Weak, Despotic, or Inclusive? How State Type Emerges from State versus Civil Society Competition

Abstract

We develop a theory of the accumulation of state capacity as the outcome of a political competition between elites and (civil) society. State capacity is accumulated by elites, and it is productive as well as useful in controlling society. However, society can fight back and accumulate its own capacity, facilitating collective action. The theory leads to three distinct equilibria depending on initial conditions. One type, a weak state, emerges when society is strong relative to the elite. Another, a despotic state, originates where the elite is initially relatively powerful. A third type, an inclusive state, emerges when the elite and society are more evenly matched. The theory has several important implications; first, variation in state capacity does not require large structural differences; second, inclusive states have the highest levels of state capacity in the long run; third, the effects of shocks or external threats like wars are conditional on the balance of power between elites and society.

Introduction

There is a great deal of variation in state capacity around the world. Simple statistical measures, such as the ratio of government tax revenues to national income, vary from close to 50% in western and northern Europe to less than 10% in many countries in sub-Saharan Africa. These differences in resources are reflected in large differences in the provisions of public goods, infrastructure, and the ability to deliver justice or hold clean elections.

Such dispersion has enormous consequences for politics. The absence of state capacity has been argued to be the main reason that societies fall into civil war (Fearon and Laitin 2003), the primary explanation for the absence of accountability and institutions of participation (Herbst 2000; Levi 1989; Tilly 2007), and an important cause of the inability of communities to collectively govern resources in desirable ways (Ostrom 1990). The presence of state capacity is argued to be the main driving force behind experiences of rapid economic growth (Amsden 1989; Evans 1995; Wade 1990) and more broadly to guarantee that the state works in the collective interest (Geddes 1994).

In this paper we propose a new theory to explain this variation. Existing explanations account for it either by structural factors such as population density, topography, or culture or via more contingent influences such as histories of warfare, colonialism, or trade. Yet, as we illustrate below, polities with very similar structural features and histories of warfare have experienced dramatic divergences in state capacity.

Our theory conceptualizes the construction of state capacity as a game between those in control of the state, who we call the elite, and civil society—or henceforth “society” for short. State capacity, accumulated by the elite, potentially has benefits for all in the sense that it allows for the provision of public goods, but at the same time it can be used to oppress citizens and thus slant all the benefits toward the elite. This trade-off is recognized in the Western world at least since the time of Locke. As Scott (2010) and Acemoglu and Robinson (2019) discuss, it is well understood in non-Western societies too. But citizens also have capacity in this game with the elite—we conceptualize this as the ability to organize and engage in collective action. Like state capacity accumulated by elites, greater societal capacity adds to productivity because it allows greater coordination and public good provision (Ostrom 1990). But it simultaneously gives society greater ability to contest with the elite (emphasized by Wood 2003).

Specifically, we assume that state and society’s capacities are used as inputs into a contest for power. Elites accumulate state capacity, which enables them to impose their wishes on and dominate society, whereas society’s capacity empowers it to resist elite schemes (which may take the form of democratic political participation, protests, collective action, strikes, and even violence). Our notion of a contest for power and the model we use is common in the literature on interest group politics, civil wars, and international relations (see Powell 1999; Skaperdas 1992; Tullock 1980). As in these literatures, we model the outcome as uncertain; for given capacities, which side wins depends on a variety of contingent factors that cannot be anticipated in advance. Whoever wins is then able to set policy and allocate resources in their own interests.

We show that there are three long-run equilibrium outcomes in this contest. ¹ Elites manage to accumulate far more capacity than society and dominate it. This leads to a despotic state. Or, in the opposite case, society dominates elites and very little state capacity develops, and a weak state emerges. In the middle, where the capacity of the state and society is balanced a third equilibria emerges where both the state and society end up with large amounts of capacity. We call this an inclusive state.

These long-run equilibrium outcomes and the dynamics leading up to them are summarized in Figure 1. We depict society’s capacity on the horizontal axis and state capacity on the vertical one. Region I illustrates the political dynamics leading to a despotic state, similar to those of early modern Prussia, a case we discuss in detail in our historical perspective section. Region III is the case of a weak state, which we illustrate with the history of Montenegro contrasted to Prussia’s. Finally, Region II depicts the middle ground where the elite and society are initially in balance, and this triggers an ongoing competition between the two, leading to an inclusive state, which we illustrate with Switzerland.

