Johns Hopkins University Press
abstract

This article builds on the large literature that discusses if frequent international wars enhance state-building, as famously argued by Charles Tilly. It integrates key insights of that literature and a series of additional arguments into a more comprehensive and systematic model of bargaining between rulers and ruled. The model specifies the conditions under which wars are likely to build states: if there are political institutions enabling such bargaining and expressing the consent of the ruled, if the population contributed substantially to the war efforts by providing soldiers and taxes, and if rulers are legitimized either through nationalism or success at war. The article expands the empirical horizon of existing quantitative research by assembling two measures of state development, ranging from the early modern period (for nearly 20 states) to the years from 1860 to the present (for 116 countries). Findings from a variety of regression models empirically support the model.

Key Words

interstate war, state formation, taxation, bargaining, legitimacy, consent

CHARLES Tilly's dictum that "war made the state, and the state made war" perhaps represents one of the most often quoted phrases ever written by a social scientist. 1 Although a half century has passed since its initial publication, Tilly's so-called bellicist thesis of state formation continues to be vividly debated. 2 To be sure, the vast literature includes other research on the rise of the modern state, including prominent arguments in economics about how state formation interacted with property rights and economic growth. 3 The bellicist thesis remains at the core of discussions in political science and historical sociology, however. Such discussions are mostly concerned with the first [End Page 188] half of Tilly's equation: whether wars built states, rather than exploring if recently strengthened states also fought more wars subsequently. This article shares this focus on the possible state-building effects of wars.

It makes a twofold contribution to this conversation. First, we use a larger universe of cases than in previous research to empirically evaluate the bellicist argument as well as its major criticisms, specifications, and extensions. Most of the empirical literature is case specific or compares across a few historical examples. Quantitative research has explored samples of countries after 1970, when data on tax revenues as a measure of state capacity become available, or three dozen war-prone states between 1870 and 2007, or a dozen European states during the early modern period. 4 We push these geographic and temporal boundaries and explore the bellicist theory of state-building for longer periods of time, stretching from the fifteenth century to the present, and for a larger, if still limited, sample of countries. We rely on two different measurements of state-building: one refers to tax revenues per capita from the early modern period to the nineteenth century and covers nineteen countries, while the other measures the government's share of gdp and is available from the late nineteenth to the end of the twentieth century for 116 countries. We develop our war data set based on a list of wars fought across the world since 1400.

Second, we build selectively on the literature to introduce a comprehensive and systematic theoretical model that specifies the conditions under which wars are likely to build states. Most existing research, by contrast, explores one or two specific hypotheses. We aim to integrate a variety of existing hypotheses into an encompassing theoretical framework and to derive a set of novel arguments from that framework. Going beyond Tilly's emphasis on coercion, the theory emphasizes consent and legitimacy. We argue and show empirically that wars were likely to build states where the population had enough negotiating power to demand public goods in return for their consent to increasing taxation during wartime and for sacrificing their young men on the battlefield. This bargaining process has three aspects: the institutionalization of consent by subjects/citizens, the resources they contribute to the war effort, and in how far they view the rulers and their war efforts as legitimate.

Wars have more state-building effects in polities with institutionalized representation of elites in assemblies (before the advent of mass democracy) or with popular representation through democratic elections [End Page 189] (in the nineteenth and twentieth centuries; the consent aspect). Wars had more state-building effects if large segments of the population participated in the fighting and if these wars were financed by taxes rather than loans, foreign aid, or oil (the resource aspect). Finally, wars were more likely to build states if they were won and if they were fought by a regime that ruled in the name of the nation rather than a dynasty or empire (the legitimacy aspect).

We first summarize Tilly's original argument and then discuss the various criticisms and modifications proposed over the past decades. The third section integrates some of these propositions into a bargaining and legitimacy theory of state formation and war and introduces a series of hypotheses informed by this model, some already discussed in the literature and some new ones derived from the theory. We then describe the data and analytical approach and present the findings. A short section concludes.

I. Tilly's Bellicist Thesis

Tilly argued that the territorial states of Western Europe emerged in the early modern period because the geopolitical environment became competitive and violent, making wars more intense, frequent, and expensive than before. Successful rulers met this challenge by building new institutions to extract taxes from the population or from its centers of wealth and to coerce men into military service. Tilly famously analogized these early modern states to organized crime, because both operate as "protection rackets," offering protection against violence by acting themselves violently. 5 This "extraction-coercion" cycle created a long-term "ratchet" effect: 6 the institutions that were built during war were maintained during peacetime and thus increased the capacity of the rulers to govern and impose their will on the population.

Later on, Tilly developed a multilinear model that also considered preexisting levels of power concentration and capital accumulation. The model foresaw various pathways to the modern state. The most effective states emerged where medium power concentration combined with medium capital accumulation, such as in England, France, and Prussia. 7 Overall, this expanded theory still maintains that the more frequent the warfare from the 1400s onwards, the more rapidly states [End Page 190] converged toward the model of the large, centralized, bureaucratic, and territorial state.

II. Criticism and Modifications

Most subsequent scholarship saw some merit in Tilly's thesis. 8 But it also identified alternative paths to state formation, specified scope conditions for the bellicist mechanism to work, or specifically discussed whether the model also operates in the contemporary world and outside of the West.

Alternative Paths to State-Building in Early M odern Europe

The first critique maintains that many states in Western Europe developed without fighting many wars. Hendrik Spruyt points out that some states did not go to war with, but rather sought the protection of, other, more militarily powerful states or pulled out of the field of inter-state military competition altogether. In cases such as Denmark, the Netherlands, Sweden, and Switzerland, this stance resulted in more pacific forms of state development. 9 In earlier work, Spruyt also shows that state centralization in France preceded the rise in war frequency and was achieved by political bargains and coalition building, rather than coercion à la Tilly. 10 Pointing to yet another alternative mechanism, Philip Gorski argued that religious asceticism was as much a motivating force for the state-builders of the Netherlands and Prussia as was the hunt for more taxation to finance larger armies. 11

Specifications of the Bellicist Thesis in the West

A second group of criticism identifies the scope conditions under which wars were likely to build states in the West. Some authors maintain that the population, or at least local and regional power-holders, needs to consent to new taxes to finance wars. 12 Along these lines, Edgar Kiser and April Linton compared the structures and roles of the English parliament and the French estate assemblies and concluded that the former "have functioned as a mechanism for cooperation and [End Page 191] credible commitment," thus increasing the state-building effects of frequent wars, while the French estates dampened these effects because of their regionally fragmented nature and their lack of veto power. 13 More recent research on France and Spain shows, however, that the fragmented and weaker provincial estates played a role similar to the unified English parliament, albeit at a different level, in enhancing the state-building efforts of kings. 14

Other authors highlight a second facilitating condition. Having a minimum degree of governmental capacity, they argue, is a precondition for wars to build states because otherwise states cannot extract resources from the population and manage expenditure effectively during and after war. 15 Scholars see this condition as one of the main reasons that the quite frequent international wars in Latin America did not have the same state-building effects during the nineteenth century as they had earlier in Europe. 16

Along similar lines, Thomas Ertman shows that wars only encouraged bureaucratic development if they increased in frequency after government positions were no longer sold to raise state revenues but given to university-educated professionals (as was the case in Sweden and Prussia). 17 The rise of a Weberian state, in other words, preceded intense war-making. 18

Several scholars suggest that only a subset of interstate wars led to state development—pointing at yet another set of scope conditions. According to Karen Rasler and Willaim Thompson, 19 only wars that were won led to state-building, because governments that lost will "pay reparations and sustain losses of territory and population … and suffer in terms of popularity, legitimacy, and [political] survival," all of which will make it more difficult to sustain wartime levels of resource extraction. 20 In earlier work, Rasler and Thompson assumed an international relations perspective to argue that only global wars, which were fought about once a century over who assumed a dominant role in the international state system, lead to the development of those states that ended up winning the global contests. 21 [End Page 192]

Herbert Obinger and Klaus Petersen argue that only wars with mass participation, such as the Second World War, had consistent effects on state development, independent of whether wars were lost or won. 22 They highlight several mechanisms, from creating popular programs during the war to maintain the loyalty of the citizenry all the way to rewarding returning soldiers and widows for their sacrifice. We also note here that according to Kenneth Scheve and David Stasavage, mass warfare provided an opportunity for governments to not only introduce but also maintain heavily progressive taxation systems, arguing after the war that its burden had fallen disproportionally on the poor whose tax burden therefore needed to lighten. 23

Wars and State-Building Outside of the West

Much of the critical assessment of Tilly's thesis came from scholars who asked if it could be applied to other parts of the world and the contemporary period, despite Tilly's reservations and ambivalences toward that end. 24 Most researchers concluded that wars did not have the same state-building effects in postcolonial societies as they had in Western Europe in the past. 25 They have brought forward several reasons for this difference. Most importantly, the sovereignty of even the weakest postcolonial states was granted by external powers (such as the former colonial master) and newly developed international norms, 26 thus blocking the evolutionary selection mechanism foreseen by Tilly's theory. Furthermore, not only were wars between states less frequent, but they resembled skirmishes rather than the full-out confrontations of earlier decades and centuries. 27

Finally, many postcolonial states developed as rentier states that depended on foreign aid or international credit rather than the resources of the population—a point already noted by Tilly. They therefore did not need to bargain with or consent to the demands of their citizens to fight wars, and the external powers that financed them did not demand [End Page 193] much state-building either. 28 The source of war financing is important beyond the contemporary Global South, however: Didac Queralt examined the long-term fiscal consequences of wars fought over the long nineteenth century. He finds that credit-financed wars did not leave a legacy of higher taxation levels at the end of the twentieth century, while wars fought without access to credit did. 29

