale (1819-1830).” Annales, pp. 700–760.

Durand, S., A. Jouanna, and E. P´elaquier.2014. “Des ´etatsdans l’Etat.” Les Etats´ de Languedoc de la Fronde `ala R´evolution,Ginebra, Droz, pp. .

Esteves, R. 2015. “Archomania, Venality and Private Finances on the Eve of the French Revolution.” Working paper.

Garfias, F. 2019. “Elite Coalitions, Limited Government, and Fiscal Capacity Development: Evidence from Bourbon Mexico.” The Journal of Politics 81:94–111.

—. 2018. “Elite competition and state capacity development: Theory and evidence from post-revolutionary Mexico.” American Political Science Review 112:339–357.

Hayhoe, J. 2016. Strangers and neighbours: Rural migration in eighteenth-century northern Burgundy. University of Toronto Press.

Henneman, J.B. 2015. Royal Taxation in Fourteenth-Century France: The Development of War Financing, 1322-1359 , vol. 1443. Princeton University Press.

Hoffman, P.T., and K. Norberg. 2002. Fiscal crises, liberty, and representative government, 1450-1789 . Stanford University Press.

Hyslop, B.F. 1968. French Nationalism in 1789, according to the General Cahiers. New York: Octagon Books, 1968 [c1934].

Keele, L.J., and R. Titiunik. 2015. “Geographic boundaries as regression discontinuities.” Political Analysis 23:127–155.

Komlos, J., M. Hau, and N. Bourguinat. 2003. “An anthropometric history of early-modern France.” European Review of Economic History 7:159–189.

Kwass, M. 2006. Privilege and the politics of taxation in eighteenth-century France: libert´e,´egalit´e,fiscalit´e. Cambridge University Press.

Lachmann, R. 1989. “Elite Conflict and State Formation in 16th- and 17th-Century England and France.” American Sociological Review 54:141–162.

Lechler, M., and L. McNamee. 2018. “Indirect colonial rule undermines support for democracy: Evidence from a natural experiment in Namibia.” Comparative Political Studies 51:1858–1898.

Legay, M. 2001. Les Etats provinciaux dans la construction de l’Etat moderne, aux XVIIe et XVIIIe si`ecles. Travaux du Grand Si`ecle,Droz.

20 Legay, M.L. 2004. “Un projet m´econnu de d´ecentralisation au temps de Laverdy (1763-1768): les grands Etats´ d’Aquitaine.” Revue historique, pp. 533–554.

Major, R. 1980. Representative Government in Early Modern France. Etudes pr´esent´ees`ala Commission internationale pour l’histoire des assembl´eesd’Etats, Yale University Press.

Mamdani, M. 1996. “Indirect rule, civil society, and ethnicity: The African dilemma.” Social Justice 23:145–150.

Mares, I., and D. Queralt. 2015. “The non-democratic origins of income taxation.” Comparative Political Studies 48:1974–2009.

Marion, M. 1923. Dictionnaire des institutions de la France aux XVIIe et XVIIIe si`ecles. A. Picard.

Markoff, J. 2010. Abolition of Feudalism: Peasants, Lords, and Legislators in the French Revolution. Penn State Press.

Mayshar, J., O. Moav, and Z. Neeman. 2017. “Geography, transparency, and institutions.” American Political Science Review 111:622–636.

Møller, J. 2017. “The birth of representative institutions: the case of the crown of Aragon.” Social Science History 41:175–200.

Motte, C., and M.C. Vouloir. 2007. “Le Site cassini. ehess. fr: Un Instrument d’Observation pour une Analyse du Peuplement.” Bulletin du Comit´efran¸caisde cartographie 191:68–84.

Mousnier, R. 1974. Les institutions de la France sons la monarchie absolue 1598-1789: Soci´et´eet Etat. Presses universitaires de France.

Naseemullah, A., and P. Staniland. 2016. “Indirect rule and varieties of governance.” Governance 29:13–30.

