The Hinges of History: State-Making and Revolt in Early Modern France Author(s): Edgar Kiser and April Linton Source: American Sociological Review, Vol. 67, No. 6 (Dec., 2002), pp. 889-910 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/3088975 Accessed: 27/05/2010 08:05

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=asa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review.

http://www.jstor.org THE HINGES OF HISTORY: STATEoMAKING AND REVOLT IN EARLY MODERN FRANCE

EDGAR KISER APRIL LINTON University of Washington Princeton University

The formation of early modern states-combining increases in taxation, warfare, and administrative centralization-was often violently opposed by subjects. A game theoretic model of strategic interactions between rulers and subjects is developed to more fully specify the relationship between state-making and revolt in France be- tween 1515 and 1789. Quantitative analyses of revolts throughout France are com- bined with a brief case study of revolts in Guyenne (the most quarrelsome French province) to test propositions derived from the model. Offensive war (but not war in general) and administrative centralization increased the likelihood of revolt, while the presence of the Estates General (the national legislative assembly) decreased it. The effect of taxation on revolt depended on the historical context-high taxes and offensive wars tended to incite revolts prior to the Fronde (a cluster of revolts in 1648-1653), but had no effect afterwards. It is argued that the outcomes of signifi- cant revolts, like the Fronde, often serve as important turning points in the history of state-making. Potentially similar events in medieval England and early modern Spain are also discussed.

HE PROCESS of early modem state- to explore the causal relationship between Imaking was often quite unpopular; it various aspects of state-making and tax re- was marked by frequent wars, increasing volts.' Both state-making and revolt are the taxation, and the growth and centralization result of choices, and these choices are al- of administrative institutions. As several ways made in light of the anticipated reac- scholars have shown, these factors often tions of other actors. Rulers consider the triggered revolts in early modern states possibility of revolt when they think about (Barkey 1991; Berce 1987; Bonney raising taxes or pursuing unpopularpolicies. 1978:214; Brustein and Levi 1987; Clark In deciding whether to revolt, subjects con- 1978; Goldstone 1991; Hopcroft 1999; sider both the likelihood of rulers using re- Kiser, Drass, and Brustein 1993; Major pression and their other options for decreas- 1960:127; Markoff 1996:12, 42; Mousnier ing their tax burden, such as tax evasion. The 1970, 1979:741-42; Tilly 1975, 1986:7, determinants of the costs of tax evasion, in 1993:17, 156-58; Zagorin 1982). part a function of rulers' decisions about ad- We use a model of the strategic interac- ministrative techniques, must thus also be tions between rulers and taxpaying subjects part of a model of revolt. To provide a rough heuristic model of the Direct correspondence to Edgar Kiser, Depart- relationship between state-making and re- ment of Sociology, Box 353340, University of volt, we use a game theoretic decision tree. Washington, Seattle WA 98195 (kiser@u. washington.edu). We thank Julia Adams, Terry 1 Revolts were caused by many factors other Boswell, Julie Brines, Cai Yong, Jerry Herting, than taxation, including religious conflicts, popu- Peter Hoff, Hyojoung Kim, James Kitts, Becky lation growth, price increases, elite conflicts, and Pettit, Kevin Quinn, Ron Schoenberg, Katherine the rise of capitalism (Barkey 1991; Goldstone Stovel, and several anonymous ASR reviewers 1991; Lachmann 1989, 2000). We focus only on for helpful comments. revolts against state taxation.