Figure 1. The Emergence and Dynamics of Weak, Despotic, and Inclusive States

This comparison is not intended to indicate the scope conditions of the model but is simply a use of Mill’s (1872) most similar research design. More broadly, our claim is that the mechanisms underlying the dynamics and equilibria of the model help to account for the patterns we see in the world with the simultaneous persistent existence of despotic states (like China), weak states (as in many parts of sub-Saharan Africa), and inclusive states (in western and northern Europe or north America). Existing theories can explain this pattern only by appealing to significant regional or continental differences in underlying conditions. Without denying the role of such differences, our ambition is to develop a global theory that can account for these rich patterns without huge structural differences across countries and regions.

Our theory and Figure 1 clarify that small differences in initial conditions can put a polity into the basin of attraction of one equilibrium rather than another. We believe this feature of the model is consistent with historical evidence, as we discussed later.

An important implication of our theory is visible in Figure 1: state capacity is greater under inclusive states than under despotic states. This is because when the state dominates society, it has less incentive to accumulate further capacity. It is competition between state and society that triggers greater investments by the elites controlling it. Our model thus highlights that weak, despotic, and inclusive states have not only different amounts of state capacity but also divergent societal capacities.

Another feature of our theory is clear from Figure 1: the effects of a rich array of structural factors are conditional. For example, factors that (exogenously) increase the power of the state could move a polity from Region III to Region II, initiating a powerful process of state capacity building. But as Figure 2 illustrates, the same factors may also push a polity previously in Region II into Region I, reducing its long-run potential to achieve high state capacity. Similarly, structural factors shift the boundaries of the basins of attraction of the three different types of states as well, but the effects of such changes are also conditional on the prevailing balance between state and society. These conditional comparative statics highlight that there is no simple version of Tilly’s “War made the state, and the state made war” (1975, 42) in our theory. Though warfare may create incentives to strengthen state capacity, its ultimate implications depend on the balance of power between state and society. This provides one explanation for why for each structural factor argued to underlie the development of state capacity, there are always counterexamples going in the opposite direction. ²

Figure 2. Conditional Comparative Statics: The Same Impulse Leads to Very Different Trajectories of Political Development

Although our theory allows for diverse long-term outcomes, it also generates several falsifiable predictions. First, the greatest state capacity emerges not when the state is overwhelmingly dominant and strong relative to society but when there is some sort of balance between the two. This would imply that one should find a nonmonotonic relationship between initial state capacity (given society’s organization) and ultimate state capacity. Second, the longer a society remains in the basin of attraction of the despotic state (or weak state), the more difficult it is for it to transition to an inclusive state. This perspective implies, for example, that in contrast to optimistic predictions about China’s political development and seamless transition to democracy based on its modernization, building inclusive state–society relationship in China will be very hard. Third, the model implies a type of weak monotonicity in initial conditions in response to shocks, which can be investigated empirically. For example, of two polities that are subject to exactly the same pro-state shock and have the same societal capacity, the one with greater state capacity is more likely to transition to a despotic state than the other and less likely to remain in the weak state’s basin of attraction.

Our theory builds on a few main ideas. First, that state capacity can be used productively, but also to control society, and that it can be accumulated. By state capacity, most scholars build on Mann’s (1993) notion of “infrastructural power,” which he defines as the capacity of a central state to “penetrate its territories and logistically implement decisions” (59). Such capacity is multidimensional and includes bureaucratic, legal, and fiscal capacities, and even the latter is itself complex, as Fauvelle-Aymar (1999) points out, as it involves not just the construction of state institutions but also people’s willingness to pay taxes, which may depend on the extent to which they see the state as legitimate or efficient. Though our model is reduced form and consistent with different interpretations of state capacity, fiscal capacity does satisfy our three conditions; it can be used to provide public goods and to equip and pay state agents of control, and more of it can be acquired via institution building, conducting surveys of assets and wealth, etc. (even if people may resist, à la Fauvelle-Aymar 1999). It is also the dominant measure in the literature (e.g., Herbst 2000; Slater 2010) and lends itself to measurement.

We pit this against society’s capacity. Our notion of societal capacity emphasizes society’s ability to solve collective action problems and define and achieve collective goals. Like our definition of state capacity, this is productive, can be used in contestation, and it can be built. Tilly (1995) studied the emergence of this capacity in eighteenth-century Britain, showing how initially “contention” was about “local people and local issues, rather than nationally organized programs and parties” (5). However, “between 1758 and 1833 a new variety of claim making had taken shape… . Mass popular politics had taken hold on a national scale” (13). Critically, this new politics was not just about new and collective issues (instead of parochial ones); it was also organized in very different ways, frequently in the form of “a special purpose association, society or club” (10). Our model captures what Tilly thought was the force driving this reorganization of society: “the contemporary reorganization … of the state, that re-shaped the state’s repressive apparatus” (23), along with expansions of “revenue, expenditure and personnel” (49), which induced on the part of society “a shift towards collective action that was large in scale and national in scope” (49).