Necker, J. 1785. De l’administration des finances de la France, vol. 1. Paris.

Nicolas, J. 2008. La r´ebellion fran¸caise: mouvements populaires et conscience sociale, 1661-1789 . Galli- mard.

North, D.C. 1993. “The paradox of the West.” Economics Working Paper Archive, Washington Univer- sity—St. Louis, Missouri, pp. .

Persson, T., G. Roland, and G. Tabellini. 1997. “Separation of powers and political accountability.” The Quarterly Journal of Economics 112:1163–1202.

Petit-Dutaillis, C. 1947. Les communes fran¸caises. Albin Michel.

Potter, M., and J.L. Rosenthal. 1997. “Politics and Public Finance in France: The Estates of Burgundy, 1660-1790.” The Journal of Interdisciplinary History 27:577–612.

Poussou, J.P. 1970. “Les mouvements migratoires en France et `apartir de la France de la fin du XV e si`ecleau d´ebutdu XIX e si`ecle: approches pour une synth`ese.”In Annales de d´emographie historique. JSTOR, pp. 11–78.

Root, H. 1994. The Fountain of Privilege: Political Foundations of Markets in Old Regime France and England. California series on social choice and political economy, University of California Press.

21 S´eguy, I. 2001. La population de la France de 1670 `a1829: l’enquˆeteLouis Henry et ses donn´ees. Ined.

Shapiro, G., T. Tackett, P. Dawson, and J. Markoff. 1998. Revolutionary demands: A content analysis of the Cahiers de Dol´eances of 1789 . Stanford University Press.

Stasavage, D. 2014. “Was Weber right? the role of urban autonomy in Europe’s rise.” American Political Science Review 108:337–354.

—. 2010. “When distance mattered: Geographic scale and the development of european representative assemblies.” American Political Science Review 104:625–643.

Swann, J. 2004. “Parlements and provincial estates.” In W. Doyle, ed. The Oxford Handbook of the Ancien R´egime. Oxford, UK: Oxford University Press, chap. 10, pp. 93–110.

—. 2003. Provincial Power and Absolute Monarchy: The Estates General of Burgundy, 1661-1790 . New Studies in European History, Cambridge University Press.

Tocqueville, A.d. 1856. L’Ancien R´egimeet la R´evolution. Paris: Gallimard.

Van Zanden, J.L., E. Buringh, and M. Bosker. 2012. “The rise and decline of European parliaments, 1188–1789.” The Economic History Review 65:835–861.

Weitman, S.R. 1968. “Bureaucracy, Democracy, and the French Revolution.” PhD dissertation, Washing- ton University.

22 A Appendix Tables

Table A.1: Provinces and consent to taxation

Province Year Year of suppression Vote of acquisition of consent Pays d’´etat:consent to taxation until 1789 Northern estates a 1659-1678 by order Pyrenean estates b 1607-1620 by order Bourgogne 1477 by order Bretagne 1534 by order Comt´eDe Foix 1607-1620 by order Languedoc 1222-1271 by head Provence 1486 by head

Pays d’´election: no consent to taxation Anjou, Aunis, Berry, Bourbonnais, Champagne, Ile de France, Maine, Nivernais, Orl´eanais, N.A. Picardie, Saumurois and Touraine. Aunis, Limousin, Marche, Poitou and Saintonge ≈ 1450

Former pays d’´etat:past consent to taxation Guyenne et Gascogne 1550-1672 Auvergne 1672 Beaujolais 1688 Bresse, Bugey and Gex 1601 1601 Dauphin´e 1349 1628 Forez 1649 Lyonnais 1625 Normandie 1655

Pays d’imposition: recently conquered Alsace 1648 1683 Maritime Flandre 1659 1659 Franche-Comt´e 1674 1704 Hainaut 1659 1659 Lorraine 1766 1766 Roussillon 1659 1789 a Flandre, Artois and Cambr´esis. b Labourd, Basse-Navarre, B´earn,Bigorre, Soule, and Marsan.