AMERICAN SOCIOLOGICAL REVIEW, 2002, VOL. 67 (DECEMBER:889-9 10) 889 890 AMERICAN SOCIOLOGICAL REVIEW

The extensive form of game theory high- province in which revolt occurred most fre- lights the sequential nature of interaction, quently in early modem France-almost one- and thus precisely specifies the relationships third of the revolts in France occurred there. between various aspects of state-making Guyenne is also a useful case because it al- (war, taxation, and administrativecentraliza- lows us to explore in more detail the impor- tion) and revolt. We use the model in a lim- tant fiscal and administrative changes in the ited way: We look at the state-making pro- Pays d'Etat (provinces retaining provincial cess from the point of view of subjects. Ex- legislative assemblies in the sixteenth cen- plaining the actions of rulers (such as tax in- tury) in southern France in the early seven- creases and war-making) that initiate the teenth century. Because it is not possible to process is beyond the scope of this paper.2 get quantitative data at the provincial level We test the relationship between aspects for our entire period (especially for tax rates, of state-making and revolt mainly via quan- prices, and population), we do not reproduce titative analyses of revolts in France be- our statistical analysis for Guyenne. How- tween 1515 and 1789. Early modern France ever, we do have sufficient data to roughly is a useful case in which to explore these re- test whether the same causal dynamics oper- lationships for several reasons. First, many ate at national and provincial levels. of the classic historical studies of state-mak- ing and revolt cited above focus on this case, STATE-MAKING AND REVOLT AS allowing us to draw on their insights in con- STRATEGIC INTERACTION structing our models and to compare their qualitative findings with our more quantita- We begin with two simple assumptions about tive results. Second, early modern France the interests of actors.5The rulers (i.e., chief contains ample variation in the main aspects executives) of states want to maximize tax of state-making (levels of taxation, war, ad- revenue while maintaining security of rule ministrative centralization) and tax revolts, (Kiser 1994; Levi 1988; North 1981). Tax- facilitating the testing of our arguments. payers collectively want public goods like Because tax rates and rates of revolt security and social order that require taxa- (among other things) varied substantially tion, but individually they want to minimize across provinces (Brustein and Levi 1987; their tax burden.6 Using these assumptions, Collins 1994:20; Major 1994:xviii; Markoff and information and argumentsfrom the his- 1996:37; Mousnier 1979:641),3 it is impor- torical literatureto specify the structuralcon- tant to supplement our analysis with an ex- ditions within which these actors were em- amination of these relationships at the pro- bedded, we first outline propositions about vincial level.4 We focus on Guyenne, the how taxes, war, and centralization affected revolt in France, and then we construct a 2An adequate test of the game-theoretic model model showing how these factors are interre- as a whole would require both endogenizing the lated. choices of rulers by explaining both tax rates and policy choices, and using comparative cases to show the conditions under which state-making TAXATION AND REVOLT goes down paths not followed in the French case. Several scholars have that there is a 3Although aggregated data on French taxation argued have been used as a measure of the tax burden in relationship between levels of taxation and several exemplary historical analyses (Braudel 1972:684; Goldstone 1991:196-212; Mann 5 The following discussion, based on these as- 1993:358-400; Skocpol 1979; Tilly 1986:61-63; sumptions, outlines a stylized model of strategic Wallerstein 1989), all of these scholars recognize interaction, and should not be taken as a histori- that provincial variations across early modern cal description. France make incomplete any test that aggregates 6 It is, of course, a simplification to group all across them. taxpayers together, but the fact that most early 4 Our intent is not to reproduce Brustein and modern tax revolts, in France and elsewhere, Levi's (1987) masterful analysis of the geogra- were class collaboration revolts involving a phy of revolt-an analysis of one province would broad cross-section of subjects (Barkey 1991; not allow us to do that. We supplement their Brustein 1985; Mousnier 1970:42) makes it a work by focusing on the timing of revolt. reasonable simplification. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 891 revolt frequency in early modern France but they realize that doing so can sometimes (Barkey 1991; Berce 1987; Hopcroft 1999; increase the likelihood of revolt. The lower Mousnier 1979; Tilly 1986; Zagorin 1982). the benefits to subjects provided by the cur- Our model suggests that this relationship is rent policy pursued by rulers (such as an un- produced by two micro-level causal mecha- popular war), the greater the potential ben- nisms. Rulers will try to get as much tax rev- efit of a successful revolt that would alter or enue as they can without inciting revolts (be- terminate that policy. cause revolts threaten their security and are This allows us to specify the relationship costly to suppress), and they will attempt to between war and revolt more precisely-it find nontax sources of revenue that will be should depend on the type of war.8 Levi less likely to instigate revolts. If rulers had (1988, 1997) argues that citizens often view perfect information about the probability of defensive wars as public goods from which revolt at various levels of taxation, and if the they would benefit and thus for which they costs of revolt were always greater than the are willing to pay taxes, while they do not benefits of tax increases, rulers would always generally support offensive wars. Hopcroft set taxes at the highest level that would not (1999:82) suggests that this was true in early incite revolt and we would find no correla- modem Europe. Although it is notoriously tion between levels of taxation and revolt. difficult to find reliable data on the percep- First, however, rulers did not have perfect tions of historical actors (Tilly 1975:9; We- information, and thus would be expected to ber 1968:5-6), there is a great deal of evi- miscalculate in setting tax rates, thus some- dence that subjects thought that offensive times triggering revolts.7 Second, even when wars were less beneficial and legitimate than rulers did have good information, there were defensive wars. Offensive wars were often some instances in which the gains from rais- seen as enriching monarchs, but not their ing taxes or the costs of not doing so out- subjects, and were thus more difficult to jus- weighed the costs of revolts, and rulers chose tify (Klaits 1976:87-88; LeRoy Ladurie to increase taxes even if they thought revolt 1980:200; Rowen 1980; Seyssel [1519] was a likely result. The most common early 1981:130-31). The influential sixteenth cen- modem example is from international rela- tury political philosopher, Claude de Seyssel tions: Rulers could have an opportunity to ([1519]1981:130-31), argued that defensive begin or continue an unusually profitable of- wars were generally more legitimate than fensive war, or fear a very costly invasion. offensive wars and that the latter should only Thus, for both of these reasons, we expect a be initiated under certain limited circum- positive relationship between levels of taxa- stances. Klaits (1976:87-88) noted that of- tion and the frequency of revolts. 8There are two main bodies of work on the re- lationship between war and revolt-one is mainly WAR AND REVOLT quantitative and focuses on the nineteenth and Our primary argument about the effects of twentieth centuries, and the second includes war on revolt is analogous to our argument mainly qualitative historical analyses of the early modern era. In the quantitative literatureon mod- about tax increases. Rulers want to pursue ern states (see Stohl [1980] for a good summary), unpopularpolicies (those that serve their in- findings range from positive to negative relation- terests more than their subjects' interests), ships, with the majority of studies showing no strong relationship between the two. In contrast, 7 Rulers would have to know the extent of the almost all qualitative historical studies of early costs imposed on subjects by taxation (which will modern states suggest that war was a primary vary with population changes and the success of cause of revolts (Barkey 1991:703; Brustein harvests), their military power relative to sub- 1985; Brustein and Levi 1987; Le Roy Ladurie jects, and the collective action capacity of sub- 1980:292; Lublinskaya 1968:331; Pillorget 1975: jects. Of course, as with rulers' decisions about 425-36; Tilly 1975, 1985, 1990, 1993). This increasing taxes and risking revolt, subjects' cal- raises an obvious question: Are the different find- culations that revolt will decrease taxes are made ings in the two literatures due to differences in with incomplete information, and they are thus the time periods studied or differences in the expected to miscalculate some proportion of the methods used? A quantitative analysis of an early time. modern state should help us find the answer. 892 AMERICAN SOCIOLOGICAL REVIEW fensive wars had "ratherdubious legal foun- almost always be preferable to revolt be- dations" and were thus frequently criticized cause it entails less cost and risk, does not in political pamphlets. Monarchs also seem require collective action, and provides clear to have been aware of this difference. In a tangible benefits in the form of lower tax discussion of offensive wars in his memoirs, payments. Therefore, subjects are expected Louis XIV ([1661-1668]1970) notes that he to revolt more often when the costs of tax fears "being accused of recklessness when he evasion increase. The costs of tax evasion rushes voluntarily into complications" (p. depend primarily on the effectiveness of rul- 123), although this fear was clearly not suffi- ers' tax administration apparatus. This is cient to stop him. We thus expect the rela- why both historians and sociologists argue tionship between war and revolt to be stron- that administrative centralization generated ger for offensive wars than for defensive revolts (Barkey 1991:700, 708-09; Bonney wars.9 1978:214; Collins 1995:52; Mousnier 1970: 34, 38; Parker 1983:95; Westrich 1972:134; Zagorin 1982:215, 219). Because adminis- ADMINISTRATIVE CENTRALIZATION trative centralization increases the effective- AND REVOLT ness of assessment and collection, and thus Subjects will choose to pay their full tax bill also increases the costs of evasion only when the options available to them to (Matthews 1958:47; Mousnier 1979:611; reduce or eliminate tax payments are more Pillorget 1975:507-25), the greater the cen- costly. Revolt is certainly the most risky and tralization of tax administration, the greater potentially the most costly strategy for sub- the expected frequency of revolt. jects who want to decrease their tax burden. Therefore, subjects are expected to revolt THE HINGES OF HISTORY: TURNING only when the costs of their other options are POINTS IN STATE-MAKING AND REVOLT prohibitively high. The most importantalter- native available to subjects is tax evasion. Neither rulers nor subjects can be expected Subjects can attempt to evade taxes, either to have perfect information about the rela- by concealing taxable assets or by colluding tionship between tax levels and revolt, so with tax collectors (e.g., offering bribes for many revolts will occur as a result of rulers' lower tax assessments).10 Tax evasion will or subjects' miscalculations (see Gartzke 1999 for a similar argumentabout the causes 9 We also test an alternative argument about of war). It is difficult to predict when such the relationship between war and revolt. States miscalculations will occur in historical are often weak during wars, and thus more likely analyses because it would require detailed to be targets of revolt (Beik 1985:186, 196-97; knowledge of the information available to Kiser et al. 1993; Tilly 1990:186, 1993:12). Sub- rulers at the time. However, we can predict jects will only revolt when they think the prob- that rulers will learn from past miscalcula- ability of the ruler capitulating or unsuccessfully tions, especially large miscalculations that using repression is greater than the probability of had clear outcomes (Lublinskaya 1968:256). rulers successfully repressing the revolt. This ar- These events usually come in the form of gument stresses the opportunities that war pro- vides to potential rebels by temporarily shifting major historical turning points (the "hinges" the balance of power in their favor, thus increas- of history) in which the outcomes of revolts ing the probability of a successful revolt. Like demonstrate to rulers (and subjects) either social movements (Tarrow 1998), revolts are thus that future tax increases will not be benefi- expected to respond to shifts in political oppor- cial for rulers (clearly successful revolts) or tunity structures.The best indication of weakness that revolting against taxes is likely to be an is when a ruler's military resources are occupied unsuccessful strategy in the future (clearly elsewhere (e.g., in wartime). If this argument is unsuccessful revolts). correct, revolts will be more frequent during pe- riods of warfare. l0 Another option is to exit the area being modern period it was very costly for most sub- taxed, but this is prohibitively costly for subjects jects, as 90 percent or more derived most of their not living close to the border of another state wealth from land. Exit, thus, was not a viable al- (North and Thomas 1973). In fact, in the early ternative to revolt. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 893

Recent developments in historical meth- fensive war). Note that the "increase taxes" ods suggest that these turning points, criti- move also includes offensive war. We view cal junctures, or transformative events are these two moves as analytically similar-the important determinants of historical out- tax rate specifies the costs to subjects, and comes and are too often ignored in main- the type of policy pursued determines the stream quantitative sociology (Abbott 1997; benefits. Throughout the game, subjects Collier and Collier 1991; Goldstone 1998; have three choices: pay, evade, or revolt Issac and Griffin 1989; Mahoney 2000; (they are not entirely mutually exclusive, as Peirson 2000; Sewell 1996). Although re- subjects will generally choose some mix of searchers may use many different terms, all pay and evade). of their arguments suggest that key events If the ruler does not increase taxes, sub- can dramatically shift the trajectory of his- jects will continue to choose some mix of torical development. They also agree that pay and evade (depending on the costs of these events can rarely, if ever, be predicted, evasion), just as they did in prior years. but can be recognized by their consequences (This assumes the conditions affecting the after the fact. Our argument focuses on the tax collection process do not change.) We microfoundations of historical turning call this situation a "customary taxation points. Historical turning points are impor- equilibrium." tant because they are likely to alter the struc- The more interesting developments follow tural factors affecting the cost-benefit calcu- from rulers' choices to increase taxes and lations of actors, and thus (in this case) also pursue offensive wars, and these choices affect the relationship between state-making were the most common in early modern and revolt. France. Subjects have the same three alter- Focusing on potential turning points is one natives, but the payoffs for each are now way to make quantitativesociology more his- quite different. The main effect of rulers torical (Issac and Griffin 1989). To begin we choosing this strategy is to increase the costs must first turn to the historical literature to of paying taxes or to decrease the benefits identify them. Historians agree that there was of the policies they fund. Therefore, both such a turning point in early modern evasion and revolt become relatively more France-the relationship between state-mak- attractive.1 Subjects are expected to choose ing and revolts changed following the col- evasion more often (because it is less costly lection of related revolts called the Fronde and less risky than revolt), but the frequency (1648-1653) (Collins 1995:77-78; Major of revolt is expected to increase somewhat. 1994:xxi; Parker 1983:119; Westrich 1972: The game then proceeds along two sepa- 140). As a result of both the information the rate lines, depending on whether subjects Fronde provided about the relative power of choose to evade taxes or revolt. If subjects rulers and subjects and the actions rulers took choose revolt, rulers then have to choose be- in its aftermath (elaborated below), subjects tween using repression or giving in to sub- viewed revolts as much more costly than be- jects' demands and lowering the tax rate. If fore. Therefore, after the Fronde, we expect they lower the tax rate, the game ends with the relationship between taxation and un- payoffs similar to the customary tax equilib- popular policies (such as offensive war) and rium. If they use repression, there is some revolt to be much weaker than before. uncertainty about the outcome-will they defeat the rebels or not? The role of chance