Ostrom’s work identified aspects of social institutions that are critical for solving collective action problems and building capacity (1990). These include: autonomous organization, monitoring capability, availability of sanctions, and effective conflict resolution mechanisms. Society lacks capacity when there is “no capacity to communicate with one another, no way to develop trust, and no sense that they share a common future” (90). Ostrom also stressed that “Such groups may need some form of external assistance to break out of the perverse logic of their situation” (21), which often comes from interactions with the state, or as she puts it “are affected by the surrounding political regime” (141). States often influence the creation of institutions to solve collective action problems “by creating and limiting the powers that can be exercised with collective-choice arrangements (creating legislative and judicial bodies, protecting rights of free speech and property etc.)” (192). Her case studies confirmed that “the orientation of the ruling political regime can make a substantial difference” (212). Recent work on collective action in civil war echoes many of these themes. Steele shows that communities in Colombia have managed to organize collectively to resist armed groups when they developed “a combination of a strict internal hierarchy and external support” (2017, 173). Arjona’s study of local order during the Colombian civil war similarly illustrates how “the quality of dispute institutions plays a central role in fostering the community’s capacity for collective action” (2016, 71). Institutions further facilitate cohesion if they can target goods and services as emphasized by Cammett (2014) and Cammett and McLean (2014).

Societal capacity also depends on cultural and behavioral factors identified by Wood (2003). In her study of the Salvadorean Civil War, collective action depends on people’s “moral commitments and emotional engagements” (18). People acted collectively “as an act of defiance of long-resented authorities and a repudiation of perceived injustices” (18). As in our theory, this type of capacity building is dynamic, as “political culture—the values, norms, practices, beliefs, and collective identity of insurgents—was not fixed but evolved in response to the experiences of the conflict itself” (41), and defiance “motivated further collective action through a recursive process” (238). Wood’s notion of society’s capacity is complemented by Viterna’s (2013) theory of how people can assume a “participation identity,” selecting from the available repertoire of identities (Swidler 1986) one that is highly congruent with collective action. As in Wood’s analysis, this can be precipitated by the influence of political institutions. “In El Salvador, the military’s attacks on rural civilians gave new meanings to existing identities” (54), which tipped people toward participation in the conflict.

A central assumption in our model is that the accumulation of both state and societal capacity is subject to (dynamic) increasing returns to scale. In particular we assume that the marginal cost of building capacity is higher below a certain threshold than afterwards. Intuitively, once one has sufficient capacity, it becomes easier to acquire more. For example, building a nascent fiscal system involves large fixed costs. Detailed studies of this, such as Brewer (1988), show that to build an effective excise tax system in Britain many elements had to be in the place so that personnel can be trained, paid, monitored, and stationed throughout Britain. Our assumption is that if one has these elements in place, then it becomes easier and cheaper to build further fiscal capacity (see also Dharmapala, Slemrod, and Wilson 2011). Our argument for society is similar: as Tilly’s (1995) study shows, more effective contention required organization and the creation of new institutions. Once these had been put in place, it subsequently became easier to accumulate greater capacity. Both Wood’s and Viterna’s theories imply increasing returns to the accumulation of what we are calling society’s capital. This is evident in Wood’s formal model (2003, 267–74) and from the peer effects connected to the formation of a collective participation identity in Viterna (2013). Increasing returns to collective action are also emphasized in Marwell and Oliver (1993) and Pearson (2000).

Our paper is closely related to several important lines of research within comparative politics. Though our findings do not support the idea that divergence in state capacity necessitates large structural differences or is a consequence of the incidence of warfare, they are consistent with other key ideas. Many scholars have studied the idea that elites and society compete and that state capacity can emerge from what Wood calls this “recursive process.” This is evident from our discussion of Tilly (1995) and central to Skocpol’s (1979) theory of social revolutions and their consequences. She defines social revolutions as “rapid, basic transformations of a society’s state and class structures, accompanied and in part carried through by class-based revolts from below” (33) and emphasizes not just “the changes that social revolutions make in the structure and function of states” (164) but also the mutual feedback such that “the changes in state structures that occur during social revolutions typically both consolidate, and themselves entail, socioeconomic changes” (164). ³ Our model formalizes and extends these ideas and mutual feedback loops and shows that mechanisms typically studied in different contexts have common roots and major implications for the distribution of state capacity. For example, our weak state configuration captures Migdal’s (1988) and Scott’s (2010) ideas that week states emerge from society’s strength. Yet, in contrast to their emphasis, our theory also reveals that, in the inclusive state configuration, society is even “stronger” (using Migdal’s terminology) when the state is stronger. Our theory’s implications also contrast with Huntington’s (1968) and Fukuyama’s (2011) claims that state capacity emerges under the auspices of powerful leaders and groups and that there is a specific sequence toward a democratic strong state—state strength first, democracy later. This is not a prediction of our model; when the state and elites become too strong, the development of state capacity and democracy (see below) is arrested. Our theory also nests Grzymala-Busse’s (2007) theory that state capacity emerged in the post-Soviet world when there was competition between political parties (as in Hungary and Poland) but did not when one party dominated (as in the Czech Republic or Slovakia). ⁴