III. A Consent and Legitimacy Theory of War and State -Building

We now develop our theory, selecting and modifying some of the arguments discussed above to fit them into as coherent and precise a model of war and state-making as possible. This model will also allow us to generate some new hypotheses. We hope that it will help to move the discussion beyond modifying, criticizing, or testing individual aspects of Tilly's original argument. We are therefore not pursuing the—equally worthwhile—goal of testing Tilly as precisely as possible (as do Kivanç Karaman and Şevket Pamuk, for example), but build upon the half a century of subsequent discussions to develop a modified and improved version of the bellicist theory. 30

Our bargaining model of state-building features institutionalized mechanisms of consent and patterns of legitimacy as crucial elements. Tilly himself tended to treat consent and legitimacy rather dismissively and instead emphasized force and coercion as the main mechanism of early modern state-building. 31 While he did argue that democracy emerged through a bargaining process between the state and the population during the long nineteenth century, ending with civilian control over the military leviathan, consent and legitimacy played no role in the analysis of its emergence. 32 We build on Spruyt, Jean-Laurent Rosenthal, Kiser and Linton, Margaret Levi, Timothy Besley and Torsten Persson, and Stasavage to argue that the ability of the state to increase taxation and enlarge the state apparatus crucially depends on the ongoing bargaining process between rulers and ruled. 33 [End Page 194]

According to Clemens Kroneberg and Andreas Wimmer's model of this negotiation process, rulers are interested in the taxes and a steady supply of soldiers while the ruled are interested in political influence over the rulers' decisions (including the decision to go to war) and in public goods. 34 These goods include, among others, protection from arbitrary violence through policing; a court system based on the rule of law; public infrastructure such as water fountains, water ways, or freeways later on; health care; access to education; and so on. Wars shift the bargaining equilibrium, we argue in this article, because they increase rulers' desire for taxes and soldiers and thus their willingness to compromise. Whether wars make states therefore depends on the balance of power between rulers and ruled. This balance of power has three aspects: an institutional, a resource control, and a legitimacy aspect, each of which we discuss in a separate section.

Institutionalized Consent

We follow Kiser and Linton, and Rosenthal in suggesting that representative institutions increase the state-building effects of war in three ways. 35 Institutionalized consent (based on budgetary veto power) allows for raising taxes in efficient ways because it reduces resistance in the form of rebellions, tax evasion, or withholding information; it increases the likelihood that rulers fight wars that they can win, which in turn increases state revenues; and it obliges rulers to provide more public goods in peace times in return for war support and increased taxation. 36

We test this argument with data on tax revenues per capita for a series of countries in the Old World from 1500 onward. Unfortunately, high quality data are available for fewer than two dozen countries only, leaving out the New World and large swaths of Africa and Asia.

—Hypothesis 1 (h1). In the early modern period, wars initiated during periods with assembly meetings will have more state-building effects (as measured by tax revenues) than wars fought during years or periods with no or few assembly meetings.

We further specify this hypothesis below by distinguishing between offensive and defensive wars. This distinction helps to address at least [End Page 195] some of the endogeneity concerns, as assemblies could have been called ahead of wars to secure their financing.

A second and novel hypothesis brings the consent argument to the modern era. The idea of representation changed dramatically with the advent of mass democracy in the second half of the nineteenth century, with important consequences for the bargaining between governing elites and the masses. 37 At the same time, the post-Napoleonic period saw the rise of universal conscription and later on of general income taxes, thus considerably increasing the contributions that nonelites made to the war effort. The consent of the general population thus gains even more significance than it did during the previous period of nonrepresentative assemblies.

—Hypothesis 2 (h2). Wars build states more effectively in democracies than in nondemocratic countries.

The three mechanisms through which institutionalized consent affects the state-building effects of wars remain the same as in the period of assemblies, but they take on a new significance. Where rulers depend on the consent of the entire population for fighting wars, the state owes its population even more public services and goods in return for the sacrifices in blood and taxes made during the war. Furthermore, democratic leaders, even more so than rulers who were constrained by the consent of assemblies, tend to fight wars that they can win, fearing to be voted out of office if they lose, 38 thus further increasing the legitimacy to expand the domains of the state after the war ends in triumph. Empirically, therefore, we have to disentangle the mechanism of institutional consent from the legitimacy that a won war provides a ruling elite. Finally, democratic elections are even more efficient than the consent of assemblies in authorizing the state to expand its purview further and further, replacing guilds, village communities, clans, and local lords as providers of security, legal protection, and public infrastructure.

We note that according to a series of scholars, the relationship between democracy and state-building in the wake of war could be confounded by an endogenous relationship between war frequency and democratization: increased demands for taxes to finance wars encouraged rulers to accept representative institutions through which they could obtain [End Page 196] a credible commitment to pay higher taxes. 39 In the empirical analysis that follows, therefore, we have to carefully assess the relationships among war frequency, democracy, and state capacity, which we do using an instrumental variable approach.

Resource Contributions by Subjects/Citizens

Our core hypothesis states that the state-building effects of wars should be larger the more resources the population contributed to the war effort, forcing rulers to provide more public goods in return. 40 Two hypotheses, already formulated in the literature, relate to this proposition.

—Hypothesis 3 (h3). Wars that cost many soldiers' lives should have greater state-building effects than wars that had fewer participants and cost fewer lives.

We note here that mass participation and casualty rates are not endogenous to state capacity: Many countries with weak prewar state capacity fielded large armies and sustained substantial casualties. For example, the correlation between prewar state capacity, as measured by the government share of the economy (see below), and war casualties is very low and negative (at −.1) for the countries that participated in either or both of the World Wars. 41

A second resource aspect concerns how governments finance wars. In line with existing arguments, we hypothesize that wars funded through international aid should produce very little state-building effects. To this we add the argument that the same should hold true for governments that finance their war activities through the sale of oil. Similar to aid-dependent countries, oil-rich regimes do not need to negotiate with their own citizens to expand their activities because they do not rely on taxes to do so. 42 Finally, and building on Queralt, 43 we hypothesize that wars financed by international credit or grants should enhance state capacity much less than wars financed by domestic taxes. In short, then,

—Hypothesis 4 (h4). Wars should not build states during periods of substantial aid-dependency or oil production or if they were financed by international credits and loans, rather than by domestic taxes. [End Page 197]

Legitimacy

We now turn to the legitimacy aspect of the theoretical model. The core hypothesis is that wars perceived as justified by the population will have more state-building effects.

—Hypothesis 5 (h5). More specifically, wars that are won should build states.

A won war provides a government with the aura of having successfully defended the population's interest so that it seems justified to expand its realms and grasp a larger share of the country's resources. 44 Victory can also increase popular identification with the nation, which, in turn, encourages and legitimizes regimes to invest in public goods. 45 Other possible mechanisms (including the loss of population and territory or reparation payments after a lost war) are obviously plausible as well and are independent of the legitimacy mechanism we envision here. 46

We suggest an additional and novel legitimacy argument that relates to the foundational ideology of states. Nationalism provided, from the early nineteenth century onward, a new principle of legitimacy prescribing that rulers and ruled should hail from the same national community and that rulers should pursue the national interest, rather than looking after the fate of their dynasty or seeking to spread civilization across the world, as empires claimed to do. 47 Compared to dynastic states or empires, citizens should therefore perceive the governments of nation-states and the wars they fight as legitimate and justified, fostering what Levi has called "conditional consent," which in turn reduces resistance to increasing taxes. 48 Conversely, legitimacy should also increase the expectation that rulers should provide members of the nation with public goods.

—Hypothesis 6 (h6). Wars should have more state-building effects once the government relies on nationalist principles to justify its rule and the wars that it fights. [End Page 198]

Note that we unfortunately cannot empirically disentangle the legitimacy of the regime from that of the wars it fights and we therefore have to leave the hypothesis with a double meaning. As with democracy, we also have to address possible endogeneity issues. It may very well be that war frequency is related to nation-state formation: the more wars fought in the past, including wars of independence, the more likely a territory should transition from empire to nation-state. We again deal with this possible complication using an instrumental variable approach.

Other Arguments

In short, institutionalized consent, large resource contributions to the war effort by subjects/citizens, and legitimacy allow the government to raise taxes without having to beat back tax revolts and fight mass tax evasion. After the fighting ends, it will be easier to keep taxation at the high war-time level and governments are encouraged to use these additional resources not only to maintain larger armies but also to provide public goods to the population.

Three hypotheses that emerged from the literature review were not compatible with this theoretical framework and we evaluate them, in a separate subsection, as competing explanations. They focus either on the nature of the international system, which is not part of our model, or on domestic factors unrelated to institutionalized consent, resource contributions, or legitimacy.

—Hypothesis 7 (h7). Only global wars enhanced the state-building of its winner.

—Hypothesis 8 (h8). Wars did not build states after the Second World War, because the international system protects even the weakest states.

—Hypothesis 9 (h9). Wars only built states if states already showed a minimal level of capability before the period of intense warfare.

Table 1 gives an overview of these nine hypotheses.

IV. Data and Methods

Only a few quantitative studies have examined Tilly's thesis to date. They analyzed data from 1975 onward either on a global scale, or for one or two continents, or for nine Central American countries from [End Page 199]

Table 1. L H
Click for larger view
View full resolution
Table 1.