A MODEL OF STA TE-MAKING IISubjects may still choose to pay the tax, if AND REVOLT the costs of evasion and revolt are too high. This Figure 1 outlines the state-making and revolt leads to an upward spiral, as rulers again increase game, combining the factors we have dis- taxes. This is not a stable equilibrium. Eventu- cussed in a model of sequential, strategic ally, the costs of paying will be higher than ei- ther evading or revolting. Subjects will then decisions. Rulers move first and decide choose one of these two options, pushing them whether to keep the tax rate the same or in- down to the lower branches of the game tree, as crease it, and what policies to pursue (spe- discussed below (indicated by dotted lines in Fig- cifically, whether to initiate an unpopularof- ure 1). 894 AMERICAN SOCIOLOGICAL REVIEW

Co 0

co

CLw cr n. Era w

Co I~~~~~~~~~~~~~~~~~~~~C

+L -

U) ~~~~~. w ct ~ ~ L a 0~~~~~~~~~

U) k o w ~~~~~~~~~~L a -~~~~U)~ ~ L STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 895 is capturedin game theory as a move by "na- Table 1. Description of Variables Used in the ture"-either the repression is effective or it Analyses: Tax Revolt Initiations in is not. If it is not, we return to something France, 1515 to 1789 similar to the customary tax equilibrium. If Variable Mean S.D. Range it is successful, rulers are able to enforce the higher tax rate and their preferred policy Tax revolt initiation .11 0-1 choices. (1 = yes) If subjects respond to tax increases or un- Total tax per capita 5.06 4.46 .87-37.12 popular policies by increasing evasion, as (in 100,000 it.) we expect, rulers will attempt to increase the Before and during .51 0-1 costs of evasion by centralizing administra- the Fronde(1 = yes) tion. As with repression, there is an element Administrativecentral- .53 - 0-1 of uncertainty regarding the outcome: Cen- ization (1 = 1642 on) tralization may or may not be effective. This is again modeled as a move by "nature."If Estates General .37 - 0-1 (1 = yes) centralization is ineffective, the result is an outcome characterized by high tax rates and War in progress .65 0-1 high evasion, with a net effect similar to the (1 = yes) customary tax equilibrium. If centralization Defensive war in .34 0-1 is effective, subjects face high costs for both progress (1 = yes) paying and evading. This is the situation in Offensive war in .50 0-1 which their last-resort option, revolt, be- progress (1 = yes) comes most likely. Rulers respond to revolt either with re- pression or by conceding and lowering the ses and provincial case study provide tests tax rate. If they choose repression, it may or of many of the links in this causal chain.'2 may not be effective (a move by "nature"as above). If it is not effective, the result is an DATA AND MEASURES unstable outcome characterized by high tax rates and frequent revolts. If it is effective, Table 1 describes the variables correspond- subjects will be forced to pay the higher tax ing to the empirical expectations discussed rates and endure the unpopularpolicies cho- above, and Appendices A and B list the dates sen by rulers. of all revolts and wars. Measures of the in- We argue that this last outcome, a high dependent variables are lagged to assure the tax equilibrium, is what eventually hap- proper temporal order in our findings. pened in early modem France. Rulers often The descriptive statistics in Table 1 docu- chose to increase taxes and fight unpopular ment several interesting features of early offensive wars, especially in the seven- modem French history. The French state was teenth century. Subjects sometimes re- forged by force, both internally and exter- volted, but most often increased their tax nally. Revolts were relatively frequent evasion. Rulers responded by centralizing, events, taking place in about 20 percent of primarily with the use of intendants. This the years between 1515 and 1789. They were centralization was in part successful, and initiated in about 11 percent of these years when it increased the costs of evasion, re- (see Appendix A). Wars were in progress volts became much more common, culmi- during about 65 percent of the years during nating in the largest revolt of all, the this period. Offensive wars were somewhat Fronde. Rulers successfully used repression more common than defensive wars-France against this revolt (and co-optation after it), was involved in offensive wars in about 50 thus convincing subjects that future revolts percent of the years, and in defensive wars would be unlikely to succeed. The result in about 34 percent. was a high tax equilibrium after the Fronde, as French rulers were able to raise taxes and 12Unfortunately, we cannot test all of the links, fight offensive wars with little resistance due to lack of data. There are no systematic data from their subjects. Our quantitative analy- on tax evasion, for example. 896 AMERICAN SOCIOLOGICAL REVIEW

REVOLTS trapolatebased on the data at hand. We have done this by (1) subtracting nontax revenue Our dependent variable is "tax revolt initia- (revenue from crown lands and the sale of tion." Coded dichotomously, it marks the offices) from the total revenue figure for a beginning of one or more tax revolts in a given year, (2) aggregating data on all rev- particularyear.'3 The data are compiled from enues contributing to total tax, or (3) carry- Heller (1991), Kiser et al. (1993), and Tilly ing established figures from year to year. (1986, 1993). We include what Tilly (1993) Because we are interested in measuring calls "revolutionarysituations," in which the the real tax burden that French citizens bore, rebels held considerable instrumentsof state we express taxation in terms of livres power, as well as regional events that he des- tournois (1.t.) per capita.'6 Ideally, we would ignates as "serious tax rebellions." These in- include in our models a control for both real surrections "stemmed from essentially the wages and taxpayers' other main expenses same processes" (Tilly 1993:156).14 Both (such as rents and tithes), but we lack reli- shared the same objective-to lessen the tax able and comprehensive data on these fac- burden-and entailed similar risks. They of- tors during the early modem period.'7 ten cost thousands of lives.

HISTORICAL TURNING POINT: TAXATION THE FRONDE Due in part to fires in the treasury archives, The Fronde was a large-scale set of revolts and in part to poor record-keeping by the with a very clear outcome. The crown monarchy, data on taxation in early modem emerged victorious and the state subse- France are incomplete and sometimes unreli- able. Historians give very different estimates 16 The population of France increased by over of the tax revenues in a particular year-a 10 million during the time span considered here. "reason for considerable caution" (Rasler Dupaquier (1979) is our principal source of and Thompson 1989:133). Thus, it is essen- population data. We also use estimates by Bean tial to know the literaturewell so as to make (1973), Bonney (1979), and Wilson and Parker wise choices among conflicting figures. Our (1977). When yearly estimates were unavailable estimates of gross total taxation draw heavily or contradictory, we smoothed the data between benchmark figures, allocating an average rate of on research by Guery (1978), Hoffman population increase/decrease per year (cf. Gold- (1986, 1996), and Rasler and Thompson stone 1991). (1989).15 To arrive at a total tax amount for 17 To explore the possibility that tax revolts re- each year, it was sometimes necessary to ex- ally reflected a more general resentment toward rising costs or shortages, we tested models in- cluding (1) the price of wheat, and (2) a standard- 13 In some cases, more than one revolt started ized tax variable that reflects the amount of in the same year (see Appendix A). An analysis wheat, in setiers (a common measure of grain in which revolt initiation is coded as the number that ranged in volume from 150 to 300 liters), of revolts beginning in a given year yields results that could be purchased with the per capita tax that are substantively similar to the ones we re- revenue for a given year (Baulant 1968). The port. wheat price variable was never significant, even 141In earlier analyses, we found that coding re- with per capita tax absent from the model. The volt initiation according to revolt severity did not results of models with standardized tax are very meaningfully alter our findings. similar to those we report in Tables 1 and 2, al- 15 Direct taxation in early modern France took though overall model fit was poorer. We also in- two main forms: the taille, levied on property, vestigated the possibility that subjects responded and capitation, levied on individuals. Collins to nominal tax increases more than to the level (1988), Clamageran (1980), Lewis (1968), and of taxation (Goldstone 1991; Le Roy Ladurie Morineau (1968) offer considerable data on both 1980:263). The annual percent change in nomi- of these. Indirect taxes primarily include a vari- nal or standardized taxes was never significant. ety of tariffs such as the gabelle, a salt tax; aides, This finding, however, does not conclusively dis- a tax levied on the consumption of a variety of credit the proposition that tax increases prompted products; and traites, duties paid when merchan- revolts-to do so would require additional case dise was imported and/or sold. studies, preferably with better data. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 897 quently became stronger and more central- learned that revolt was likely to be unsuc- ized (Hoffman 1994:247; Mousnier 1979: cessful, so they shifted to more cautious and 571, 624). Tilly (1986) argues that the "mid- covert forms of disobedience (Mettam seventeenth century served as a hinge in the 1988:3-4, 11-12; Parker 1983:102-03).18 history of contention" (p. 40). Because the crown emerged victorious and In the aftermath of the state's "military the state subsequently became stronger and victory" in the Fronde (Mousnier 1979:624; more centralized, it is probable that the also see Lachmann and Adams 1988:154), Fronde altered the relationship between several changes increased the power of the state-making and revolt. We therefore in- state relative to potential rebels. After the clude a variable coded 1 for the years before Fronde, a new law was passed confiscating and during the Fronde (1515-1653), and 0 half of town revenues for the central state, for the years thereafter. and intendants took more direct control of town spending (Berce 1987:16-17; Doyle ADMINISTRATIVE CENTRALIZATION 1996:19; Mousnier 1979:572). Town walls also atrophied after the Fronde. Towns ex- Direct tax collection in early modern states panded; tree-lined streets replaced old was generally decentralized. In almost all moats; rings of boundary stones replaced states, taxes were assessed and collected by town walls (Berce 1987:54-55). Subjects officials elected or appointed and controlled lost power in judicial institutions that had locally, not by the central state. France was been used to spread information about crown no exception. Rulers generally negotiated policies contrary to their interests. Moote lump sum payments from regions and towns, (1971) argues that the limitation of the right and left it to the locals to collect these taxes of remonstrance in law courts (parlements) as they saw fit (Markoff 1996:237). Al- until after laws had been registered was "by though direct tax collection was never fully far the most successful aspect of the continu- centralized in early modern France (even the ing royalist revival after the Fronde" (pp. revolution did not totally change this), there 356-57). Localities and provinces lost an was a major turning point in the seventeenth important political tactic, substantial rev- century: the use of intendants (Kettering enue, and the ability to defend against royal 1986:6). Parker (1983) notes that intendants attacks. The development of a standing army were the "principal agent through which the also increased the costs of revolt by making concentration of authority was effected" (p. quick repression more likely (Ranum 81; also see Collins 1995:53). The inten- 1993:346). In these ways, the Fronde "extin- dants were part of a long list of royal at- guished local resistance to the fiscal offen- tempts to use ad hoc commissioned officials, sive of the crown" (Berce 1987:17; also see centrally appointed and controlled, to super- Parker 1983:119; Pillorget 1975:900-10). vise local tax collection (dating back to the The aftermath of the Fronde was marked medieval bailiffs). Although Richelieu used not just by repression but by concessions and intendants, their numbers and their powers co-optation as well (Lachmann and Adams were significantly increased in 1642, when 1988; Tilly 1986:40), and this too made re- they were given full authority to supervise volt less likely. Opposition to the state was tax collection (Bonney 1988:82). This was "subdued but not eradicated" after the the most important administrative transfor- Fronde (Moote 1971:355). Therefore, Louis mation in direct tax administration prior to XIV and subsequent rulers had to learn how the revolution, so we include a dummy vari- to compromise (Mettam 1988:3-4, 11-12). able for "administrative centralization" to They were more careful not to disrupt tradi- capture this shift. tional privileges, and they provided addi- tional lucrative state jobs to those elites 18 Tilly (1993) remarks whose support they required (Lachmann and that post-Fronde re- volts were different in substance from earlier up- Adams 1988:155-58). risings. Although nominally still tax revolts, they As our argument suggests, the one impor- were more importantly protests against "the ex- tant result of this clear state victory, coupled pense, arbitrariness, and corruption of the gov- with co-optation, is that potential rebels ernment" (p. 163). 898 AMERICAN SOCIOLOGICAL REVIEW