Much scholarship in comparative politics has attempted to explain not just patterns of state building but also their consequences for regimes. Herbst (2000), building on ideas initially developed by Levi (1989), argues that state weakness in Africa leads to nondemocracy because states that do not raise taxes are not forced to negotiate with their citizens. Ertman (1997) develops a dichotomy of state capacity (bureaucratic or patrimonial), combining it with regime types (constitutional or autocratic rule). Though we do not formally model the process of democratization, our theory implies that democracy cannot easily arise under despotic or weak states. In the former, powerful elites can block the participation of society (as in many standard theories of democracy (e.g., Acemoglu and Robinson 2006; Rueschemeyer, Stephens, and Stephens 1992). In the latter, the weakness of the central state does not allow for the emergence of effective democratic institutions. It is therefore in the middle, where there is state capacity, but it is forced to be responsive to a capacitated society that one would expect democracy to emerge.

In the historical perspective section, we also develop an Early Modern European case study in the spirit of Mill’s (1872) “most similar” research design (Skocpol 1979, discusses the strengths and weaknesses of this approach). Our case selection focuses on the large differences within Europe in terms of our main dependent variable. Why did Switzerland develop an inclusive state that had capacity but was very accountable to its people, whereas Prussia created a despotic state with less capacity? Why did Montenegro never really create a state, except a powerless theocracy, until the twentieth century?

We develop this case study as two linked pairs, Prussia–Switzerland and Switzerland–Montenegro. This design is “most similar” in the sense that we emphasize both common institutional histories (Prussia–Switzerland) and social structures and ecologies (Switzerland–Montenegro). Prussia and Switzerland in 1600 (before Brandenburg had merged with Prussia) were both part of the Holy Roman Empire and had inherited several political and economic institutions in common from the Germanic tribes, the Carolingians, and feudalism. Switzerland and Montenegro had deep institutional roots in common (Roman Empire), but our focus is on social structures (clans) and also mountainous ecology. The role of clans is well established in Montenegro, of course, but it is also a common theme in Swiss studies. For example, Steinberg (2016, 17) notes, “[Swiss] medieval clan structures had little to do with our images of democratic forms but these peasants were ‘free,’” and he notes the extent of communal activities and organization. It is also interesting the extent to which the early development of the Old Swiss Confederacy was focused on conflict resolution and the management of feuds, which were an incessant problem in Montenegro as well. The main difference between the cases is the initial strength of the state: no state in Montenegro, a nascent, noncentralized state in Swiss cantons, and a stronger and more centralized state in Brandenburg (Prussia). In our theory it is these relatively small differences in the initial strength of the state that led to the very different dynamics and outcomes.

Finally, our theory is built on the theory of dynamic contests (e.g., Hirshleifer 1989). Our results on stronger incentives to build capacity when elites and citizens are evenly matched are related to the discouragement effect in contests emphasized in Harris and Vickers (1985).

The rest of the paper is organized as follows. In the next section, we introduce our main model. Next, we characterize the dynamic equilibrium and steady states of this model. To maximize transparency, this section uses a number of simplifying assumptions, many of which are relaxed in the Appendix. We then outline how the same model with forward-looking players leads to similar results. The historical perspective section discusses our case study of early modern European state divergence. The last section concludes, and the online Appendix presents the proofs of the results stated in the text, details of several generalizations mentioned in the text, some additional technical material, and also a generalization of our model in which the inclusive state becomes feasible only after the capacities of both the state and society are above certain thresholds.

Basic Model

In this section, we introduce our basic model, which is then analyzed in the next several sections.