List of Hypotheses

1900 to 2000, or for a sample of thirty-nine countries (all of which fought interstate wars) from 1870 and 2007, or for eight to twelve European countries in the early modern period. 49 We build on and go beyond these existing studies by expanding the geographical horizon and by including both the early modern period and the late nineteenth and twentieth centuries. We thus offer a composite universe of cases made from different samples and time periods. Seen together, they generate an almost global perspective on the world of independent states and a long time horizon stretching from ca. 1500 to 2000. Notable omissions are the many non-Western states before their incorporation into Western colonial empires for which we lack data on tax revenues.

State Capacity

For the early modern period, we follow the bulk of the literature and use tax revenues per capita (converted into gold grams) to measure [End Page 200] levels of state-building. We assembled the data from several sources and harmonized various measurements. 50 The variable contains information for 50-year periods, stretching from 1500 to 1900, for 19 countries, including the non-Western cases of the Ottoman Empire, Russia, China, and India.

How can we pursue this story from the nineteenth to the end of the twentieth century? Taxation data are available for a large number of countries from 1975 onward only. We therefore looked for a new measurement of state capacity that would cover the nineteenth and twentieth centuries and as many countries as possible. A brief revisit of how to define state capacity is in order here. It is increasingly common among scholars to distinguish between three aspects: An organizational aspect, which refers to the bureaucratic capacity to implement policies (whatever these are) across the territory of a state; an extractive aspect, which mainly revolves around the fiscal capacity to raise taxes from citizens (as in the measurement discussed above); and finally, the coercive capacity to establish a monopoly of violence within the state's territory and defend it against outside aggressors. 51

Tilly was most interested in the extractive and coercive aspects of state capacity, but he also repeatedly linked fiscal capacity to bureaucratic capacity. And indeed, as Jonathan Hanson and Rachel Sigman show, the two aspects do not form, from an empirical, inductive point of view, clearly distinguishable aspects of state capacity. 52 They therefore suggest to use a latent variable approach based on two dozen individual indicators, most of which become available from the 1970s onward only, however.

For our purposes, we need a measurement with longer time coverage. To our knowledge, the best available is the ratio of the central government's expenditure per capita to the gdp per capita of each country, or the government share of the economy for short. Michael Mann used this measurement in his pioneering study of English finances, and, subsequently, a range of economic historians have used it as well. 53

Our data on government expenditure were adopted from the monumental compendium of historical statistics assembled by Mitchell, which we complemented with additional sources for the Ottoman Empire [End Page 201] as well as the Soviet Union (for details see Appendix 1 in the online supplementary material). 54 All figures were converted into US constant dollars using time-varying conversion rates and standardized per gdp. 55 To increase the confidence in the government expenditure data, massive data cleaning and a fair amount of hand corrections were necessary, for example, for currency devaluations or the introduction of new currencies, shifts in the standards of what counts as government expenditure, and so on.

Conceptually, the government share of the economy relates to all three aspects of state capacity mentioned above: it is higher the larger the tax base of the state, the more a country spends on the military and police, and the more it maintains a bureaucracy that can implement policies effectively. And indeed, Appendix 1 in the supplementary materials shows that the government share of the economy is correlated with various indicators of taxing capability, with Hanson and Sigman's composite index of state capacity, as well as with a range of measures of administrative/bureaucratic quality—all of which are available only for a limited number of years and thus not suitable as dependent variables on their own.

One concern with using these data as a dependent variable is that they could be endogenous to war frequency, especially if these are financed by domestic or international debt and thus increase the government share of the economy after the war through debt payments. However, using data from Rosella Cappella Zielinski for forty countries on how they financed their wars since 1816, we find no relationship between the government share of the economy five or ten years after the war and the degree to which wars were financed by debt. 56

Independent Variables

We are primarily interested in the effects of the cumulative number of wars on state-building. Given the mutually reinforcing relationship between war-making and state-building, a cumulative war count captures the supposed mechanism in an appropriate way, including the ratchet effects that Kiser and Linton discuss. 57 [End Page 202]

To create such a war count, we use a list of wars that historian Peter Brecke assembled—by far the highest quality data source with the long time horizon needed for this study. 58 Brecke defines war as violent conflict between two armed parties, at least one of which represents a government, resulting in at least thirty-two battlefield deaths. This represents a relatively low threshold compared to the 1,000 battle deaths in the often-used Correlates of War data set. Following this definition, massacres of civilians or armed conflict between warlords do not count as a war, as is also the case in other war data sets. Using a large number of case-specific literatures, we code for each war if it was an interstate or civil war, who the participants were, and where the major battlefields were located. On that basis, we create a count of interstate wars that have been fought since 1400 (for more details, see Appendix 2.1 in the online supplementary material).

Hypothesis 1 states that early modern wars had more state-building effects where and when there were assemblies that could consent to the war efforts and to increasing taxation. We use data that are available for fifty-year periods (the temporal granularity of the dependent variable) wherever available or for centuries. For the half century data, we aggregated Andrej Kokkonen and Jorgen Möller's list of years during which assemblies were held. 59 Jan Luiten van Zanden, Eltjo Buringh, and Maarten Boske offer data on the percentage of years with assembly meetings in each century. 60 We complemented these two sources with data from Stasavage for Venice and Tuscany as well as our own coding for the non-European countries as well as for the nineteenth century (using the years in which parliaments were elected according to the V-Dem data set). 61

To test the argument that the association between state-building and wars should be stronger in democracies during the modern, post-Napoleonic period (h2), we created separate war counts for the years during which a country was a democracy and those when it was not, using the conventional six as a cut-off point in the combined democracy/autocracy score of Polity2. 62 For robustness purposes, we also use a continuous coding of this variable and report results below. [End Page 203]

To test whether wars had more state-building effects in nation-states, compared to empires or dynastic kingdoms (h6), we counted the number of wars fought before the transition to the nation-state and those that occurred after that transition, using the data on transition years from Wimmer and Brian Min. 63

To test the effects of aid dependency (h4), we used oecd data, available from 1960 onward, on the total amount of official development aid (in constant US dollars). 64 We then counted the number of wars fought while a country received aid and those fought without receiving any aid. Similarly, we use an oil production per capita variable 65 to distinguish wars fought during oil-producing years and those fought by countries or during years without oil production. For robustness purposes, we also use continuous, rather than dichotomized, codings of these variables and report results below. Again relying on data from Capella Zielinski for a reduced number of forty countries, we also count the number of wars financed without domestic taxes and those at least partially funded by taxes. 66 The same goes for the number of wars funded at least partially by outside sources, such as loans, grants, or plunder, or funded without relying on any such outside resources.

To determine whether wars won have more state-building effects than those lost (h5), we use war data from the Correlates of War (cow) project. 67 They are available from 1816 onward. We also use cow data on battlefield deaths to test h3 and scale it per square kilometer because population data for earlier periods are unreliable (for robustness purposes, however, we standardized by population as well).

To test whether only global wars won contribute to state development (h7), we used Rasler and Thompson's list of global wars, identified these in our war list, determined the winners of these wars, and coded two separate war counts on that basis. 68 And finally, we test hypothesis 8 by exploring whether wars built states before and after the Second World War in the Global North and South, given the changed international environment of the postwar world.

Several control variables test other arguments that seek to explain state development and that need to be taken into account to evaluate [End Page 204] our revised theory of state-building and war-making. Appendix 2.2 in the online supplementary material describes them in detail.

Analytical Strategy and Model Specifications

For the early modern period, when we use gold gram taxes per capita as the outcome variable, the units of observation are the historical states that existed at the time (e.g., Venice or Prussia). We coded all variables, including the war count, on these units. For the nineteenth and twentieth centuries and the the government share of the economy as the dependent variable, the units of observations are the countries that existed in 2001. We coded all variables on these units, including the war count. If we had coded historical states for the nineteenth and twentieth centuries as well, we would not have been able to establish a long-term war count, for example, and it would have become difficult to evaluate our revised bellicose theory of state formation over the long run.

But is taking contemporary states as units of observation substantially justifiable? For an answer, we refer the reader to Appendix 3 in the supplementary online material, which shows empirically that most contemporary countries emerged from imperial provinces or predecessor states with similar borders. Furthermore, empires often relied on indirect rule that conserved preexisting polities. Finally, the complete eradication of a polity is extraordinarily rare and even more rarely permanent. These three findings combine to justify the continuity assumptions underlying our research design.

Data analysis proceeds in two steps. In the first step, we focus on early modern Europe and use gold grams of taxes as the dependent variable. Since these data vary over time, we can use time fixed effects as well as country fixed effects to take into account unmeasured differences in context, both across states and over time. All independent variables are lagged one period (half a century or a century).

The second step is based on the government share of the economy, which takes the analysis to the second half of the nineteenth century and the entire twentieth century. Since the dependent variable is also varying over time, we can again use country fixed-effects regressions. Time trends were taken into account with natural cubic splines on calendar year, a technique that allows us to capture possible nonlinear baseline trends (such as reversals; using year dummies resulted in substantially identical findings). All independent variables are lagged one year. Changing to a ten or twenty year lag did not affect the substantial [End Page 205] results, as Appendix Table A1 shows. 69 This data set covers time periods for which there are many more variables available, which allows us to test all nine hypotheses introduced above. Because the data are large, imbalanced, and characterized by all kinds of temporal dependencies, we used Driscoll-Kraay standard errors. 70

Before we present the findings, a note on potential endogeneity between war frequency and state-building is in order—other endogeneity concerns are dealt with in a subsection below. In the models with the government share of the economy as the time-varying dependent variable, we are more confident that our estimates are not severely biased. Additional analyses (not shown) demonstrate that the levels of stateness in the second half of the fourteenth century or during the first half of the nineteenth century—as measured by the state antiquity index assembled by Valerie Bockstette, Areendam Chanda, and Louis Putterman—are not predictive of the number of wars fought from 1860 onward—thus raising doubt about the states-made-wars half of the Tillyean equation, at least for the late nineteenth and twentieth centuries. 71 We arrive at the same conclusion using shorter time lags and data from 1860 onward: In a structural equation model (results available upon request), the government share of the economy five years prior does not predict the propensity to fight wars today, while the wars fought in the immediate past predict government shares in the economy five years later, in line with the main findings to which we now turn.