WARS Both of these possibilities, pushing in oppo- site directions, are consistent with the micro- Our war data come from Kiser et al. (1993, foundations of our model. We thus have no 1995), with additions for the eighteenth cen- clear prediction about the effects of repre- tury from Dupuy and Dupuy (1977) (see Ap- sentative institutions; it will depend on pendix B). The war variables are dichoto- which effect is stronger.We include both na- mous. A value of 1 indicates that the state tional and provincial representative institu- was involved in a conflict during a given tions as control factors. year; a value of 0 for war or for both offen- The broadest of these institutions was the sive and defensive war means that no war Estates General, a national legislative as- took place. This allows us to assess the gen- sembly comprised of representatives of the eral relationship between war and revolt ini- clergy, nobility, and the primarily bourgeois tiation, as well as to contrast the impact of "third estate." It met intermittently until defensive and offensive wars. 1614, consulting kings on importantmatters such as war and taxes, and presenting lists of grievances. We therefore include a NATIONAL AND PROVINCIAL dummy variable for the existence of the Es- REPRESENTATIVE INSTITUTIONS tates General, coded 1 prior to 1615, and 0 AS CONTROL VARIABLES after that. We also explore the role of the Subjects can attempt to negotiate with rulers provincial estates, even though we cannot collectively to lower their tax obligations or include them in our regression analyses. In to modify policies, usually throughrepresen- Table 4, we note, by region, whether a pro- tative political institutions. There are two vincial estate existed when revolts were ini- contrasting arguments about how the pres- tiated. We discuss the effects of provincial ence of representative institutions might af- estates along with our regression results. fect revolt. First, they were important sites for the exchange of informationbetween rul- ers and subjects (Mousnier 1979:215). By DATA ANALYSIS AND RESULTS increasing the flow of information between REVOLTS IN FRANCE: could decrease the level of un- parties, they REGRESSION ANALYSES certainty and thus decrease the frequency of revolt. Second, however, representative po- Some degree of autocorrelation is common litical institutions could enhance the collec- in time-series analysis. The logit regression tive action capacity of subjects and thus in- model does not provide for unmeasured fac- crease the frequency of revolt (Bonney tors affecting the occurrence of revolts, nor 1978:347-48; Brustein and Levi 1987).19 does it allow for possible serial correlation due to the dependence of the observations. 19Collective action capacity is important, but While this means that the estimates of the the main question involves the level at which coefficients are not efficient, they are con- these institutions are most effective. Is it the sistent (Robinson 1982). For statistical infer- country as a whole (Estates General), provinces ence we likeli- (provincial estates), parishes (ad hoc meetings of use QML (quasi-maximum groups of communities), or local communes? Un- hood) standard errors, which are consistent til its demise in 1614, the Estates General inter- under misspecification of the model mittently brought together representatives from all over France. It served as a viable forum for cally based institutions were more important de- negotiation about taxes (Mousnier 1979:215; terminants of collective action capacity. Berce Parker 1983:81), but was probably too large and (1987:75-76) stresses the importance of collec- heterogeneous to facilitate collective action, tive action at the parish level, with several com- which was almost never organized at the level of munes (usually led by local notables) assembling the country as a whole. Provincial estates, which to discuss shared interests, including possible re- existed (decreasingly over time) in some regions volts. Unfortunately, systematic data on the but not others, were forums for negotiation power and activities of these assemblies at the (Collins 1994:23; Hoffman 1994:230; Mousnier parish and commune level are not available for 1979:610), but may also have been too large to this period, so we cannot add this as a variable in facilitate collective action. We suspect that lo- our quantitative analyses. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 899

Table 2. Logit Coefficients from the Regression of Tax Revolt Initiation on Selected Independent Variables: France, 1515 to 1789

IndependentVariable Model 1 Model 2 Model 3 Model 4 Total tax per capita (in 100,000 i.t.) -.114* -.249 -.062 (lagged one year) (.049) (.275) (.134) Before and duringthe Fronde 1.296** -1.407 2.532** (.450) (.310) (.864)

Before and during the Fronde .710* x total tax per capita (.310) Administrative centralization 1.852* (lagged one year) (.845) Constant -1.161*** -2.914*** -1.107 -4.060*** (.266) (.388) (1.724) (.984) AIC 180.35 175.89 157.75 165.89 Note: N = 275 for each model. QML standarderrors are shown in parentheses. *p <.05 ** <.01 *** < .001 (two-tailed tests)

(Arminger 1995; Gourieroux, Monfort, and French history which changed the relation- Trognon 1984; White 1982).20 The AIC ship between state-making and revolt (Akaike information criterion) fit statistic (Collins 1995:77-78; Major 1994:xxi; provides an accounting of use of degrees of Mousnier 1979:571, 624). This is easy to see freedom, and is especially useful when com- graphically as well; although the Fronde di- paring nonnested models such as these. vides the period covered in our study almost Comparatively small values indicate an im- exactly in half, there were more than three provement in fit. times as many revolts prior to its conclusion The models in Table 2 test the degree to than after it. As we argued above, the deci- which the level of taxation, the Fronde, and sive defeat of the rebels signaled to all sub- administrative centralization influence the jects that revolts were unlikely to be suc- likelihood of revolt initiation. By itself, the cessful strategies for opposing taxation level of taxation is negatively related to re- (Berce 1987:17; Mettam 1988:3-4, 11-12; volt initiation (Model 1). Taxes were highest Pillorget 1975:900-910). After the Fronde, in the eighteenth century; revolts were least rulers in France were able to raise taxes frequent then. Models 2 and 3 incorporate a without fear of revolt. dummy variable for the period before and Model 4 incorporates administrative cen- during the Fronde-alone and in interaction tralization after 1642. Here it appears that with taxation. Model 3 specifically examines both the period before and during the Fronde the effect of the Fronde on the relationship and the years of direct tax centralization between taxation and revolt initiation. The were times in which revolts were likely to impact of taxation on the probability of re- begin. Ten revolt initiations-a third of the volt is positive before and during the Fronde, total-occurred in the years for which posi- whereas after the Fronde it is not significant. tive values for these variables overlap: 1643 This finding demonstrates quantitatively through 1653. In substantive terms, this ef- what many historians have argued-that the fect is explained by opposition to the use of Fronde was indeed a key turning point in intendants, central semibureaucratic royal agents that increasingly intervened in direct 20 We also ran a model of revoltinitiation that tax administration, eliminating some forms includeda laggeddependent variable (revolt ini- of local corruption that had benefited sub- tiationin the previousyear), and found very simi- jects at the expense of the monarch lar results. The lag did not contributesignifi- (Mousnier 1979:611; Pillorget 1975:507- cantlyto any of our models. 25). There is thus some support for the argu- 900 AMERICAN SOCIOLOGICAL REVIEW