Preferences and Conflict

We start with a discrete time setup, where period length is ∆ > 0 and will later be taken to be small so that we work with differential rather than difference equations in characterizing the dynamics. At time t, the stock variables inherited from the previous period are $ \left({x}_{t-\Delta},{s}_{t-\Delta}\right)\in {\left[0,1\right]}^2 $ , where the first element corresponds to the capacity of society and the second to state capacity controlled by the elite. ⁵

At each point, the elite is represented by a single player and society is also represented by a single player. In the next two sections, we study both the case in which these players are short-lived and are immediately replaced by another player (so that we have a nonoverlapping generations model with myopic players) and the case in which players are long-lived and maximize their discounted sum of utilities.

At time, players simultaneously choose their investments, $ {i}_t^x\hskip0.35em \ge \hskip0.35em 0 $ and $ {i}_t^s\hskip0.35em \ge \hskip0.35em 0 $ , which determine their current capacity according to the equations:

(1) $$ {x}_t\hskip0.35em =\hskip0.35em {x}_{t-\Delta}+{i}_t^x\Delta -\delta \Delta, $$

and

(2) $$ {s}_t\hskip0.35em =\hskip0.35em {s}_{t-\Delta}+{i}_t^s\Delta -\delta \Delta, $$

where $ \delta >0 $ is the depreciation of the capacities of both parties between periods. Both investment and depreciation are multiplied by the period length, ∆, as they represent “flow” variables, and when period length is taken to be small, they will be suitably downscaled. ⁶

The cost of investment for society during a period of length ∆ is given as $ \Delta \cdot {\tilde{C}}_x\left({i}_t^x,{x}_{t-\Delta}\right) $ , where

$$ {\tilde{C}}_x\left({i}_t^x,{x}_{t-\Delta}\right)\hskip0.35em =\hskip0.35em \left\{\begin{array}{cc}{c}_x\left({i}_t^x\right)& \hskip0.2em \mathrm{if}\;{x}_{t-\Delta}>{\gamma}_x,\\ {}{c}_x\left({i}_t^x\right)+\left({\gamma}_x-{x}_{t-\Delta}\right){i}_t^x& \mathrm{if}\;{x}_{t-\Delta}\hskip0.35em \le \hskip0.35em {\gamma}_x.\end{array}\right. $$

This cost function is multiplied by ∆, as it is the cost of investing an amount $ {i}_t^x $ during the period of length ∆, as captured by Equation 1. The presence of the term $ {\gamma}_x>0 $ , on the other hand, captures the “increasing returns” nature of capacity accumulation mentioned in the Introduction: starting from a low level of capacity, it is more costly to build up this capacity. We specify this in a very simple form here, with the cost of investments increasing linearly as last period’s capacity falls below the threshold $ {\gamma}_x $ . This increasing returns aspect plays an important role in our analysis as we emphasize below.

The cost of investment for the elite during a period of length ∆ is similarly given as $ \Delta \cdot {\tilde{C}}_s\left({i}_t^s,{s}_{t-\Delta}\right) $ , where

$$ {\tilde{C}}_s\left({i}_t^s,{s}_{t-\Delta}\right)\hskip0.35em =\hskip0.35em \left\{\begin{array}{cc}{c}_s\left({i}_t^s\right)& \mathrm{if}\;{s}_{t-\Delta}>{\gamma}_s,\\ {}{c}_s\left({i}_t^s\right)+\left({\gamma}_s-{s}_{t-\Delta}\right){i}_t^s& \hskip-0.1em \mathrm{if}\;{s}_{t-\Delta \hskip0.35em }\le \hskip0.35em {\gamma}_s.\end{array}\right. $$

In these expressions, it will often be more convenient to eliminate investment levels and directly work with the two stock variables, $ {x}_t $ and $ {s}_t $ , especially when we take ∆ to be small and transition to continuous time. In preparation for this transition, let us substitute out the investment levels and observe that the cost function for society and state can be written as

$$ {C}_x\left({x}_t,{x}_{t-\Delta}\right)={c}_x\left(\frac{x_t-{x}_{t-\Delta}}{\Delta}+\delta \right)+\max \left\{{\gamma}_x-{x}_{t-\Delta},,0\right\}\left(\frac{x_t-{x}_{t-\Delta}}{\Delta}+\delta \right), $$

and

$$ {C}_s\left({s}_t,{s}_{t-\Delta}\right)\hskip0.35em =\hskip0.35em {c}_s\left(\frac{s_t-{s}_{t-\Delta}}{\Delta}+\delta \right)+\max \left\{{\gamma}_s-{s}_{t-\Delta},0\right\}\left(\frac{s_t-{s}_{t-\Delta}}{\Delta}+\delta \right), $$

where the increasing returns to scale nature of the cost function is now captured by the max term. ⁷