Sources: Barbiche(2015), Mousnier(1974), Major(1980) .

23 Table A.2: Rebellion typology

Riot type Count Freq. Count Freq.

Fiscal: Resistance to the state’s fiscal or para-fiscal system 3284 41.1% 3284 41%

Local elites: Hostility to the seigneurial system and its agents 382 4.8% 770 10% Actions against notables 142 1.8% Action against municipal authority 141 1.8% Hostility to the Church 96 1.2% Hostility to nobility and noble privilege 9 0.1%

Food riots: 1257 15.7% 1257 16%

State (non-fiscal): Resistance to the judiciary, military or police state apparatus 1192 14.9% 1241 16% Rejection of state reforms 49 0.6%

Other: Religion, beliefs 260 3.3% 1448 18% Work conflict 432 5.4% Regional particularism 16 0.2% Miscalleneous 740 9.3% Notes: Source and typology from Nicolas(2008).

24 Table A.3: All grievance subjects: coding levels 1 and 2

Level 1 Level 2

Misc colonies; foreign policy; local subjects; attitudes toward non-catholics General general Constitution constitution; estates-general-estates general; powers of nation; political liberties; pow- ers of the monarch Economy economy; agriculture; commerce; finance; industry and manufacturing; transporta- tion Government government; administrative agencies; government: finances; the king; military; re- gional and local government; government taxation Judiciary judiciary; criminal prosecution and penalties; the courts; civil law and procedure; due process; enforcement agents of the court; legal professions Religion religion; church finances; the clergy; church organization; church-state (and church- rome) relations; dime; morality and religious practices Stratification system stratification system; criteria of mobility; economic class relations; seigneurial regime; blank subject field Notes: Source and typology from Shapiro et al.(1998).

25 Table A.4: Royal taxation grievance subjects: coding levels 3 and 4

Level 3 Level 4 government taxation - taxation- misc; taxation- general; droit d’aubaine; domaine du roi; franc, fief; finances misc ordinaires; regie general; taxation- miscellaneous; taxation- general tax advantages tax advantages - misc; tax advantage - general; bourgeois; capitalistes; clergy; cred- itors of the state; guilds; locality; negociant; noblemen; province; regular clergy; rentiers; seigneur; venal officers, direct tax agencies direct tax agencies- miscellaneous; direct tax agencies- general; bureaux de finances; collecteurs des tailles; contrainte solidaire; controleurs du vingtieme; directeurs du vingtieme; recette generale; receveurs des tailles; direct taxation agencies, taille: tre- soriers de finances direct taxes existing direct taxes- miscellaneous; existing direct taxes- general; impots accessoires a la taille; capitation; deuxieme brevet; premier brevet; impots personnels: impots reels: taille; taille personnelle; taille reelle; taille tarifee; vingtieme des biens fonds; vingtieme de l’industrie; vingtiemes; vingtieme des charges et offices indirect taxes indirect tax agencies- misc; indirect tax agencies- general; commis, indirect tax agen- cies; collecteurs, indirect tax agencies; company of general farmers; indirect taxation agencies, droits de controle; indirect taxation agencies, , droits domaniaux: employes, indirect tax agencies; fermier, indirect tax agencies; fermes generales; gabellous; gre- niers a sel; existing indirect taxes- miscellaneous; existing indirect taxes- general: aides; centieme denier accessoires: cuir; droits de controle; droits domaniaux; droits sur la fabrication; droits joints aux aides; droits d’entree et de sortie; fer; gabelle: huiles; insinuation; octrois des villes; centieme denier des offices: sel d’impot new taxes new taxes- miscellaneous; new taxes- general; dime royale; impot territorial: impot unique: luxe tax administration tax administration- miscellaneous; tax administration- general: perception, collection of taxes; repartition of taxes; rolls, tax administration Notes: Source and typology from Shapiro et al.(1998).