Table 3. Logit Coefficients from the Regression of Tax Revolt Initiation on Selected Independent Variables: France, 1515 to 1789

IndependentVariable Model 1 Model 2 Model 3 Model 4 Model 5 Total tax per capita -.185 .022 -.013 -.081 -.23 (in 100,000 It.) (lagged one year) (.168) (.115) (.107) (.107) (.221) Before and duringthe Fronde 1.732* .168 1.487 -.853 1.531* (.768) (.963) (.814) (1.071) (.730) Estates General -2.497*** -2.233*** (.639) (.693) War in progress (lagged one year) 1.984** (.661) Defensive war in progress .647 .659 .198 (lagged one year) (.409) (.431) (.435) Offensive war in progress 1.378** -.541 1.131* (lagged one year) (.507) (.758) (.581) Before/during the Fronde 2.910** x offensive war in progress (1.006) Constant -1.516 -4.787*** -4.085*** -2.268** -2.109 (1.201) (1.113) (.999) (.962) (1.253) AIC 151.08 156.89 156.22 145.90 140.82 Note: N = 275 for each model. QML standarderrors are shown in parentheses. *p <.05 ** < .01 *** < .001 (two-tailed tests) ment that administrative centralization was Although we could not include provincial an additional aspect of state-making that estates in our regression analyses, we did stimulated revolts, but its effect is tempo- gather data to assess their effects on revolt rally limited. (see Table 4). Our data neither support our The models in Table 3 explore the effects argument that their presence decreased re- of representativeinstitutions and wars on the volts, nor Brustein and Levi's (1987) conten- probability of revolt. The odds of a revolt tion that they facilitated revolt. Rather, we initiation were almost 50 percent lower in find no significant difference in revolt ini- years when the Estates General convened. In tiation across provinces with and without Model 1, both a restraining effect of the Es- provincial estates, or across time in areas tates General and a positive before/during- where provincial estates met during only the-Fronde effect are prominent. Half of the some of the years from 1514 to 1789. Table revolt initiations that we consider occurred 4 shows that 17 tax revolts occurred in prov- between 1616 and 1653-years before or inces where and at times when provincial during the Fronde, but after the Estates Gen- estates met, whereas 18 took place where or eral ceased to meet. Including both "before when they did not. When we compare blocks and during the Fronde" and "Estates Gen- of total province-years in which provincial eral" markers in the same model (Model 1) estates met (3,544 years) and did not meet effectively shows that, although revolts were (7,681 years), we see that revolt initiations much more likely to take place before or dur- happened .46 percent of the time in the first ing the Fronde, this was not the case when a block of years and .23 percent of the time in central legislative body existed and met. This the second. Revolts thus seem to be slightly finding suggests that the Estates General more likely when provincial estates were worked not as a collective action mechanism present, thus providing a bit more support facilitating revolt, but as a forum for nego- for Brustein and Levi's view than for ours, tiation and information exchange that pro- but the difference is too small to be substan- vided a less costly alternative to revolt. tively significant. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 901

Table 4. Provincial Estates and Revolts by Province: France, 1514 to 1789

Revolt Initiations Numberof Years Numberof Years When Provincial Provincial ProvincialEstates When EstatesDid Province Estates EstatesMet Did Not Meet EstatesMet Not Meet Anjou No 0 275 0 1 Auvergne No 0 275 0 0 Bearne Yes 275 0 3 0 Berry No 0 275 0 0 Bigorre Yes until 1622 108 167 0 0 Boulonnais Yes 275 0 1 0 Bourbonnaise No 0 275 0 1 Bresse Yes 275 0 0 0 Brittany Yes 275 0 2 0 Burgundy Yes 275 0 0 0 Champagne No 0 275 0 0 Dauphind Yes until 1628 114 161 0 0 Foix Yes 225 0 0 0 Gavardan Yes until 1622 108 167 0 0 Guyenne Yes until 1622 108 167 5 7 he de France No 0 275 0 1 Ile de R6 No 0 275 0 0 Ile d'Oleron No 0 275 0 0 Labourd Yes until 1622 108 167 0 0 Languodoc Yes 275 0 4 0 Limosin No 0 275 0 1 Maine No 0 275 0 0 Marche No 0 275 0 0 Marsan Yes until 1622 108 167 0 0 Metz No 0 275 0 0 Navarre Yes 275 0 0 0 Nebouzan Yes until 1622 108 167 0 0 Nivernais No 0 275 0 0 Normandy Yes until 1655 141 134 1 0 Orleanais No 0 275 0 1 Picardy No 0 275 0 0 Poitou No 0 275 0 4 Provence Yes 275 0 1 0 Quatre-Vallees Yes until 1622 108 167 0 0 Saintonge No 0 275 0 2 Saumurois No 0 275 0 0 Soissonnais No 0 275 0 0 Soule Yes until 1622 108 167 0 0 Toul No 0 275 0 0 Touraine No 0 275 0 0 Verdun No 0 275 0 0 Total 3,544 7,681 17 18 902 AMERICAN SOCIOLOGICAL REVIEW