During the lifetime of each generation, a polity with state capacity $ {s}_t $ and societal capacity $ {x}_t $ produces income/output given by

(3) $$ f\left({x}_t,{s}_t\right), $$

where $ f, $ the production function, is assumed to be nondecreasing and differentiable. ⁸ We adopt a reduced-form approach in modeling the contribution of state and society capacities and assume that they contribute to aggregate income as well as affecting the conflict between them. In reality, the ability of society to coordinate and the infrastructural power of the state increase the productivity of producers, which then affects the income to be divided between different parties. ⁹ The dependence of the total output of the economy on state capacity captures the various efficiency-enhancing roles of the state. In addition, we allow for output to depend on the capacity of civil society as well because its greater cooperation and coordination also improve economic efficiency (Ostrom 1990).

We next discuss how the aggregate income is distributed between the elite (controlling the state) and society. At date t, if the elite and society fight and one side wins and captures all of the income of the economy, the other side receives zero. Winning probabilities are functions of relative capacities. In particular, the elite will win if

(4) $$ {s}_t\hskip0.35em \ge \hskip0.35em {x}_t+{\sigma}_t, $$

where $ {\sigma}_t $ is drawn from the distribution H independently of all past events. We denote the density of the distribution function H by h. The existence of the random term $ {\sigma}_t $ captures the fact that various stochastic factors affect the outcome of any conflict. ¹⁰

This specification of a stochastic contest function, and a symmetry assumption that we will impose shortly, implies that when the capacities of society and state are given, respectively, by x and s, the probability that the elite will win the conflict is H(s–x) and the probability that the society will do so is $ 1-H\left(s-x\right)\hskip0.35em =\hskip0.35em H\left(x-s\right), $ a property we will use frequently below. ¹¹

Simplifying Assumptions

We next introduce three assumptions. The first one is a simplifying assumption, which we impose initially and then relax subsequently:

Assumption 1 $ f\left(x,s\right)\hskip0.35em =\hskip0.35em 1 $ for all $ x\in \left[0,1\right] $ and $ s\in \left[0,1\right] $ .

This assumption makes it transparent that the multiple equilibria and their dynamics—our main focus—are driven by the dynamic contest between the elite and society, not because of changes in the value of the prize in this contest. It will be relaxed in the Appendix.

The next two assumptions are imposed throughout.

Assumption 2

1. 1. $ {c}_x $ and $ {c}_s $ are continuously differentiable, strictly increasing, and weakly convex over ℝ₊, and satisfy $ {\lim}_{x\to \infty }{c}_x^{\hbox{'}}(x)\hskip0.35em =\hskip0.35em \infty $ and $ {\lim}_{s\to \infty }{c}_s^{\hbox{'}}(s)\hskip0.35em =\hskip0.35em \infty $ .

2. 2. $ {c}_s^{\hbox{'}}\left(\delta \right)\ne {c}_x^{\hbox{'}}\left(\delta \right). $

3. 3. $ \frac{\left|{c}_s^{\hbox{'}\hbox{'}}\left(\delta \right)-{c}_x^{\hbox{'}\hbox{'}}\left(\delta \right)\right|}{\min \left\{{c}_x^{\hbox{'}\hbox{'}}\left(\delta \right),{c}_s^{\hbox{'}\hbox{'}}\left(\delta \right)\right\}}<\frac{1}{\sup_z\left|{h}^{\hbox{'}}(z)\right|}. $

4. 4. $ {c}_s^{\hbox{'}}(0)+{\gamma}_s\hskip0.2em >\hskip0.2em {c}_x^{\hbox{'}}\left(\delta \right)\hskip0.35em \mathrm{and}\hskip0.35em {c}_x^{\hbox{'}}(0)+{\gamma}_x>{c}_s^{\hbox{'}}\left(\delta \right). $

Part 1 of Assumption 2 is standard. Part 2 is imposed for simplicity and rules out the nongeneric case where the marginal cost of investment at $ \delta $ is exactly equal for the two parties. Part 3 is also imposed for technical convenience and is quite weak. For example, if the gap between $ {c}_x^{\hbox{'}\hbox{'}}\left(\delta \right) $ and $ {c}_s^{\hbox{'}\hbox{'}}\left(\delta \right) $ is small, this condition is automatically satisfied. We will flag its role when we come to our analysis, but anticipating that discussion, it makes it much easier for us to establish the instability of some substantively uninteresting equilibria. Part 4 ensures that the marginal cost of each player in the increasing returns region (when $ x<{\gamma}_x $ or $ s<{\gamma}_s $ ) when making zero investment is greater than the marginal cost of the other player outside this region when evaluated at $ \delta $ —the marginal cost on the right-hand side is evaluated at $ \delta $ because, as our above transformation showed, the level of investment necessary for maintaining any positive equilibrium level of capacity is $ \delta $ . We will flag the role of this assumption when we come to our formal analysis.