26 Table A.5: Seigneurial regime grievance subjects: coding levels 3 and 4

Level 3 Level 4 misc misc/general; land improvements banalites misc/general; four; moulin; pressoir protection rights misc/general; cens en commande; fouage; le guet et la garde; taille seigneuriale symbolic deference misc/general; droits honorifiques; droit de port d’armes; droits de requisition; credit patterns force feudistes, terriers feudistes terriers commissaires a terriers fairs and markets misc/general; redevances seigneurs hauts jus- misc/general; desherence; foundling seigneur’s obligations ticiers labor services labor services periodic rental dues misc/general; cens; champart; cens et rentes; rente fonciere; rente seigneuriale; so- lidite dues on property trans- misc/general; lods et ventes; rachat; retrait feodal fers recognition of la directe misc/general; aveu et denombrement; commise saisie; foi et hommage recreational privileges misc/general; chasse; colombiers; garennes; droit de peche seigneurial agents seigneurial agents serfdom misc/general; formariage; mainmorte; poursuite seigneurial monopolies misc/general; ban de fauchaison; ban de moisson; ban vin; ban de vendange tolls tolls seigneurial courts misc/general; efficiency, effectiveness; procureur fiscal; qualified judges Notes: Source and typology from Shapiro et al.(1998).

27 Table A.6: Summary statistics

Panel A: Third Estate

Pays d’´election Pays d’´etat

Statistic N Mean St. Dev. Min Max N Mean St. Dev. Min Max

N grievances 89 246.539 131.536 42 702 53 248.302 126.888 45 730 N grievance subjects 89 148.27 54.894 26 272 53 146.491 52.572 34 300 Government 89 0.33 0.067 0.137 0.517 53 0.324 0.073 0.215 0.639 Constitution 89 0.093 0.048 0.007 0.357 53 0.087 0.041 0.024 0.224 Economy 89 0.133 0.054 0 0.321 53 0.146 0.059 0.02 0.31 Justice 89 0.211 0.062 0.08 0.431 53 0.211 0.066 0.079 0.422 Religion 89 0.092 0.04 0 0.222 53 0.096 0.038 0 0.179 Society 89 0.087 0.039 0 0.199 53 0.083 0.038 0 0.175 Royal taxation 89 0.16 0.048 0.052 0.326 53 0.138 0.04 0.061 0.227 Seigneurial regime 89 0.068 0.033 0 0.135 53 0.069 0.036 0 0.144 Panel B: Nobility

Pays d’´election Pays d’´etat

Statistic N Mean St. Dev. Min Max N Mean St. Dev. Min Max

N grievances 87 171.023 79.494 35 439 32 144.438 77.194 17 329 N grievance subjects 87 102.931 35.363 26 183 32 91 40.098 7 180 Government 87 0.397 0.077 0.23 0.594 32 0.36 0.058 0.188 0.483 Constitution 87 0.157 0.063 0.012 0.343 32 0.182 0.089 0.042 0.471 Economy 87 0.082 0.054 0 0.379 32 0.062 0.042 0 0.144 Justice 87 0.193 0.059 0.084 0.319 32 0.202 0.071 0 0.396 Religion 87 0.056 0.044 0 0.196 32 0.081 0.064 0 0.252 Society 87 0.061 0.036 0 0.182 32 0.056 0.032 0 0.135 Royal taxation 87 0.156 0.063 0.04 0.411 32 0.117 0.05 0 0.219 Seigneurial regime 87 0.022 0.024 0 0.112 32 0.019 0.019 0 0.07 Panel C: Villages