V. Main Findings

Wars and Taxes in the Early Modern Period

In the first step of the analysis, tax revenues per capita are the outcome variable, which average to about four grams of gold per capita per year. The percent years in which representative assemblies were held—such as the British Parliament, French estates, or Spanish cortes—serves as the moderator of the impact of wars. This modeling strategy allows us to test hypothesis 1, which states that the presence of assemblies increased the state-building effects of war. [End Page 206]

As mentioned above, however, parliamentary consent for wars and increased taxation may be endogenously related to wars: rulers may call for assemblies to meet and grant them tax increases in view of a costly war that the ruler would like to fight. 72 To mitigate this potential problem, we went through the hundreds of wars in the list and decided, on the basis of detailed research for each individual conflict, whose troops initiated the fighting and who was initially on the defensive side (we did not code allies who did not field battle troops). The following analysis focuses on wars that were forced upon a polity by its enemies, thus minimizing (though obviously not entirely eliminating) the endogeneity concerns mentioned above since assemblies were less likely to be called to raise taxes in defensive wars. 73

In Table 2, the main independent variable of interest is the interaction between the percentage of years during which representative assemblies were held during 50- or 100-year periods and the number of wars forced upon a country during these periods. In models 1 and 2, data entries are for nineteen countries in eight half-century periods between 1500 and 1900, the level of temporal aggregation of most of the taxation data. Because some of the data on assembly meetings between 1500 and 1800 are only available in century intervals, models 3 and 4 repeat the analysis with all data aggregated to 100-year periods.

Models 1 and 3 estimate the associations of tax revenues with wars and assembly meetings. These associations are positive and statistically significant, even though the coefficient for assemblies in model 3 is significant only at the 0.1 level (in a two-tailed test). Models 2 and 4 add an interaction between wars and assembly meetings, which has a positive and statistically significant coefficient in both models and improves the model fit, as multiple indices at the bottom of Table 2 show. The beta coefficient for this interaction is sizable in both models, but the interpretation is not straightforward. However, we can easily compare the predicted change in tax revenues due to fighting more wars in periods [End Page 207]

Table 2. R T R G G C , 1500−1900a
Click for larger view
View full resolution
Table 2.

Regressions of Tax Revenue in Gold Gram per C apita, 1500−1900a

[End Page 208]

with no assembly meetings to the predicted changes in periods with an average number of assembly meetings. In model 2, a standard deviation increase in the number of wars forced upon a polity is associated with an increase of 0.19 standard deviations in tax revenues in periods with an average percent of years with assemblies (39.63) and with a 0.02 standard deviation decrease in periods with no assembly meetings (a net difference of 0.21 standard deviations). In model 4, a standard deviation increase in the number of forced wars is associated with a 0.14 standard deviation increase in tax revenues in periods with an average percent of years with assemblies (33.24) and with a 0.12 standard deviation decrease in periods with no assembly meetings (a net difference of 0.26 standard deviations). The interaction effect is thus quite substantial.

The results support hypothesis 1: The more wars were fought during periods when assemblies met more often, the higher the tax revenues during the subsequent period (fifty-year period in model 2 or a century in model 4). 74 In light of the limited number of countries included in this analysis, we consider the findings an encouraging but preliminary validation of the theoretical argument. 75 The following section tests not only the consent argument, but the full set of hypotheses using a variable-rich data set with large coverage, shifting the focus from early modernity to the second half of the nineteenth and the twentieth centuries. The outcome variable now is the government share of the economy. We use yearly observations and include important control variables, briefly discussed in Appendix 2.2 in the supplementary online material.

Wars and the Government's Share of gdp after 1860

Model 1 in Table 3 shows that, on average, fighting more wars in the past is associated with an increased government share of the economy in the present, thus supporting the bellicose argument in its most general, unspecified form. 76 How substantially important is this bellicose effect compared to that of other important factors that are known to influence the government share of the economy, such as transitioning to democracy 77 or economic development? A standard deviation more wars (or 12 wars) increases the government share of the economy by [End Page 209]

Table 3. C F E R G E /GDP 1860 O
Click for larger view
View full resolution
Table 3.

Country Fixed Effects Regression of Government E xpenditure/GDP from 1860 Onward

[End Page 212]

0.20 standard deviations, while the transition to democracy and economic growth increase the government share by 0.18 and 0.27 standard deviations, respectively. Clearly, wars are not the only factor driving state development in the late nineteenth and twentieth centuries, but their effect is of comparable size to other processes discussed prominently in the literature.

Model 2 returns to the institutionalized bargaining part of the theory, as specified for the nineteenth and twentieth centuries. In line with theoretical expectations (h2), we find that wars fought during democratic periods of governance boost the government share of the economy, while those initiated by nondemocratic regimes are not associated with the outcome variable and the sign of the coefficient is negative. Endogeneity concerns are addressed further below.

Next we explore the resource aspect of the theory. In line with h3, model 3 shows that the number of battle deaths sustained in previous wars is positively associated with an increased government share of the economy later on. As noted above, we do not believe that there is an endogeneity problem with this measurement, as there is almost no correlation between the casualties suffered during World War I and/or World War II and the government share of the economy five years before the outbreak of these wars.

Turning to the three other aspects of the resource hypothesis, we find that both wars fought during periods of oil production (h4) and those fought without oil revenues increase the government share of the economy (model 4). But the size of the standardized coefficients (not shown in Table 3) differs substantially: non–oil funded wars have a beta coefficient of .177, while the coefficient of those fought by oil producers is about half the size (.085). Model 5 shows that wars increase the government share of the economy if the country did not receive foreign aid during the same period, but no such effects are noted for wars fought by aid-receiving countries. Wars not financed by any taxes have no state-building effects while tax-funded wars do (model 6). 78 In model 7, the effect of wars not funded by sources from abroad is 2.5 times stronger (with a beta coefficient of .339) than the effect of wars funded at least partly by sources from abroad, such as grants, credits, aid, or loot (the beta coefficient is .109). Note that models 6 and 7 are based on a reduced data set with only forty countries. [End Page 213]

The next two models explore the legitimacy aspect of the theory. Model 8 confirms that only wars that were won increase the government share of the economy (h5). Model 9 shows that wars fought after a state became ruled in the name of the nation, and no longer of a dynasty or empire, increase the government share of gdp, while wars fought before the transition to the nation-state even significantly decrease the government share of gdp (h6).

A couple of notes about this last finding are in order. The results of model 9 remain substantially the same if we reduce the sample to independent states, thus not considering colonial and imperial dependencies (results not shown). Since we control for regime type in all models, the effect of nationalist principles of legitimacy are not confounded with those of democracy. Finally, nation-states are not more likely to win wars than empires and dynastic states in this data set, thus avoiding confounding the effects of the two legitimacy mechanisms. 79 Unfortunately, however, we cannot distinguish between the legitimacy of the regime and that of the wars it fights, as noted above.

VI. Supplementary Analysis

Endogeneity and Confounding Mechanisms, Robustness Checks

As also mentioned above, an endogenous relationship might exist between democracy and nation-statehood on one hand and the frequency of war on the other. States that fought more wars may transition to democracy sooner since leaders will seek the credible commitment of the population to pay taxes. 80 States that fought many wars and thus developed bureaucracies that are able to interfere in the daily life of their subjects, raising taxes and recruiting soldiers, might produce a nationalist counter-reaction as subjects begin to demand being ruled by "our own people." 81 To clarify, these are not endogenous relationships between dependent and independent variables. Our argument is not that democracies or nation-states fight more wars, but that the wars they fight show more state-building consequences.

The problem, therefore, may be that our estimates of the moderating effects of democracy and nation-statehood on the state-building [End Page 214] effects of frequent wars may be distorted by the association between war frequency and the likelihood of a transition to nation-statehood and democracy. We thus implement a two stage instrumental variable regression, instrumenting democracy with the percentage of democracies in the vicinity of each country, as the comparative politics literature has done previously. 82

To explore whether the exclusion restriction is violated, we perform two placebo regressions that test whether the proportion democratic neighbors is associated with the government share of the economy in a subsample of highly democratic and a subsample of highly autocratic countries (using the combined autocracy/democracy scale of Polity IV). We do not find this to be the case (results not shown). The percent of democratic countries in the neighborhood has a beta coefficient of 0.77 in a regression of regime type and is thus highly relevant (the second criterion for a good instrumental variable besides the exclusion criterion). We then interact the posterior probability of being a democracy with the number of war onsets for each country-year.

Doing so allows us to create a new cumulative count of democratic wars. In the second stage of the IV regression, we included that new measure as an independent variable in a regression of government's share of the gdp. The coefficient is positive and statistically significant (see model 1 in Table 4). This finding reinforces our conclusion that wars make a more substantial contribution to state-building when fought by democratic countries—even if an endogenous relationship may exist between the frequency of war and transitioning to democracy.

To look at nation-statehood, we again used a two-stage instrumental regression approach by weighing wars by the posterior probability of being a nation-state. To calculate these posterior probabilities, we regress nation-state formation on those independent variables in Wimmer and Feinstein's study that are likely to satisfy the exclusion criterion (the instruments are listed in a note in Table 4). 83 We then create a cumulative measure of weighted war participation.