Getting back to Table 3, Models 2 and 3 Model 4 tests the proposition that the re- contrast two ways of measuring war.21 lationship between offensive wars-an un- Model 2 considers all wars together; Model popular policy-and revolt was weaker af- 3 separates offensive and defensive wars. ter the Fronde. Indeed, we find that offen- Although the general war variable has the sive wars affect the likelihood of revolt expected effect, comparison reveals that the only in the period before and during the type of war is of crucial importance: Offen- Fronde. sive wars were responsible for the positive Model 5 best fits the observed pattern of effect of war on revolt initiation. This test revolt initiations.23It accounts for the com- allows us to adjudicate between the three bined effects of taxation, the Fronde, the Es- possible causal mechanisms linking war to tates General, and wars. Here we see clear revolt. evidence of the multiple causal mechanisms It seems clear now that the relationship linking state-making and revolts. Tax revolts between war and revolt was not entirely due were not simply a response to high taxes; to the costs of war. Although we unfortu- they were facilitated or hindered by other nately have no data on the costs of each factors that are key to the state-making pro- war, there is no reason to expect offensive cess-notably offensive wars and the pres- wars to be significantly more expensive, ence or absence of representative institu- and this would be accounted for by the tax tions. variable in any case. Nor is the relationship due only to the weakening of the state due A PROVINCIALCASE STUDY: to war, because if anything, defensive wars REVOLTS IN GUYENNE might be expected to weaken the state more (Tilly 1990:186; 1993:12). The strong ef- Figure 2 shows the geographic distribution fect of offensive war in Model 3 suggests of tax revolts in France in three periods: the that subjects cared about how their tax reign of Francis I (1515-1548), the period money was spent. They were willing to pay between the death of Francis I and the be- for defensive wars that protected them from ginning of the Fronde (1549-1648), and the external threats, but not for offensive wars years during and after the Fronde (1649- from which they expected few if any ben- 1789). In the first period, all of the revolts efits and could pay very high costs occur in the south and southwest. The two (Hopcroft 1999; Klaits 1976:87-88; Rowen later periods show more geographical varia- 1980; Seyssel 1981:130-35). They seem to tion, although most revolts are still found in have thought that defensive wars were pub- the south and southwest. This pattern has lic goods, but that offensive wars served been documented and explained by Brustein narrower interests (Levi 1988, 1997). Of and Levi (1987) and Hopcroft (1999). Al- course, it would take a detailed cultural though our focus is on temporal variation, analysis of subjects' reactions to different the geographic concentration of revolts in types of wars to fully demonstrate this, and southern and southwestern France suggests that is beyond the scope of this paper.22 that it would be wise to look more closely at these areas to provide an additional and 21 The Frondevariable approaches significance more precise test of our arguments. in these models(p = .080 in Model2 and .068 in We selected the province with the highest Model 5). rate of revolt in France, Guyenne -12 of the 22 Buena de Mesquita, Silverson, and Woller 36 tax revolts in early modern France oc- (1992) report that the outcome of wars, in con- curred in this southwestern province. junction with who initiated them, has a strong ef- Guyenne was part of the dramatic change in fect on revolt. We tried a fourfold differentiation the status of the Pays d'Etat in the seven- like theirs: successful offensive wars, lost offen- teenth century, involving increases in taxa- lost defen- sive wars, successful defensive wars, tion and centralization, and diminishing or sive wars. We found, however, that whether an offensive (or defensive) war was won or lost did not change the relationships we report. Variables 23 The fit is significantlybetter than that of that reflect outcome were nonsignificant in our modelsin which tax centralizationis substituted models. for the EstatesGeneral variable. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 903

* ~~~~~~~~~*

f * f~~~

* 1514-1548

* *+ *i ** 1549-1648 * 1649-1789

* Figue 2 R* D

Figure 2. Tax Revolt Initiations in France, 1515 to 1789 A eliminating the provincial estates.24 Prior to about 1620 and 1670 (Collins 1988:105), in the seventeenth century, Pays d'Etat gener- what Major (1994) calls a shift from a "re- ally had more autonomy from the crown, naissance monarchy" to an "absolute mon- both legislatively (through provincial es- archy." tates) and administratively,which usually al- Tax increases were thus a key component lowed them to minimize their tax burden of this transformation. As Collins (1994) relative to the more centrally controlled Pays puts it, "[K]ings saw the Pays d'Etat as an d'IElection.25This began to change between untapped (or poorly tapped) financial re- source. After 1628, [Louis XIII] dramati- cally raised taxes in all of them" 24 (p. 152). Le Roy Ladurie (1980:201) argues that the Although precise data are not available, the general factors revolts were similar causing significant tax increases that hit all of the throughout southern France, and historians dis- cussing the southern provinces mention the same Pays d'Etat began in Guyenne in the 1620s factors we have discussed for France as a (Ranum 1993: 220). whole-taxes, war, centralization, the role of To increase taxes, the administrative and provincial estates, and changes brought about by legislative autonomy of the Pays d'Etat were the Fronde (Beik 1985; Collins 1994; Le Roy decreased, and in some cases terminated en- Ladurie 1980:198, 201-202, 286; Major 1966; tirely (Barkey 1991:702). The process of ad- Mousnier 1979:622-23; Westrich 1972; Wolff ministrative centralization (imposition of in- 1969:210, 230). tendants) was complicated in Guyenne. In- 25 The assault on the Pays d'Etat did not affect all provinces equally. Guyenne was a relatively rich province (as indicated by high levels of meat it an appealing place for monarchs to look for consumption [Ranum 1993:217-18]), known more revenue, Guyenne was hit especially hard, then and now primarily for the export of great losing its Pays d'Etat status (Languedoc, for ex- wine (Bordeaux). In part because its wealth made ample, retained that status). 904 AMERICAN SOCIOLOGICAL REVIEW tendants were imposed, abolished, and then were initiated in roughly 8 percent of years re-imposed several times between 1616 and experiencing warfare, but in less than 1 per- 1653 (after which they remained in place).26 cent of the peaceful years. Revolts were The elimination of representative institu- slightly more likely to occur during offen- tions mattered less is Guyenne than else- sive wars, but that effect is not strong. Re- where. Even when it was a Pays d'Etat, the volts were initiated in about 5 percent of provincial estates were always weak in years with offensive wars, and in 3.6 percent Guyenne and were virtually eliminated after of years in which France was fighting a de- 1621 (Collins 1994:154; Major 1966:369; fensive war. Parker 1983:81).27 The presence of representativeinstitutions With these rough measures of our inde- had no effect on revolt in Guyenne. Neither pendent variables at the provincial level, we the Estates General nor the provincial estates can now explore their effects on revolts in affected revolt initiation. (Because the pro- Guyenne. Taxes alone had little effect-re- vincial estates were always weak in volts were instigated in 4.7 percent of the Guyenne [Major 1966], we would not expect years prior to the major increases in 1623 them to affect revolts.) and in 4.2 percent of the years after those Overall, our results at the provincial level tax increases. However, just as for France as are similar to those for France as a whole- a whole, adding the turning point of the taxation, administrative centralization, and Fronde clarifies the relationship between war (especially offensive war) are important taxation and revolt. Revolts were three times causes of revolts, and the effects of the first as likely in Guyenne before the Fronde, and two factors interact strongly with the tax increases before the Fronde were over Fronde. Because these results are based on five times as likely to cause revolts as were only a few cases (12 revolts) and very rough tax increases after the Fronde (13 percent of measures of some of our causal factors (i.e., years with tax increases resulted in revolts taxation), we cannot make any strong con- before the Fronde, compared with only 2.2 clusions based on this analysis of Guyenne. percent after the Fronde). However, its consistency with our national- The effects of administrative centraliza- level findings provides good evidence for tion on revolts are similar-they too are the robustness of our conclusions. highly conditioned by the Fronde. In the pre- Fronde period, the imposition of indendants CONCLUSION more than doubled the likelihood of revolt initiation (revolts started in about 13 percent We model the relationship between state- of years with intendants but in less than 5 making and tax revolt as the product of stra- percent of years without intendants prior to tegic interactions between rulers and sub- the Fronde in Guyenne). Just as it did with jects. Both groups are making cost-benefit tax increases, the Fronde virtually eliminates calculations in which an important part of the relationship between administrativecen- the decision-making process is their antici- tralization and revolt-the use of intendants pation of the actions of others. Neither rul- is more than five times more likely to pro- ers nor subjects have perfect information, so duce revolt before the Fronde as after it. both are expected to miscalculate some- War had a very strong effect on revolt in times. However, both are also expected to Guyenne-11 of the 12 revolts in Guyenne learn over time, especially when outcomes were initiated during a French war. Revolts are clear and decisive, so experience should lead to fewer miscalculations.28 26 Intendants were present in Guyenne in 1616, 1619, 1622-1625, 1627-1628, 1630-1648, 1650-1651, and 1653-1788. 28 We do not want to overstate the extent of 27 The estates in Languedoc, for example, were learning in this case. Although rulers did have always more powerful and met more frequently. some information about the policies of prior rul- But although they survived the onslaught on the ers (and reactions to those policies), Peirson Pays d'Etat, they were weakened substantially af- (2000) is correct to note that learning is more dif- ter 1663 (Hoffman 1994:230, 241; Major 1994: ficult in the context of political institutions than 346-47; Mousnier 1979:610-23). in the context of markets. STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 905