Pays d’´election Pays d’´etat

Statistic N Mean St. Dev. Min Max N Mean St. Dev. Min Max

N grievances 424 35.354 23.868 2 172 106 44.302 26.663 10 125 N grievance subjects 424 25.608 14.688 1 105 106 31.472 16.263 9 82 Government 424 0.465 0.156 0 1 106 0.418 0.163 0 0.8 Constitution 424 0.039 0.06 0 0.464 106 0.051 0.065 0 0.3 Economy 424 0.165 0.121 0 0.781 106 0.17 0.118 0 0.615 Justice 424 0.127 0.101 0 0.5 106 0.121 0.105 0 0.4 Religion 424 0.084 0.093 0 0.595 106 0.086 0.084 0 0.375 Society 424 0.071 0.081 0 0.5 106 0.112 0.096 0 0.404 Royal taxation 424 0.353 0.15 0 1 106 0.236 0.117 0 0.588 Seigneurial regime 424 0.07 0.082 0 0.5 106 0.107 0.097 0 0.39 Notes: “Government” through “Society” variables correspond to the level one subjects in Markoff & Shapiro’s classification of grievances.

28 Table A.7: OLS: Consent and share of grievances about royal taxation

Dependent variable: share of grievances about royal taxation

Third Estate Nobility

(1) (2) (3) (4) (5) (6)

consent −0.049∗∗∗ −0.067∗∗∗ 0.003 −0.039∗∗∗ −0.042∗∗∗ −0.039∗ (0.019) (0.016) (0.051) (0.010) (0.010) (0.022)

mean dep var 0.31 0.309 0.32 0.149 0.152 0.153 N 153 132 121 127 111 111 N cluster 30 27 25 30 27 25 Consent provinces All Weak TE Strong TE All Weak TE Strong TE

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors in parentheses, clustered at the g´en´eralit´elevel for TE and N. TE: Third Estate; N: Nobility

Table A.8: Consent and living standards in the short and long run

Ancien R´egime post-1789

soldiers’ age at age at % stunted mortality % urban % urban pop height death death conscripts growth (1) (2) (3) (4) (5) (6) (7) (8)

consent −0.056∗ −0.101 −0.253∗ 0.207 −0.181 0.026 0.103 0.027 (0.030) (0.129) (0.137) (0.166) (0.269) (0.236) (0.264) (0.087)

Timeframe 1716-1784 1740-1790 1800-1829 1818-1830 1836 1800 1851 1800-1851 Geographic commune village village canton arr. dep. dep. commune unit Observations 21,896 1,424 1,150 2,289 224 77 68 29,508 R2 0.037 0.170 0.150 0.310 0.190 0.515 0.537 0.016 Adj. R2 0.036 0.159 0.137 0.306 0.131 0.396 0.404 0.016

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. (1) Soldiers’ height (in mm) data from Komlos, Hau, and Bourguinat(2003). (2) and (3) Civil registry data on a 1/100 sample of rural communes between 1740 and 1829, amounting to 358 communes in the territory of interest, from S´eguy(2001). (4) Cantonal data on conscripts height from Demonet, Dumont, and Le Roy Ladurie (1976). (5-7) Census data. (8) commune-level population data from Motte and Vouloir(2007). Models (1-3) include decade fixed effects. Standards errors are clustered at the g´en´eralit´elevel.

29 Table A.9: OLS: Consent and type of rebellion (1700-1788)

Rebellions 1700-1788

Share against taxes Share against local elites

(1) (2) (3) (4) (5) (6) consent −0.161∗∗∗ −0.156∗∗∗ −0.127∗∗ 0.092∗∗∗ 0.118∗∗∗ 0.033 (0.045) (0.050) (0.063) (0.016) (0.013) (0.034)

mean dep var 0.432 0.481 0.433 0.107 0.091 0.097 N 3209 2727 2720 3209 2727 2720 N cluster 31 28 28 31 28 28 Consent provinces included All Weak TE Strong TE All Weak TE Strong TE

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors in parentheses, clustered at the g´en´eralit´elevel. Rebellion data and typology from Nicolas(2008).