Assumption 3

1. 1. h exists everywhere and is differentiable, single-peaked, and symmetric around zero.

2. 2. For each $ z\in \left\{x,s\right\} $ ,

   $$ {c}_z^{\hbox{'}}(0)>h(1). $$

3. 3. For each $ z\in \left\{x,s\right\} $ ,

   $$ \min \hskip-0.12em \left\{h(0)-{\gamma}_z;h\left({\gamma}_z\right)\right\}>{c}_z^{\hbox{'}}\left(\delta \right). $$

Part 1 contains the second main assumption for our analysis—single peakedness and symmetry of h around 0 (differentiability is standard). This assumption not only simplifies our analysis as it ensures that $ h\left(x-s\right)\hskip0.35em =\hskip0.35em h\left(s-x\right) $ and $ {h}^{\hbox{'}}\left(x-s\right)\hskip0.35em =\hskip0.35em -{h}^{\hbox{'}}\left(s-x\right) $ but also implies that incentives for investment are strongest when x and s are close to each other. We highlight the role of this feature below as well. ¹²

Part 2 imposes that when a player has the maximum gap between itself and the other player, it has no further incentives to invest. Part 3, on the other hand, ensures that at or near the point where capacities are equal, there are sufficient incentives to increase capacity. Both of these assumptions restrict attention to the part of the parameter space of greater interest to us.

Equilibrium with Short-Lived Players

We now present our main results about the dynamics of the capacity of state and society, focusing on the nonoverlapping generations setup, where at each point, each side of the conflict is represented by a single short-lived agent who will be replaced by a new agent from the same side next period.

Preliminaries

Suppose that the above-described polity is populated by nonoverlapping generations of agents—on the one side representing the elite and on the other, society.

With these assumptions, at each time t, society maximizes

$$ H\left({x}_t-{s}_t\right)-\Delta \cdot {C}_x\left({x}_t,{x}_{t-\Delta}\right) $$

by choosing $ {x}_t $ (or equivalently $ {i}_t^x $ ), taking $ {x}_{t-\Delta} $ as given. Simultaneously, the elite maximize

$$ H\left({s}_t-{x}_t\right)-\Delta \cdot {C}_s\left({s}_t,{s}_{t-\Delta}\right) $$

by choosing $ {s}_t $ , taking $ {s}_{t-\Delta} $ as given. A dynamic (Nash) equilibrium with short-lived players is given by a sequence $ {\left\{{x}_{k\Delta}^{\ast },{s}_{k\Delta}^{\ast}\right\}}_{k\hskip0.1em =\hskip0.1em 0}^{\infty } $ such that $ {x}_{k\Delta}^{\ast } $ is a best response to $ {s}_{k\Delta}^{\ast } $ given $ {x}_{\left(k-1\right)\Delta}^{\ast } $ , and likewise $ {s}_{k\Delta}^{\ast } $ is a best response to $ {x}_{k\Delta}^{\ast } $ given $ {s}_{\left(k-1\right)\Delta}^{\ast } $ .

The investment decisions of both elites and society are then determined by their respective first-order conditions (with complementary slackness). In particular, at time $ t $ , we have ¹³

$$ {\displaystyle \begin{array}{cc}h\left({x}_t\hskip0.15em -\hskip0.15em {s}_t\right)\hskip0.15em \le \hskip0.15em {c}_x^{\hbox{'}}\left(\frac{x_t\hskip0.15em -\hskip0.15em {x}_{t-\Delta}}{\Delta}+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_x\hskip0.15em -\hskip0.15em {x}_{t-\Delta}\right\}& \hskip-0.9em \mathrm{if}\;\frac{x_t - {x}_{t-\Delta}}{\Delta}\hskip0.15em =\hskip0.30em -\delta\;\mathrm{or}\;{x}_t\hskip0.15em =\hskip0.15em 0,\\ {}h\left({x}_t\hskip0.15em -\hskip0.15em {s}_t\right)\hskip0.15em \ge \hskip0.15em {c}_x^{\hbox{'}}\left(\frac{x_t\hskip0.15em -\hskip0.15em {x}_{t-\Delta}}{\Delta}+\delta \right)+\hskip-0.2em \max \left\{0;{\gamma}_x\hskip0.15em -\hskip0.15em {x}_{t-\Delta}\right\}& \mathrm{if}\;{x}_t\hskip0.15em =\hskip0.15em 1,\\ {}h\left({x}_t\hskip0.15em -\hskip0.15em {s}_t\right)\hskip0.15em =\hskip0.15em {c}_x^{\hbox{'}}\left(\frac{x_t\hskip0.15em -\hskip0.15em {x}_{t-\Delta}}{\Delta}+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_x\hskip0.15em -\hskip0.15em {x}_{t-\Delta}\right\}& \mathrm{otherwise},\end{array}} $$