Model 2 in Table 4 shows that this new war count has a positive and statistically significant coefficient. This finding provides substantial support for the idea that the development of nation-states reinforced the link between war-making and state-building independent of the fact that the frequency of previous war (including anti-imperial [End Page 215]

Table 4. I D N-S F
Click for larger view
View full resolution
Table 4.

Instrumenting Democracy and Nation-State F ormation

wars of national liberation) enhanced the transition from empire to nation-state.

A similar potential problem is that our analysis may confound two mechanisms: Wars fought by democracies might lead to state-building not because institutionalized consent reduces resistance to taxation, as we argue, but because democracies choose to fight wars they can win, such that the association between wars fought by democracies and [End Page 216] state-building would come about indirectly, through the legitimacy of won wars. Similar to the analysis of the early modern period above, we address this issue by distinguishing between wars initiated by a state and those forced upon it, using two different data sources. Table 5 presents the findings.

They show that wars that were forced upon democratic governments have by far the strongest association with the government's gdp share. More precisely, the coefficients for wars initiated by democratic or nondemocratic regimes are positive and statistically significant in only one of the two models (model 1) and insignificant in the other (model 2). But the coefficient for wars forced upon democratic regimes

Table 5. C F E R G E /GDP W I F D N R
Click for larger view
View full resolution
Table 5.

Country Fixed Effects Regression of Government E xpenditure/GDP on Wars Initiated by or Forced upon Democratic and Nondemocratic Regimes

[End Page 217]

is significant in both models and roughly three times the size of the coefficient for democracy-initiated wars in model 1. Autocracies that have to defend themselves from attacks by other states are not seeing a subsequent boost in the government share of gdp (both in models 1 and 2) despite that a defensive war would arguably be perceived as legitimate as that fought by a democratic regime. This analysis thus helps to disentangle consent from legitimacy and suggests that the former's effect is independent from the latter.

Next, we address a possible omitted variable problem resulting from our approach to test each variable in a separate model. We proceed in this way because the samples differ considerably from variable to variable. A combined model would, therefore, result in substantial loss of observations. Data on foreign aid, for example, are available from 1960 onward only; those on the source of war financing for forty countries only, and so forth. More importantly, we do not include several variables in the same model because the cumulative war counts (e.g., number of wars fought as a democracy since 1860, the number of wars fought as a nation-state since 1860) are not mutually exclusive: the same war might be initiated by a democracy that is also a nation-state that is also an oil producer (e.g., Britain's Falklands War). Entering multiple war counts into the same model would thus produce substantial collinearity problems.

Our approach, although substantially meaningful, therefore creates a potential omitted variable problem: a focal variable in one model may be statistically significant because a more influential variable (that is examined in a separate model) is not included in the analysis. We find a partial solution to address this issue by creating war counts with cumulative, combined conditions (i.e., whether a war was initiated while being a democracy and nation-state or whether it was fought by an autocracy and a nation-state) and then comparing the explanatory power of several such combined war counts with each other. We focus on democracy, nation-statehood, and oil because these are relevant for the same list of wars and cover the entire sample, while the other conditions apply to subsamples only. As a word of caution, note that this is a rather indirect (and necessarily imperfect) way of testing an argument that essentially would amount to a three-way interaction in a fully specified model.

Appendix Table A2 in the supplementary material shows the results. We find that democracy is the most robust and most consistent moderator of the association between war frequency and the government's share of the gdp. This moderation persists even when controlling for [End Page 218] nation-stateness and oil (as shown in the first set of contrasts in Table A2 in the supplementary material). Second, being a nation-state is the second most significant moderator, but only if the country is not an oil producer (the second set of contrasts demonstrates this). Third, net democracy and nation-stateness, oil production does not seem to act as a moderator (this conclusion is based on the findings of the third pair of contrasts). Fourth, and most importantly, war frequency has the strongest association with the government's share of the gdp when fought by a country that is both a democracy and a nation-state (the last column in Table A2 highlights this finding). These additional results provide further evidence, in our understanding, for considering consent and legitimacy as major factors conjointly reinforcing the state-building effects of war.

Finally, we want to mention the results of a series of robustness checks with alternative measurements of key independent variables. Instead of a war count, we rerun the analysis with a war-year count, as one could argue that years at war is a more precise measurement of war efforts than is a simple war count. We present the findings in Appendix Table A3 in the online supplementary material. Results hold.

Three of the moderator variables are available in continuous forms: democracy, oil production, and foreign aid. In the above analysis, we dichotomized these (by coding wars fought as a democracy or not, for example) to explore interactions with the war count. We replicate the analysis with continuous versions of these three variables in Appendix Table A4 in the online supplementary material. The results are similar to those presented in Table 3 , except that the coefficient for wars fought with foreign aid is now statistically significant and negatively associated with the government share of gdp, further supporting our argument.

Appendix Table A5 reestimates the effect of war casualties by standardizing these by population, rather than the surface of a country. The coefficient remains positive and borderline significant (with a p value of .06; if we were to use a less stringent one-tailed test, the p value would be .03). Using population size for standardization reduces the sample considerably. We prefer to use the country size version in the main table.

Alternative Arguments: Global Wars, Initial S tate Capacity, and Postwar Security Environment

In this last empirical section, we briefly evaluate the three hypotheses, derived from the existing literature, that were not part of our revised theory of consent and legitimacy. To recall, these hypotheses state that [End Page 219] only global wars lead to state expansion and only by their winners (h7); that wars did not build states in the postcolonial world after World War II due to a changed international environment (h8); or that wars only built states if states were already somewhat centralized at the beginning of the period of intensifying warfare (h 9). Hypotheses 7 and 9 demand that we shift to a new dependent variable that is available for very long stretches of time and for the entire world. Starting the time series in 1860, when data on the government share of the economy become available, reduces the number of global wars to 1 (the two World Wars combined form one war according to the theory); and the period of intensifying war that hypothesis 9 refers to began in the early 1400s, hundreds of years before. We use the state antiquity index, as developed by Bockstette, Chanda, and Putterman, as a dependent variable to test these two arguments. 84

The index is coded separately for each 50-year period over the past 2,000 years and notes (1) whether a government "above the tribal level" ruled over the territory of today's countries; (2) how much that government was controlled by local elites rather than foreign imperial powers; and (3) how much of today's territory was governed by that government. The index runs from 0 to 50. It captures Tilly's understanding of the state quite well, which he defined as a "coercion-wielding organization that [is] distinct from households and kinship groups and exercises clear priority in some respects over all other organizations within substantial territories." 85

Table 6 shows the results of these additional analyses. Using the state antiquity index from 1801 to 1850 as a dependent variable and running a linear regression, model 1 does not support hypothesis 9: the state-building effects of early modern wars are not dependent on an already high level of stateness at the beginning of the fifteenth century (the same holds true if we restrict the sample to Western Europe; not shown). Model 2 uses country fixed-effects regression to analyze changes in countries' scores on the state antiquity index from 1400 to 1850. The findings indicate that the winners of global wars in Europe did not experience an increase in autonomous statehood (in contrast to h7), while all other wars did show such effects.

The next two models (3–4) evaluate whether the changed international norms and global political environment after World War II prevented the bellicose mechanism from operating during this period as [End Page 220]

Table 6. Testing Alternative Arguments
Click for larger view
View full resolution
Table 6.

Testing Alternative Arguments

[End Page 222]

well (h8). We do not find support for this conjecture: Wars built states both before World War II (model 3) and after (model 4; though the war variable just misses standard levels of significance in model 4 with a p value of .053 in a two-tailed test). A changed international environment, which guaranteed every state's security and continued existence, did not seem to reduce the incentives to build stronger states. Models 5–8, however, show that whereas prior to World War II, wars built states both in the West (model 5) and in the rest (model 6), in the postwar era, wars continued to build states in the West (model 7) but not in the rest (model 8). Without further research into the matter, it is unclear whether this is due to the role of aid and other nontax sources of war financing or because a changed international environment affected non-Western countries more than Western ones.

VII. Conclusion

Selectively building on, systematizing, and expanding the scholarship on war and state-building, this article introduced a unified theoretical framework to explain why and under which conditions frequent warfare enhanced the building of centralized states capable of extracting resources from the population. We argued that wars are more likely to build states where rulers and the ruled bargained an agreement to expand and further centralize the state during and after war. This situation is more likely to occur where institutions to negotiate such an agreement have emerged; where the population contributed manpower and taxes to the war effort; where rulers cannot rely on foreign aid, credit, or natural resources to finance their wars; and where the war appears more legitimate because it was won or fought to defend the nation's interests, rather than those of an empire or dynasty.

Under these conditions the state can expand its domains to raise taxes and to centralize functions hitherto assumed by other governing bodies or by religious and social organizations. What are the mechanisms through which these conditions increase state capacity? The consent of the population or its representatives reduces the likelihood that citizens will resist the expanding state by mounting armed rebellions or using "weapons of the weak" strategies to undermine its functioning in daily life. 86 The blood and taxes that the population offered during the war are rewarded during peacetimes with increased public goods provision. A won war provides legitimacy for the state and thus reduces [End Page 223] resistance to its expansion, while nationalism offers an ideological rationale for the state to take care of ever more needs of its people.

The theoretical model presented here integrates into a unified framework disparate arguments that have emerged in the past fifty years of debate. The model reinterprets these arguments to fit into the overall theoretical architecture and adds a series of new hypotheses (about the role of nationalism as a state ideology or of electoral democracy) derived from the theory.