We find that the three main aspects of suppress revolt in France as a whole, sug- early modern state-making-war, tax in- gesting that it worked as a forum for nego- creases, and administrative centralization- tiation and information exchange in early all incited tax revolts in early modern modem France, but it had no effect on re- France. However, in each case, our findings volts in Guyenne. Moreover, we found no show that these relationships are complex effect for provincial estates-they did not and often contingent on other factors. affect revolts in Guyenne, and had virtually Our initial analyses show an effect of war no effect in our rough analysis of France as on revolt, but this effect operates only a whole. through offensive wars. When we decom- The effects of taxation and offensive war pose wars into offensive and defensive on revolt are complicated, due primarily to types, our quantitativeresults show that only the existence of a significant historical turn- offensive wars consistently generated re- ing point in our period. As we anticipated, volts. Our provincial analysis of Guyenne the relationship between taxation and revolt also shows a slightly stronger effect of of- was very different before the Fronde than fensive wars. This is consistent with our hy- after it. Prior to the Fronde, tax increases pothesis, but the effect is stronger than we tended to generate revolts; after the Fronde anticipated. The opportunity for revolt they rarely did. The effects of offensive war (based on the weakness of the state) was on revolt are similar, in that they too were clearly not all that mattered-subjects pri- strongly affected by the Fronde. Offensive marily reacted to how their tax money was wars generated revolts prior to the Fronde. being spent. When wars were fought to de- As with tax increases, the Fronde proved to fend them, subjects were more willing to be a decisive turningpoint-indicated in this pay; when rulers pursued offensive wars that case by the fact that offensive wars follow- seemed to provide few benefits, subjects fre- ing the Fronde had no effect on revolt. quently revolted. As Levi (1988, 1997) ar- These findings clearly demonstrate that gues, subjects' compliance is in part a func- history matters. Particularhistorical turning tion of whether they think the state is pro- points change the causal relationships be- viding them with public goods or not. tween variables (Goldstone 1998; Issac and These findings help us explain the differ- Griffin 1989; Mahoney 2000; Peirson 2000). ence between the strong effects of war on In this case, the Fronde was a clear and de- revolt found in most qualitative studies of cisive victory for the state over rebellious early modern states and the weak effects subjects. The combination of repression and found in quantitative studies of the nine- co-optation that followed it significantly teenth and twentieth centuries. Our quanti- changed the costs and benefits of revolt for tative analysis reinforces the qualitative more than a century. After the Fronde, sub- findings for early modern states, and sug- jects less often thought of revolt as a poten- gests that the early modern period is indeed tially successful strategy for decreasing different-that it is thus necessary to specify taxes or opposing offensive wars, and state- a scope condition for the relationship be- making proceeded relatively unimpeded. tween war and revolt. Perhaps the relation- Although our empirical analysis is limited ship between warfare and revolt changes to France, we believe many of our arguments with the advent of nationalism in the nine- could apply to other early modern states in teenth century (Hechter 2000). When "sub- the process of state formation. For example, jects" are transformed into "citizens" with other states could also have experienced turn- some stake in the wars of "their"country, a ing points that transformed the relationship "rally 'round the flag" reaction may become between state-making and revolt. Although more common than revolt. However, be- further research would be necessary to test cause our data do not extend to the nine- this, it is possible that the "GreatRevolt" of teenth century, we cannot test this hypoth- 1381 in England and the Comuneros revolt esis. in Spain (1520-1521) may have played roles Our control variable measuring the exist- similar to the Fronde. Both were responses ence of representative institutions produced to state-making, increased taxation, and ad- mixed results. The Estates General tended to ministrativecentralization (Elliott 1963:141- 906 AMERICAN SOCIOLOGICAL REVIEW

49; Fryde 1981:9-11; Hopcroft 1999:74-75; Edgar Kiser is Professor of Sociology at the Uni- Thompson 1994:182-85). Fryde (1981) and versity of Washington. He has published articles Hopcroft (1999) argue that the English tax in sociology, political science, and economics journals on revolt remained in popular memory and topics including the determinants of war and revolt, the development and decline of shaped the fiscal evolution of the English voting institutions, the centralization and bu- state for centuries by defining the limits of reaucratization of state administration, and the taxation (specifically, an invasive tax like the methodology of historical sociology. poll tax would not be tolerated). The Comuneros revolt in Spain, like the Fronde, April Linton is currently a visiting research fel- low at the Office of Population Research, resulted in a decisive victory for the state Princeton University, and will be, effective July (Elliott 1963: 149; Thompson 1994:184). 2004, Assistant Professor of Sociology at the After thatpoint, the Cortes was marginalized, University of California, San Diego. Her disser- administrative centralization and taxes in- tation, completed in July 2002 at the Department creased, yet "there were no more revolts in of Sociology, University of Washington, is titled Castile against the power of the crown" "Spanish for Americans? The Politics of Bilin- gualism in the United States." Currently she is (Elliott 1963:149). These and other turning pursuing research about bilingualism and lan- points may well be keys to increasing our un- guage policy, the fair trade movement, and revo- derstanding of early modern state formation lutions that incite large-scale social transforma- and reactions against it. tions.

APPENDIX A Tax Revolts in France, 1514 to 1789

Year Location Description 1514 Agen, Guyenne Taxation and political exclusion 1528 Bordeaux, Guyenne Excise tax on wine 1539 Marseille, Provence Gabelle 1542 La Rochelle and Les Iles, Saintonge Gabelle 1544 St. Maixent, Poitou Taxation, religion 1545 Perigeux, St Foy, and Duras, NW Guyenne Gabelle 1545 Sarlat, Niort, and Saintes, Poitou Taxation 1545 St. Maixent, Poitou Taxation 1548-1551 Guyenne Pitaud insurrection 1594-1595 Guyenne Croquant rebellions 1621-1622 Bearn First Bearnese Revolt 1625-1626 B6arn Second Bearnese Revolt 1627-1629 Bearn Third Bearnese Revolt 1629-1630 Guyenne Croquant uprising 1635-1636 Guyenne Croquant uprising 1635-1636 Languodoc Croquant uprising 1637-1641 Perigord and Quercy, Guyenne Croquant uprising 1639 Normandy Nu-pieds 1640 Moulins, Bourbonnais Taxation 1640 Rennes, Brittany Taxation 1641 Poitou Taxation 1641 Saintonge Taxation 1643-1644 Languodoc Rouerge revolt in the South and West 1645 Montpellier, Languodoc Taxation 1648-1653 Angers, Anjou The Fronde (Continued on next page) STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 907

(Appendix A continued) Year Location Description 1650-1651 Bordeaux, Guyenne The Fronde 1650 Tulle, Limosin The Fronde 1652 Guyenne, Ormee of Bordeaux 1655-1657 Guyenne Tardanizat rebellion 1658 Orldanais Sabotiers rebellion 1661-1662 Guyenne Benauge rebellion 1662 Boulonnais (Artois) Lustucru rebellion 1663 Roussillon, Languodoc Angelets guerilla warfare 1663-1670 Gascony, Guyenne Audijos rebellion 1675 Brittany Papier Timbre, Bonnets Rouges (Torreben) rebellions 1789 French Revolution

Sources: Heller (1991); Kiser et al. (1993); Tilly (1986, 1993).

APPENDIX B Wars Involving the French, 1515 to 1788

Year War Type of Wara

1521-1526 First Italian War Defensive 1526-1530 Second Italian War Defensive 1535-1538 Third Italian War Offensive 1542-1544 Fourth Italian War Offensive 1544-1546 Anglo-French War Defensive 1547-1559 Hapsburg-Valois War Offensive 1549-1550 Anglo-French War Defensive 1557-1561 Anglo-French War Defensive 1589-1598 War of the Three Henries Offensive 1609-1614 War of the Julich Succession Offensive 1624-1648 French Thirty Years War Offensive 1627-1628 Anglo-French War Defensive 1628-1631 War of the Mantuan Succession Defensive 1642-1696 Iroquois-French War Defensive 1648-1659 Franco-Spanish War Offensive 1667-1668 War of Devolution Offensive 1672-1689 Third Dutch War Offensive 1676-1679 Messian Rebellion Offensive 1688-1697 War of the Grand Alliance Offensive 1701-1714 Spanish Succession Offensive 1718-1720 Quadruple Alliance Defensive 1733-1738 Polish Succession Offensive 1744-1748 King George's War Defensive 1754-1763 French and Italian War Offensive 1756-1763 Seven Years War Defensive 1764-1769 Corsican War Offensive 1775-1783 American Revolution Offensive a We define wars initiated by powers other than France as defensive wars. Sources: Dupuy and Dupuy (1977); Kiser et al. (1993, 1995). 908 AMERICAN SOCIOLOGICAL REVIEW

APPENDIX C Pearson Correlation Coefficients for Variables Used in the Regression Analyses of Tax Revolt Initiation in France, 1515 to 1789

Variable Variable (1) (2) (3) (4) (5) (6) (7) (8)

(1) Revolt Initiation 1.000 (2) Total tax per capita -.129* 1.000 (3) Before/during the Fronde .183** -.776*** 1.000 (4) Administrative centralization -.094 .776*** -.923*** 1.000 (5) Estates General -.073 -.705*** .754*** -.816*** 1.000 (6) War .185** .039 -.152** .211*** -.227*** 1.000 (7) Defensive war .068 .012 -.223*** .334*** -.279*** .516*** 1.000 (8) Offensive war .185** .077 -.055 .135* -.206*** .741*** .120* 1.000 Note: All independent variables are lagged one year. *p <.05 ** <.01 *** < .001 (two-tailed tests)