Table A.10: OLS: village grievances

Dependent variable: share of grievances

Taxation Seigneurial regime

(1) (2) (3) (4) (5) (6)

consent −0.066∗ −0.064∗ −0.046 0.045∗ 0.060∗∗ −0.007 (0.034) (0.037) (0.043) (0.025) (0.030) (0.028)

mean dep var 0.332 0.338 0.348 0.078 0.079 0.07 N 504 474 428 504 474 428 N cluster 19 17 17 19 17 17 Consent provinces included All Weak TE Strong TE All Weak TE Strong TE

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors in parentheses, clustered at the g´en´eralit´elevel. TE: Third Estate. Strong/weak refers to the TE having veto power over decisions in the provincial estates.

30 Table A.11: OLS: Third Estate grievances about the seigneurial system

Dependent variable: share of grievances

Third Estate Villages

(1) (2) (3) (4) (5) (6)

consent 0.019 0.031∗∗ −0.009 0.045∗ 0.060∗∗ −0.007 (0.014) (0.014) (0.027) (0.025) (0.030) (0.028)

mean dep var 0.133 0.138 0.129 0.078 0.079 0.07 N 154 133 123 504 474 428 N cluster 30 27 26 19 17 17 Consent provinces All Weak TE Strong TE All Weak TE Strong TE

Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors in parentheses, clustered at the g´en´eralit´elevel. Pays d’´etat with weak Third Estate denote provinces with estates that voted by order. .

31 B Appendix Figures

Figure B.1: Pays d’´etatand Pays d’´elections

No consent Weak consent up to mid−15th c.

Consent up to 17th c. BOULONNAIS FLANDRE Consent Late territorial additions 17−18th c. ARTOIS HAINAUT CAMBRESIS

PICARDIE

NORMANDIE

ILE DE FRANCE

LORRAINE CHAMPAGNE

ALSACE MAINE BRETAGNE

ORLEANAIS

ANJOU

FRANCHE COMTE TOURAINE NIVERNAIS BOURGOGNE SAUMUROIS BERRI

POITOU BOURBONNAIS GEX AUNIS BRESSE MARCHE BEAUJOLAIS BUGEY

LYONNAIS SAINTONGE FOREZ

LIMOUSIN AUVERGNE

DAUPHINE

GUYENNE

VENAISSIN GASCOGNE LANGUEDOC PROVENCE

LABOURD BASSE−NAVARREBEARN SOULE BIGORRE QUATRE−VALLEES COUSERANSCOMTE DE FOIX

ROUSSILLON

Notes: Consent = pays d’´etat;No consent + former pays d’´etat= pays d’´electionin 1789. Late territorial additions are not included in the analyses. Provincial estates categorization based on Major(1980).

Figure B.2: Examples of coding of taxation grievances

Source: Revolutionary Demands, Shapiro et al.(1998)

32 Figure B.3: Balance checks

wheat ●

rugged ●

dist. Paris ●

dist. river ●

dist. sea ●

dist. border ●

dist city ●

N cities ●

dist. intendance ●

N grievances ●

N subjects 1 ●

N subjects 2 ●

N subjects 3 ●

N subjects 4 ●

Noble age at death ●

−0.5 0.0 0.5

Notes: Points represent the estimated effect of consent for each dependent variable. The unit of analysis for “wheat suitability”-“distance from intendance capital” variables is 1793 commune locations. The unit of analysis for 1789 grievance variables is bailliage district locations. The unit of analysis for “Noble age at death” is the location of death for 8,782 members of the nobility. All models include latitude/longitude linear interaction and geographic controls. 95% confidence interval based on robust standard errors clustered at the g´en´eralit´elevel.