We empirically evaluated this model, together with some other hypotheses that have been formulated in the literature, with data sets that cover a large number of states from around the world and across long spans of time. Going beyond existing quantitative scholarship that has been confined to smaller samples of countries and limited temporal spans, we developed two different measures of state-building from different sources and tested the argument with about twenty European and Asian states for the early modern period and with a sample of 116 states for the late nineteenth and twentieth centuries.

Important limitations remain, leaving aside the common problems of analyzing historical processes with quantitative data. First, expanding the measurements of tax revenues and assembly meetings beyond the nineteen cases that we use for the early modern period would be desirable (but extraordinarily difficult). Second, while our measurements of state-building improve on existing research, more fine-grained data on bureaucratic capacity, taxation, and domestic policing (the three crucial aspects of state capacity mentioned above) would be highly desirable—if extraordinarily difficult to generate for long periods of time and across the world. Inter alia, such data would allow us to include an analysis of the effects of colonial wars of conquest on state capacities around the world. Without such data and a thorough analysis of the role of imperialism, the global historical record of how wars and state-building relate to each other remains incomplete.

Third, to understand state-building beyond the war mechanism, other factors highlighted in the historical literature need to be taken into account, such as the nature of bureaucratic organizations on which state builders can rely or the political coalitions that support them. 87 Although we were able to control for some basic demographic, political, and historical forces that the literature has discussed, a full assessment of the various determinants of state formation and thus of the relative importance of the bellicose mechanism remains for future research to explore. [End Page 224]

Supplemental Materials

[ Click to Download (PDF) ]

Yuval Feinstein
Department of Sociology, University of Haifa, Haifa, Israel
Andreas Wimmer
Department of Sociology, Columbia University, New York, New York
Yuval Feinstein

Yuval Feinstein is an associate professor in the sociology department at the University of Haifa in Israel. His main line of research examines how complex, contested, situated, and dynamic national identities impact political attitudes and policy preferences in areas such as peace and wars, immigration, and inter-ethnic relations. His most recent book is Rally 'round the Flag: The Search for National Honor and Respect in Times of Crisis (Oxford, 2022). He can be reached at fyuval@soc.haifa.ac.il.

Andreas Wimmer

Andreas Wimmer is the Lieber Professor of Sociology and Political Philosophy at Columbia University. His research assumes a long-term historical and globally comparative perspective. It asks how states are built and nations formed, how ethno-racial boundaries and hierarchies form or dissolve in the process, and when these inequalities will lead to armed conflict and war. His most recent book is Nation Building: Why Some Countries Come Together While Others Fall Apart (Princeton University Press, 2018). He can be reached at andreas.wimmer@columbia.edu.