REFERENCES Abbott, Andrew. 1997. "On the Concept of Turn- 1500-1700." Theory and Society 12:445-68. ing Point." Comparative Social Research Brustein, William and MargaretLevi. 1987. "The 16:85-105. Geography of Rebellion: Rulers, Rebels, and Arminger, Gerhard. 1995. "Specification and Es- Regions, 1500-1700." Theory and Society timation of Mean Structures:Regression Mod- 16:467-95. els." Pp. 77-177 in Handbook of Statistical Buena de Mesquita, Bruce, Randolph M. Modeling for the Social and Behavioral Sci- Silverson, and Gary Woller. 1992. "War and ences, edited by G. Arminger, C. Clogg, and the Fate of Regimes: A Comparative Analy- M. Sobel. New York: Plenum. sis." American Political Science Review 86: Barkey, Karen. 1991. "Rebellious Alliances: The 638-45. State and Peasant Unrest in Early Seventeenth- Clamageran,Jean Jules. [1867] 1980. Histoire de Century France and the Ottoman Empire." l'impot en France (The history of taxation in American Sociological Review 56:699-715. France). Reprint, Geneva, Switzerland: Bean, Richard. 1973. "War and the Birth of the Slatkine Reprints. Nation State." Journal of Economic History Clark, Samuel. 1978. "The Importance of Agrar- 33:203-21. ian Classes: Agrarian Class Structure and Col- Beik, William. 1985. Absolutism and Society in lective Action in Nineteenth-Century Ireland" Seventeenth Century France. Cambridge, En- British Journal of Sociology 29:22-40. gland: Cambridge University Press. Collier, Ruth Berins and David Collier. 1991. Berce, Yves-Marie. 1987. Revolt and Revolution Shaping the Political Arena: Critical Junc- in Early Modern Europe. New York: St. Mar- tures, the Labor Movement, and Regime Dy- tin's Press. namics in Latin America. Princeton, NJ: Bonney, Richard J. 1978. Political Change in Princeton University Press. France under Richelieu and Mazarin, 1624- Collins, James B. 1988. Fiscal Limits of Absolut- 1661. Oxford, England: Oxford University ism. Berkeley, CA: University of California Press. Press. 1979. "The Failure of the French Rev- . 1994. Classes, Estates, and Order in enue Farms, 1600-60." Economic History Re- Early Modern Brittany. Cambridge, England: view 32:11-32. Cambridge University Press. .1988. Society and Governmentin France . 1995. The State in Early Modern France. under Richelieu and Mazarin, 1624-61. New Cambridge, England: Cambridge University York: St. Martin's Press. Press. Braudel, Fernand. 1972. The Mediterranian Doyle, William. 1996. Venality. Oxford, En- World in the Age of Philip I. New York: gland: Clarendon. Harper and Row. Dupaquier, Jacques. 1979. La population Brustein, William. 1985. "Class Conflict and fran!Vaise au XVIIE et XVIIIE siecles (The Class Collaboration in Regional Rebellions, French population in the 17th and 18th centu- STATE-MAKING AND REVOLT IN EARLY MODERN FRANCE 909

ries). Paris, France: Presses Universitaires de ology 22:305-25. France. . 1995. "Ruler Autonomy and War in Dupuy, R. Ernest and Trevor N. Dupuy. 1977. Early Modern Western Europe." International The Encyclopedia of Military History. New Studies Quarterly 39:109-38. York: Harper and Row. Klaits, Joseph. 1976. Printed Propaganda under Elliott, J. H. 1963. Imperial Spain. Cambridge, Louis XIV. Princeton, NJ: Princeton University England: Cambridge University Press. Press. Fryde, E. B. 1981. The Great Revolt of 1381. Lachmann, Richard. 1989. "Elite Conflict and London, England: Historical Association. State Formation in 16th- and 17th-CenturyEn- Gartzke, Erik.1999. "War Is the ErrorTerm." In- gland and France." American Sociological Re- ternational Organization 53:567-87. view 54:141-62. Gourieroux, C. A. Monfort, and A. Trognon. . 2000. Capitalists in Spite of Themselves: 1984. "Pseudo-Maximum- Likelihood Meth- Elite Conflict and Economic Transitions in ods: Theory." Econometrica 52:681-700. Early Modern Europe. New York: Oxford Uni- Goldstone, Jack. 1991. Revolution and Rebellion versity Press. in the Early Modern World. Berkeley and Los Lachmann, Richard and Julia Adams. 1988. Angeles, CA: University of California Press. "Absolutism's Antinomies: Class Formation, . 1998. "Initial Conditions, General Laws, State Fiscal Structures, and the Origins of the Path Dependence, and Explanation in Histori- French Revolution." Political Power and So- cal Sociology." American Journal of Sociology cial Theory 7:135-76. 104:829-45. Le Roy Ladurie, Emmanuel. 1980. The Peasants Guery, Alain. 1978. "Les finances de la of Languedoc. Chicago, IL: University of Illi- monarchie franqaise sous l'ancien regime" (Fi- nois Press. nances of the French monarchy under the old Levi, Margaret. 1988. Of Rule and Revenue. Ber- regime). Annales 33:216-34. keley, CA: University of California Press. Heller, Henry. 1991. Iron and Blood. Montreal . 1997. Consent, Dissent, and Patriotism. and Kingston, Quebec, Canada: McGill- Cambridge, England: Cambridge University Queen's University Press. Press. Hechter, Michael. 2000. Containing Nationalism. Lewis, Peter 1968. Late Medieval France. Lon- New York: Oxford University Press. don, England: MacMillan. Hoffman, Philip T. 1986. "Taxes and Agrarian Louis XIV. [ 1661-1668] 1970. Memoirs for the Life in Early Modern France: Land Sales, Instruction of the Dauphin. Translated by P. 1500-1730." Journal of Economic History Sonino. New York: Free Press. 46:73-55. Lublinskaya, Aleksandra D. 1968. French Abso- 1994. "Early Modern France, 1450- lutism: The Crucial Phase, 1620-1629. Cam- 1700." Pp. 226-52 in Fiscal Crises, Liberty, bridge, England: Cambridge University Press. and Representative Government, 1450-1789, Mahoney, James. 2000. "Path Dependence in edited by P. Hoffman and K. Norberg. Historical Sociology." Theory and Society 29: Stanford, CA: Stanford University Press. 507-48. . 1996. Growth in a Traditional Society. Major, J. Russell. 1960. Representative Institu- Princeton, NJ: Princeton University Press. tions in Renaissance France. Madison, WI: Hopcroft, Rosemary. 1999. "Maintaining the University of Wisconsin Press. Balance of Power: Taxation and Democracy in . 1966. "Henry IV and Guyenne: A Study England and France, 1340-1688" Sociological Concerning the Origins of Royal Absolutism." Perspectives 42:69-95. French Historical Studies 4:363-83. Issac, Larry and Larry Griffin. 1989. . 1994. From Renaissance Monarchy to "Ahistoricism in Time-Series Analyses of His- Absolute Monarchy. Baltimore, MD: Johns torical Process: Critique, Redirection, and Il- Hopkins University Press. lustrations from U.S. Labor History." Ameri- Mann, Michael. 1993. The Sources of Social can Sociological Review 54:873-90. Power. Vol 2. Cambridge, England: Cam- . 1986. Patrons, Brokers, and Clients in bridge University Press. Seventeenth Century France. New York: Ox- Markoff, John. 1996. TheAbolition of Feudalism. ford University Press. University Park, PA: Pennsylvania State Uni- Kiser, Edgar. 1994. "Markets and Hierarchies in versity Press. Early Modern Tax Systems: A Principal-Agent Matthews, George. 1958. The Royal General Analysis." Politics and Society 22:284-315. Farms in Eighteenth-Century France. New Kiser, Edgar, Kriss A. Drass, and William York: Columbia Universisty Press. Brustein. 1993. "The Relationship between Mettam, Roger. 1988. Power and Faction in Revolt and War in Early Modern Western Eu- Louis XIV's France. London, England: rope." Journal of Political and Military Soci- Blackwell. 91 0 AMERICAN SOCIOLOGICAL REVIEW

Moote, A. Lloyd. 1971. The Revolt of the Judges: Stohl, Michael. 1980. "The Nexus of Civil and The Parlement of Paris and the Fronde, 1643- InternationalConflict." Pp. 297-329 in Hand- 1652. Princeton, NJ: Princeton University book of Political Conflict, edited by T. R. Press. Gurr. New York: Free Press. Morineau, Michel. 1968. "Y'a-t-il eu une Tarrow, Sidney. 1998. Power in Movement: So- revolution agricole en France au XVIIIe cial Movements and Contentious Politics. siecle?" (Was there an agricultural revolution Cambridge, England and New York: Cam- in 18th-century France?). Revue Historique bridge University Press. 234:299-326. Thompson, I. A. A. 1994. "Castile: Absolutism, Mousnier, Roland. 1970. Peasant Uprisings in Constitutionalism, and Liberty." Pp. 181-225 Seventeenth-Century France, Russia, and in Fiscal Crises, Liberty, and Representative China. New York: HarperTorchbooks. Government, 1450-1789, edited by P. . 1979. The Institutions of France under Hoffman and K. Norberg. Stanford, CA: the Absolute Monarchy, 1598-1789. Chicago, Stanford University Press. IL: University of Chicago Press. Tilly, Charles, ed. 1975. The Formation of Na- North, Douglass. 1981. Structure and Change in tion States in Western Europe. Princeton, NJ: Economic History. New York: Norton. Princeton University Press. North, Douglass and Robert Thomas. 1973. The . 1985. "Retrieving European Lives." Pp. Rise of the Western World. Cambridge, En- 11-52 in Reliving the Past: The Worlds of So- gland: Cambridge University Press. cial History, edited by 0. Zunz. Chapel Hill, Parker, David. 1983. The Making of French Ab- NC: University of California Press. solutism. London, England: Edward Arnold. . 1986. The Contentious French. Cam- Pierson, Paul. 2000. "Increasing Returns, Path bridge, MA:Belknap. Dependence, and the Study of Politics." . 1990. Coercion, Capital, and European American Political Science Review 94:251-67. States, AD 990-1990. Cambridge, MA: Pillorget, Rene. 1975. Les mouvements insur- Blackwell. rectionnels de Provence entre 1596 et 1715 1993. European Revolutions, 1492- (Insurrectional movements in Provence, 1596 1992. Cambridge, MA: Blackwell. to 1715). Paris, France: Editions A. Pedone. Wallerstein, Immanuel. 1989. The Modern Ranum, Orest. 1993. The Fronde: A French World-System.Vol. 3. New York: Academic. Revolution, 1648-1652. New York: Norton. Weber, Max. [1922] 1968. Economy and Society. Rasler, Karen A. and William R. Thompson Vol. 1. Edited by G. Roth and C. Wittich. 1989. War and State Making: The Shaping of Translated by E. Fischoff et al. Reprint, New Global Powers. Boston, MA: Unwin Hyman. York: Bedminster. Robinson, Peter. 1982. "On the Asymptotic Prop- Westrich, Sal Alexander. 1972. The Ormee of erties of Estimators of Models Containing Bordeaux. Baltimore, MD: Johns Hopkins Limited Dependent Variables." Econometrica University Press. 50:27-42. White, H. 1982. "Maximum-Likelihood Estima- Rowen, Herbert. 1980. The King's State. New tion of Misspecified Models." Econometrica Brunswick, NJ: Rutgers University Press. 48:817-838. Sewell, Willian H., Jr. 1996. "Three Temporali- Wilson, Charles and Geoffrey Parker, eds. 1977. ties: Toward an Eventful Sociology" Pp. 245- An Introduction to the Sources of European 80 in The Historic Turn in the Human Sci- Economic History, 1500-1800. London, En- ences, edited by T. McDonald. Ann Arbor, MI: gland. Weidenfeld and Nicolson. University of Michigan Press. Wolff, Philippe. 1969. Documents de l'histoire Seyssel, Claude de. [1519] 1981. The Monarchy du Languedoc (Historical documents of of France. New Haven, CT: Yale University Languedoc). Paris, France: Univers de la Press. France, Collection d'histoire regional. Skocpol, Theda. 1979. States and Social Revolu- Zagorin, Perez. 1982. Rebels and Rulers, 1500- tions. Cambridge, England: Cambridge Uni- 1660. Cambridge, England: Cambridge Uni- versity Press. versity Press.