33 Figure B.4: RD: grievances about taxation

Third Estate Nobility

taxation ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● taxation ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2 1 1

indirect indirect ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2 1 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● direct ● ● ● ● ● ● ● ● ● ● direct ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2 1 1

● ● ● ● ● ● ● ● ● ● ● taille ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● taille ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2 1 1

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● universal ● ● ● ● ● ● ● ● ● ● universal ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2

1 ● ● ● ● ● ● ● 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● tax collection ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● tax collection ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2 1 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● tax distribution ● ● ● ● ● ● ● ● ● ● ● ● ● ● tax distribution ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50

xs2 xs2 1 1 ● ● ● ● ● ● ● ● ● ● ● ● tax privilege ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● tax privilege ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

20 25 30 35 40 45 50 20 25 30 35 40 45 50 Notes: The black points are the OLS estimates of the effect of pays d’´etaton eight dependent variables. The white points are RD estimates of the same relationship on 10 to 50 km bandwidths. 95% confidence interval, based on standard errors clustered at the g´en´eralit´elevel. Indirect and direct denote the share of grievances mentioning respectively indirect and direct taxation. Universal denotes two universal taxes which target both privileged and non-privileged individuals: capitation and vingti`eme. Taille denotes the main direct taxes, which fell almost exclusively on the peasantry.

34 Figure B.5: Robustness checks: drop border sections

Taxation Seigneurial regime 1 1

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● drop Brittany 0 drop Brittany 0

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

10 20 30 40 50 10 20 30 40 50

xs xs 1 1

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● drop Northern 0 drop Northern ● ● ● ● ● ● ● 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

10 20 30 40 50 10 20 30 40 50

xs xs 1 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

drop Burgundy 0 drop Burgundy 0

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1

10 20 30 40 50 10 20 30 40 50

xs xs

1 ● ● ● ● ● ● ● ● 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

drop Southern 0 drop Southern 0

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

10 20 30 40 50 10 20 30 40 50

xs xs 1 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

drop Pyrenean 0 drop Pyrenean 0

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −1 −1 ● ● ● ● ● ● ● ●

10 20 30 40 50 10 20 30 40 50

Notes: Points represent the estimated effect of consent for RD models in a bandwidth of 10 to 50 km bandwidth. All models include latitude/longitude linear interaction and geographic controls. 95% confidence interval based on robust standard errors clustered at the g´en´eralit´elevel.

35 Figure B.6: Regression discontinuity plots

No controls No controls

●

0.5 ● ● ● ● 0.15

● ● ● ● ● ● ● ●

● ● ● 0.4 ● ● ● ● ●

● ● ● ● 0.10 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.3 ● ● ● ● ● % grievances about taxation % grievances ● ● ●

● about seigneurial regime % grievances 0.05

● ● 0.2 ●

0.00 ●

0.1 −60 −30 0 30 60 −60 −30 0 30 60 distance to pays d'etat distance to pays d'etat Controls Controls

●

0.5 ● ● ● ● 0.15

● ● ● ● ● ● ● ●

● ● ● 0.4 ● ● ● ● ●

● ● ● ● 0.10 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.3 ● ● ● ● ● % grievances about taxation % grievances ● ● ●

● about seigneurial regime % grievances 0.05

● ● 0.2 ●

0.00 ●

0.1 −60 −30 0 30 60 −60 −30 0 30 60 distance to pays d'etat distance to pays d'etat

36 Figure B.7: Social conflict RD: robustness check

Exclude border: 1 Brittany North Bourgogne Languedoc/Provence SouthWest

Share rebellions against taxes 0 −1

10 20 30 40 50

xs 1

Share rebellions

against taxes, 0 salt tax excluded −1

10 20 30 40 50

xs 1

Share rebellions against local elites 0 −1

10 20 30 40 50

Notes: Lines represent the estimated effect of consent for RD models in a bandwidth of 10 to 50 km bandwidth (1 km increments). All models include latitude/longitude linear interaction and geographic controls. 95% confidence interval based on robust standard errors clustered at the g´en´eralit´elevel. Each color corresponds to one excluded segment of the border between provinces with and without consent.

37