Corresponding author. E-mail: andreas.wimmer@columbia.edu

References

Acemoglu, Daron, and James A. Robinson. 2012. Why Nations Fail: The Origins of Power, Prosperity, and Poverty. New York, N.Y.: Crown Books.
Anderson, Perry. 1974. Lineages of the Absolutist State . London, UK: Verso.
Besley, Timothy, and Torsten Persson. 2008. "Wars and State Capacity." Journal of the European Economic Association 6, no. 2/3: 522–30. At https://www.jstor.org/stable/40282661.
Besley, Timothy, and Torsten Persson. 2009. "The Origins of State Capacity: Property Rights, Taxation, and Politics." American Economic Review 99, no. 4: 1218–44. At https://www.aeaweb.org/articles?id=10.1257/aer.99.4.1218.
Besley, Timothy, and Torsten Persson. 2011. Pillars of Prosperity: The Political Economics of Development Clusters. Princeton, N.J.: Princeton University Press.
Bockstette, Valerie, Areendam Chanda, and Louis Putterman. 2002. "States and Markets: The Advantage of an Early Start." Journal of Economic Growth 7: 347–69. At https://doi.org/10.1023/A:1020827801137.
Bonney, Richard. 2012. "The Rise of the Fiscal State in France, 1500–1914." In Bartolomé Yun-Casalilla and Patrick K. O'Brien, eds., with Francisco Comín Comín, The Rise of Fiscal States: A Global History, 1500–1914. Cambridge, UK: Cambridge University Press.
Brandt, Loren, Debin Ma Ma, and Thomas G. Rawski. 2014. "From Divergence to Convergence: Reevaluating the History behind China's Economic Boom." Journal of Economic Literature 52, no. 1: 45–123. At https://www.jstor.org/stable/24433858.
Brecke, Peter. 2012. Conflict Catalog (Violent Conflicts 1400 A.D. to the Present in Different Regions of the World).
Brubaker, Rogers. 2010. "Charles Tilly as a Theorist of Nationalism." The American Sociologist 41, no. 4: 375–81. At https://doi.org/10.1007/s12108-010-9107-9.
Bueno de Mesquita, Bruce, James D. Morrow, Randolph M. Siverson, and Alastair Smith. 1999. "An Institutional Explanation of the Democratic Peace." American Political Science Review 93, no. 4: 791–807. At https://doi.org/10.2307/2586113.
Cappella Zielinski, Rosella. 2016. How States Pay for Wars . Ithaca, N.Y.: Cornell University Press.
Centeno, Miguel A., Atul Kohli, and Deborah J. Yashar. 2017. "Unpacking States in the Developing World: Capacity, Performance, and Politics." In Miguel A. Centeno, Atul Kohli, and Deborah J. Yashar, eds., States in the Developing World. Cambridge, UK: Cambridge University Press: 1–34.
Centeno, Miguel Angel. 1997. "Blood and Debt: War and Taxation in Nineteenth-Century Latin America." American Journal of Sociology 102, no. 6: 1565–1605. At https://doi.org/10.1086/231127.
Chaudhry, Kiren Aziz. 1989. "The Price of Wealth: Business and State in Labor Remittance and Oil Economies." International Organization 43, no. 1: 101–45. At https://www.jstor.org/stable/2706728.
Cingolani, Luciana, Kaj Thomsson, and Denis de Crombrugghe. 2015. "Minding Weber More than Ever? The Impacts of State Capacity and Bureaucratic Autonomy on Development Goals." World Development 72: 191–207. At https://doi.org/10.1016/j.worlddev.2015.02.016.
Comín, Francisco Comín and Bartolomé Yun-Casalilla. 2012. "Spain." In Bartolomé Yun-Casalilla and Patrick K. O'Brien, eds., with Francisco Comín Comín, The Rise of Fiscal States: A Global History, 1500–1914. Cambridge, UK: Cambridge University Press: 233–66.
Coppedge, Michael, John Gerring, Carl Henrik Knutsen, Staffan I. Lindberg, Jan Teorell, David Altman, Michael Bernhard, M. Steven Fish, Adam Glynn, Allen Hicken, et al. 2020. "V-Dem [Country–Year/Country–Date] Dataset v10." At https://doi.org/10.23696/vdemds20.
Cowen, Deborah. 2008. Military Workfare: The Soldier and Social Citizenship in Canada. Toronto, Canada: University of Toronto Press.
Daunton, Martin. 2012. "The Politics of British Taxation, from the Glorious Revolution to the Great War." In Bartolomé Yun-Casalilla and Patrick K. O'Brien, eds., with Francisco Comín Comín, The Rise of Fiscal States: A Global History, 1500–1914. Cambridge, UK: Cambridge University Press: 111–42.
Desch, Michael C. 1996. "War and Strong States, Peace and Weak States?" International Organization 50, no. 2: 237–68. At https://www.jstor.org/stable/2704078.
Dincecco, Mark. 2011. Political Transformations and Public Finances: Europe, 1650–1913. Cambridge, UK: Cambridge University Press.
Dincecco, Mark, Giovanni Federico, and Andrea Vindigni. 2011. "Warfare, Taxation, and Political Change: Evidence from the Italian Risorgimento." The Journal of Economic History 71, no. 4: 887–914. At https://doi.org/10.1017/S002205071100218X.
Dincecco, Mark, and Mauricio Prado. 2012. "Warfare, Fiscal Capacity, and Performance." Journal of Economic Growth 17: 171–203. At https://doi.org/10.1007/s10887-012-9079-4.
Downing, Brian M. 1988. "Constitutionalism, Warfare, and Political Change in Early Modern Europe." Theory and Society 17: 7–56. At https://doi.org/10.1007/BF00163725.
Ertman, Thomas. 1997. Birth of the Leviathan: Building States and Regimes in Medieval and Early Modern Europe. Cambridge, UK: Cambridge University Press.
Fearon, James D., and David Laitin. 2014. "Does Contemporary Armed Conflict Have 'Deep Historical Roots'?" At https://leitner.yale.edu/sites/default/files/files/persistenceofarmedconflict3.pdf.
Finer, Samuel E. 1975. "State- and Nation-Building in Europe: The Role of the Military." In Charles Tilly, ed., The Formation of National States in Western Europe. Princeton, N.J.: Princeton University Press: 84–163.
Gill, Graeme. 2003. The Nature and Development of the Modern State. Basingstoke, UK: Palgrave Macmillan.
Golden, Miriam, and Brian Min. 2013. "Distributive Politics around the World." Annual Review of Political Science 16: 73–99. At https://doi.org/10.1146/annurev-polisci-052209-121553.
Gorski, Philip S. 2003. The Disciplinary Revolution: Calvinism and the Rise of the State in Early Modern Europe. Chicago, Ill.: University of Chicago Press.
Gupta, Bishnupriya, Debin Ma, and Tirthankar Roy. 2016. "States and Development: Early Modern India, China, and the Great Divergence." In Jari Eloranta, Eric Golson, Andrei Markevich, and Nikolaus Wolf, eds., Economic History of Warfare and State Formation . Singapore: Springer: 51–69.
Haldén, Peter. 2017. "The Realm as a European Form of Rule: Unpacking the Welfare Thesis through the Holy Roman Empire." In Lars Bo Kaspersen and Jeppe Strandsbjerg, eds., Does War Make States? Investigations of Charles Tilly's Historical Sociology. Cambridge, UK: Cambridge University Press: 154–80.
Hanson, Jonathan K., and Rachel Sigman. 2021. "Leviathan's Latent Dimensions: Measuring State Capacity for Comparative Political Research." The Journal of Politics 83, no. 4. At https://doi.org/10.1086/715066.
Hariri, Jacob Gerner. 2012. "The Autocratic Legacy of Early Statehood." American Political Science Review 106, no. 3: 471–94. At https://doi.org/10.1017/S0003055412000238.
Harris, Kevan. 2017. A Social Revolution: Politics and the Welfare State in Iran. Berkeley, Calif.: University of California Press.
Hechter, Michael. 2000. Containing Nationalism. Oxford, UK: Oxford University Press.
Henrickson, Phil. 2019. "Predicting the Costs of War." The Journal of Defense Modeling and Simulation: Application, Methodology, and Technology 17, no. 3: 285–308. At https://doi.org/10.1177/1548512919826375.
Herbst, Jeffrey. 1990. "War and the State in Africa." International Security 14, no. 4: 117–39. At https://doi.org/10.2307/2538753.
Heydemann, Steven, ed. 2000. War, Institutions, and Social Change in the Middle East. Berkeley, Calif.: University of California Press.
Hiers, Wesley, and Andreas Wimmer. 2013. "Is Nationalism the Cause or Consequence of Imperial Breakdown?" In John Hall and Siniša Malešević, eds., Nationalism and War. Cambridge, UK: Cambridge University Press.
Hoechle, Daniel. 2007. "Robust Standard Errors for Panel Regressions with Cross Sectional Dependence." The Stata Journal 7, no. 3: 281–312. At https://doi.org/10.1177/1536867X0700700301.
Jackson, Robert H. 1993. Quasi-States: Sovereignty, International Relations and the Third World. Cambridge, UK: Cambridge University Press.
Karaman, K. Kivanç, and Şevket Pamuk. 2013. "Different Paths to the Modern State in Europe: The Interaction between Warfare, Economic Structure, and Political Regime." American Political Science Review 107, no. 3: 603–26. At https://doi.org/10.1017/S0003055413000312.
Kaspersen, Lars Bo, and Jeppe Strandsbjerg, eds. 2017. Does War Make States? Investigations of Charles Tilly's Historical Sociology. Cambridge, UK: Cambridge University Press.
Kasza, Gregory J. 2002. "War and Welfare Policy in Japan." The Journal of Asian Studies 61, no.2: 417–35. At https://doi.org/10.2307/2700296.
Kisangani, Emizet F., and Jeffrey Pickering. 2014. "Rebels, Rivals, and Postcolonial State-Building: Identifying Bellicist Influences on State Extractive Capability." International Studies Quarterly 58, no. 1: 187–98. At https://www.jstor.org/stable/24017857.
Kiser, Edgar, and April Linton. 2001. "Determinants of the Growth of the State: War and Taxation in Early Modern France and England." Social Forces 80, no. 2: 411–48. At https://www.jstor.org/stable/2675585.
Kokkonen, Andrej, and Jørgen Møller. 2020. "Succession, Power-Sharing and the Development of Representative Institutions in Medieval Europe." European Journal of Political Research 59, no. 4: 954–75. At https://doi.org/10.1111/1475-6765.12381.
Krause, Keith. 1996. "Insecurity and State Formation in the Global Military Order: The Middle Eastern Case." European Journal of International Relations 2, no. 3: 319–54. At https://doi.org/10.1177/1354066196002003002.
Kroneberg, Clemens, and Andreas Wimmer. 2012. "Struggling over the Boundries of Belonging: A Formal Model of Nation Building, Ethnic Closure, and Populism." American Journal of Sociology 118, no. 1. At https://doi.org/10.1086/666671.
Leander, Anna. 2003. "Wars and the Un-Making of States: Taking Tilly Seriously in the Contemporary World." In Stefano Guzzini and Dietrich Jung, eds., Contemporary Security Analysis and Copenhagen Peace Research. Oxfordshire, UK: Routledge: 85–96.
Levi, Margaret. 1989. Of Rule and Revenue. Berkeley, Calif.: University of California Press.
Levi, Margaret. 1997. Consent, Dissent, and Patriotism. Political Economy of Institutions and Decisions. Cambridge, UK: Cambridge University Press.
Levi, Margaret. 1998. "Conscription: The Price of Citizenship." In Robert H. Bates, Avner Grief, Margaret Levi, Jean-Laurent Rosenthal, and Barry R. Weingast, eds., Analytic Narratives. Princeton, N.J.: Princeton University Press: 109–47.
Lewis, Joanna. 2000. Empire State-Building: War & Welfare in Kenya, 1925–52. Columbus, Ohio: The Ohio State University Press.
Maddison, Angus. 2003. The World Economy: Historical Statistics. Paris, France: OECD.
Magalhaes, Leandro de, and Francesco Giovannoni. 2019. "War and the Rise of Parliaments." Bristol Economics Discussion Papers 19/709. Bristol, UK: School of Economics, University of Bristol.
Mann, Michael. 1980. "State and Society, 1130–1815: An Analysis of English State Finances." In Maurice Zeitlin, ed., Political Power and Social Theory: A Research Annual. Greenwich, Conn.: Jaicon Press: 165–208.
Marshall, Monty G., Ted Robert Gurr, and Keith Jaggers. 2017. Polity IV Project: Political Regime Characteristics and Transitions, 1800–2016. Vienna, Austria: Center for Systemic Peace.
Mitchell, Brian R. 1998, 2003, 2016. International Historical Statistics (3 Volumes). New York, N.Y.: Palgrave.
North, Douglass C. 1981. Structure and Change in Economic History. New York, N.Y.: Norton.
O'Brien, Patrick Karl. 2006. "Contentions of the Purse between England and its European Rivals from Henry V to George IV: A Conversation with Michael Mann." Journal of Historical Sociology 19, no. 4: 321–43. At https://doi.org/10.1111/j.1467-6443.2006.00287.x.
Obinger, Herbert, and Carina Schmitt. 2011. "Guns and Butter? Regime Competition and the Welfare State during the Cold War." World Politics 63, no. 2 (April): 246–70. At https://www.jstor.org/stable/23018786.
Obinger, Herbert, and Klaus Petersen. 2017. "War and the Welfare State." In Berno Marin, ed., The Future of Welfare in a Global Europe. Basingstoke, UK: Routledge: 135–72.
OECD. 2016. Geographical Distribution of Financial Flows: Flows to Developing Countries. Paris, France: OECD.
Officer, Lawrence H., and Samuel H. Williamson. 2020. The Price of Gold, 1257–2014. At https://www.measuringworth.com/datasets/gold/.
Paul, T. V. 2014. The Warrior State: Pakistan in the Contemporary World. Cambridge, UK: Cambridge University Press.
Pezzolo, Luciano. 2012. "Republics and Principalities in Italy." In Bartolomé Yun-Casalilla and Patrick K. O'Brien, eds., with Francisco Comín Comín, The Rise of Fiscal States: A Global History, 1500–1914. Cambridge, UK: Cambridge University Press: 267–84.
Porter, Bruce D. 1994. War and the Rise of the State: The Military Foundations of Modern Politics. New York, N.Y.: Free Press.
Queralt, Didac. 2019. "War, International Finance, and Fiscal Capacity in the Long Run." International Organization 73, no. 4: 713–53. At https://doi.org/10.1017/S0020818319000250.
Rasler, Karen A., and William R. Thompson. 1985. "War Making and State Making: Governmental Expenditures, Tax Revenues, and Global Wars." American Political Science Review 79, no. 2: 491–507. At https://doi.org/10.2307/1956662.
Rasler, Karen A., and William R. Thompson. 1989. War and State Making: The Shaping of the Global Powers. Basingstoke, UK: Routledge.
Rasler, Karen A., and William R. Thompson. 2017. "War Making and the Building of State Capacity: Expanding the Bivariate Relationship." In William R. Thompson, ed., The Oxford Encyclopedia of Empirical International Relations Theory. Oxford, UK: Oxford University Press.
Rosenthal, Jean-Laurent. 1998. "The Political Economy of Absolutism Reconsidered." In Robert H. Bates, Avner Grief, Margaret Levi, Jean-Laurent Rosenthal, and Barry R. Weingast, eds., Analytic Narratives. Princeton, N.J.: Princenton University Press: 64–108.
Ross, Michael. 2012. The Oil Curse. How Petroleum Wealth Shapes the Development of Nations. Princeton, N.J.: Princeton University Press.
Sambanis, Nicholas, Stergios Skaperdas, and William C. Wohlforth. 2015. "Nation-Building through War." American Political Science Review 109, no. 2: 279–96. At https://doi.org/10.1017/S0003055415000088.
Sarkees, Meredith Reid, and Frank Whelon Wayman, eds. 2010. Resort to War: A Data Guide to Inter-State, Extra-State, Intra-State, and Non-State Wars, 1816–2007. Washington, D.C.: CQ Press.
Scheve, Kenneth, and David Stasavage. 2016. Taxing the Rich: A History of Fiscal Fairness in the United States and Europe. Princeton, N.J.: Princeton University Press.
Scott, James C. 1990. Domination and the Arts of Resistance . New Haven, Conn.: Yale University Press.
Skocpol, Theda. 1995. Protecting Soldiers and Mothers . Cambridge, Mass.: Harvard University Press.
Sørensen, Georg. 2001. "War and State-Making: Why Doesn't It Work in the Third World?" Security Dialogue 32, no. 3: 341–54. At https://www.jstor.org/stable/26298020.
Spruyt, Hendrik. 1996. The Sovereign State and its Competitors: An Analysis of Systems Change. Princeton, N.J.: Princeton University Press.
Spruyt, Hendrik. 2017. "War and State Formation: Amending the Bellicist Theory of State Making." In Lars Bo Kaspersen and Jeppe Strandsbjerg, eds., Does War Make States? Investigations of Charles Tilly's Historical Sociology. Cambridge, UK: Cambridge University Press.
Stasavage, David. 2010. "When Distance Mattered: Geographic Scale and the Development of European Representative Assemblies." American Political Science Review 104, no. 4: 625–43. At https://doi.org/10.1017/S0003055410000444.
Stasavage, David. 2016. "Representation and Consent: Why They Arose in Europe and Not Elsewhere." Annual Review of Political Science 19: 145–62. At https://doi.org/10.1146/annurev-polisci-043014-105648.
Teschke, Benno. 2017. After the Tilly Thesis: Social Conflict, Differential State Formation and Geopolitics in the Construction of the European System of States. In Lars Bo Kaspersen and Jeppe Strandsbjerg, eds., Does War Make States? Investigations of Charles Tilly's Historical Sociology. Cambridge, UK: Cambridge University Press: 25–51.
Thies, Cameron G. 2004. "State-Building, Interstate and Intrastate Rivalry: A Study of Post-Colonial Developing Country Extractive Efforts, 1975–2000." International Studies Quarterly 48, no. 1: 53–72. At https://www.jstor.org/stable/3693563.
Thies, Cameron G. 2006. "Public Violence and State-Building in Central America." Comparative Political Studies 39, no. 10: 1263–82. At https://doi.org/10.1177/0010414005284380.
Thies, Cameron G. 2007. "The Political Economy of State-Building in Sub-Saharan Africa." The Journal of Politics 69: 716–31. At https://doi.org/10.1111/j.1468-2508.2007.00570.x.
Tilly, Charles. 1975. "Reflections on the History of Western State-Making." In Charles Tilly, ed., The Formation of National States in Western Europe . Princeton, N.J.: Princeton University Press: 3–83.
Tilly, Charles. 1985. "War Making and State Making as Organized Crime." In Catherine Bestemen, ed., Violence: A Reader. New York, N.Y.: New York University Press: 35–60.
Tilly, Charles. 1992. Coercion, Capital, and European States, AD 990–1990. Oxford, UK: Blackwell.
van Zanden, Jan Luiten, Eltjo Buringh, and Maarten Bosker. 2012. "The Rise and Decline of European Parliaments, 1188–1789." The Economic History Review 65, no. 3: 835–61. At https://www.jstor.org/stable/23271553.
Wang, Erik H., and Yiqing Xu. 2018. "Awakening Leviathan: The Effect of Democracy on State Capacity." Research & Politics 5, no. 2. At https://doi.org/10.1177/2053168018772398.
Wimmer, Andreas, and Brian Min. 2006. "From Empire to Nation-State. Explaining Wars in the Modern World, 1816–2001." American Sociological Review 71, no. 6: 867–97. At https://doi.org/10.1177/000312240607100601.
Wimmer, Andreas, and Yuval Feinstein. 2010. "The Rise of the Nation-State across the World, 1816 to 2001." American Sociological Review 75, no. 5: 764–90. At https://doi.org/10.1177/0003122410382639.
Wimmer, Andreas. 2002. Nationalist Exclusion and Ethnic Conflicts: Shadows of Modernity. Cambridge, UK: Cambridge University Press.
Yun-Casalilla, Bartolomé, and Patrick O'Brien, eds, with Francisco Comín Comín. 2012. The Rise of Fiscal States: A Global History, 1500–1914. New York, N.Y.: Cambridge University Press.

Acknowledgments

Both authors contributed equally to this article. We are grateful to the colleagues who shared their (sometimes unpublished) data on early modern taxes and assemblies or gave us advise on where to find them: Scott Abramson and Carles Boix, Marc Dincecco, Kivanç Karaman, Andrej Kokkonen, Jørgen Møller, Didac Queralt, and David Stasavage. Rarely have we encountered such a generous community of scholars. We also acknowledge the advice and feedback given by Rhiannon Stephens and Didac Queralt as well as the comments and critical suggestions made by the audiences at All Souls College of Oxford, where Stathis Kalyvas invited us to present this paper; the interdisciplinary Political Violence and its Legacies workshop at Yale University, to which we were invited by Jason Lyall; the Department of Political Science at mit, where Evan Lieberman suggested we give a talk; the colloquium of the Sociology Department at the University of California Irvine convened by Evan Schofer and Kristin Turney; the departments of political science and sociology at the European University Institute in Florence for which Ellen Immergut offered an invitation; as well as the annual meeting of the American Sociological Association in Philadelphia. The literature review part of the paper was also presented at a presidential panel dedicated to the work of Charles Tilly under the auspices of the president of the Eastern Sociological Association, Alejandro Portes. Last but not least, we gratefully acknowledge the outstanding research assistance of Eva Beeman, Nicholas Pang, and Durana Saydee who all worked on the war data set.

Footnotes

7. For empirical support, see Karaman and Pamuk 2013.

8. A rare exception is Teschke 2017.

11. Gorski 2003. For other critical arguments, which we do not explore in this study, see Gill 2003, 154–56.

18. Ertman's analysis is more complex, however, as he seeks to explain not only levels of bureaucratization of the state but also political regimes (i.e., absolutism versus constitutionalism).

22. Obinger and Petersen 2017. See also the revised argument and findings of Rasler and Thompson 2017.

24. On one hand, Tilly states that "the tendency for war to build state structure hold[s] through much of world history"; Tilly 1992, 15. On the other hand, he offers a series of reasons why this could be less the case in the postcolonial world. What Tilly seems to have in mind is that wars continue to build states, but different kinds of states: military dictatorships rather than democratic national states as was the case earlier in Western Europe.

25. Thies 2004 is an exception.

27. Sørensen 2001; Desch 1996 ; Herbst 1990; Krause 1996; Kisangani and Pickering 2014; for an exception, see the state-building effects of the Iran-Iraq War, described by Harris 2017.

31. E.g., Tilly 1985; see also Brubaker 2010.

34. For a formal model, see Kroneberg and Wimmer 2012.

36. Levi 1989; for empirical evidence, see Dincecco 2011. See Comín and Yun-Casalilla 2012 . Besley and Persson 2011; for empirical evidence for the nineteenth century, see Dincecco 2011, 108–20.

37. For a formal model, see Levi 1998.

39. Most recently, see Queralt 2019. For a dissenting view, see Downing 1988.

40. On the effects of war on welfare development, see Cowen 2008; Kasza 2002; Skocpol 1995 ; Lewis 2000; Obinger and Schmitt 2011; Dincecco 2011.

41. For an attempt to explain battle deaths in interstate war, see Henrickson 2019.

44. Ruling elites that won a war also dispose of more resources to keep state activities at the war level (rather than paying for reparation, for example).

49. For 1975 onward, see Besley and Persson 2008; Besley and Persson 2009. For one or two continents, see Thies 2004; Thies 2007. For Central American countries, see Thies 2006. For the small sample of thirty-nine countries, see Rasler and Thompson 2017 . For European countries in the early modern period, see Dincecco 2011; Karaman and Pamuk 2013.

53. Mann 1980; for a critical discussion, see O'Brien 2006.

54. Mitchell various years.

55. Data on GDP are from Maddison 2003. We note here that using government expenditure per capita, rather than per GDP, produced nearly identical results.

56. Cappella Zielinski 2016. Results available upon request.

58. Brecke 2012. The data have recently been used by other political scientists; see Fearon and Laitin 2014.

69. We also re-estimated the models in Appendix Table A1 with a lagged dependent variable as an additional predictor. Except for one variable (number of battle deaths per territory size), all results (not shown) hold.

72. For four European countries (there is a contemporaneous association only between wars that were lost and the calling of parliament, however; this association may suggest that assemblies were called after a financially ruinious war, rather than before); Magalhaes and Giovannoni 2019. Stasavage does not find an association between war frequency and the number of parliamentary meetings for a larger sample of states; Stasavage 2010.

73. The military history of most countries includes offensive and defensive wars. Including only the latter therefore creates a risk of biased estimates due to an omitted variable problem. To test the robustness of the findings in Table 2, we added a variable counting the number of wars initiated by each policy in each period. The focal interactions coefficients in Table 2 maintain direction and statistical significance in this more advanced analysis (not shown here).

74. In an additional analysis (results available upon request), we recoded the dependent variable as the tax revenue in the last year of a period (or the average of the last few years if annual data were unavailable) and removed the period lag from the independent variable. Results hold even when we added lagged tax revenue figures as an additional independent variable.

75. The analysis in Table 2 is also limited by the reduced sample size due to incomplete information on tax revenues for many periods.

76. This is in line with the findings of Dincecco and Prado 2012.

78. We control for democracy in all models of Table 3. The effect of source of war funding is therefore net regime type. However, regime type and funding sources are correlated: democracies rely more on taxes (85 percent of wars in Capella Zielinski's data set) than autocracies (53 percent of wars). See https://sites.bu.edu/cappella/confronting-the-cost-of-war-data/.

79. But see Hiers and Wimmer 2013 for a larger sample of countries and years.

80. According to Queralt 2019.

Share