and

$$ {\displaystyle \begin{array}{cc}h\left({s}_t\hskip0.15em -\hskip0.15em {x}_t\right)\ \le\ {c}_s^{\hbox{'}}\left(\frac{s_t\hskip0.15em -\hskip0.15em {s}_{t-\Delta}}{\Delta}+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_s\hskip0.15em -\hskip0.15em {s}_{t-\Delta}\right\}& \hskip-0.50em \mathrm{if}\;\frac{s_t\hskip0.15em -\hskip0.15em {s}_{t-\Delta}}{\Delta}\hskip0.15em =\hskip0.30em -\delta\;\mathrm{or}\;{s}_t\hskip0.15em =\hskip0.15em 0,\\ {}h\left({s}_t\hskip0.15em -\hskip0.15em {x}_t\right)\hskip0.15em \ge \hskip0.15em {c}_s^{\hbox{'}}\left(\frac{s_t\hskip0.15em -\hskip0.15em {s}_{t-\Delta}}{\Delta}+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_s\hskip0.15em -\hskip0.15em {s}_{t-\Delta}\right\}& \mathrm{if}\;{s}_t\hskip0.15em =\hskip0.15em 1,\\ {}h\left({s}_t\hskip0.15em -\hskip0.15em {x}_t\right)\hskip0.15em =\hskip0.15em {c}_s^{\hbox{'}}\left(\frac{s_t\hskip0.15em -\hskip0.15em {s}_{t-\Delta}}{\Delta}+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_s\hskip0.15em -\hskip0.15em {s}_{t-\Delta}\right\}& \mathrm{otherwise}.\end{array}} $$

The first line of either expression applies when the relevant player has chosen zero investment so that its stock variable shrinks as fast as it can (at the rate $ \delta $ ) or is already at its lower bound $ {x}_t\hskip0.35em =\hskip0.35em 0 $ or $ {s}_t\hskip0.35em =\hskip0.35em 0 $ . In this case, we have the additional cost of investment on the right-hand side, and also the optimality condition is given by a weak inequality, as at this lower boundary, the marginal benefit could be strictly less than the marginal cost of investment. The second line, on the other hand, applies when the stock variable takes its maximum value, 1, and in this case the marginal benefit could be strictly greater than the marginal cost of investment. Away from these boundaries, the third line applies and requires that the marginal benefit equal the marginal cost. Note also that the marginal benefit for society is the same as the marginal benefit for the elite—as $ h\left({s}_t-{x}_t\right)\hskip0.35em =\hskip0.35em h\left({x}_t-{s}_t\right) $ . On the other hand, we also have from Assumption 3 that changes in the marginal benefits of the two players are the converses of each other—that is, $ {h}^{\hbox{'}}\left({s}_t-{x}_t\right)\hskip0.35em =\hskip0.35em -{h}^{\hbox{'}}\left({x}_t-{s}_t\right) $ .

Now letting $ \Delta \to 0 $ , we obtain the following continuous-time first-order optimality (and thus equilibrium) conditions

(5) $$ {\displaystyle \begin{array}{cc}\hskip-0.5em h\left({x}_t-{s}_t\right)\hskip0.35em \le \hskip0.35em {c}_x^{\hbox{'}}\left({\dot{x}}_t+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_x-{x}_t\right\}& \mathrm{if}\;{\dot{x}}_t\hskip0.35em =\hskip0.35em -\delta\;\mathrm{or}\;{x}_t\hskip0.35em =\hskip0.35em 0,\\ {}h\left({x}_t-{s}_t\right)\hskip0.35em \ge \hskip0.35em {c}_x^{\hbox{'}}\left({\dot{x}}_t+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_x-{x}_t\right\}& \mathrm{if}\;{x}_t\hskip0.35em =\hskip0.35em 1,\\ {}h\left({x}_t-{s}_t\right)\hskip0.35em =\hskip0.35em {c}_x^{\hbox{'}}\left({\dot{x}}_t+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_x-{x}_t\right\}& \mathrm{otherwise},\end{array}} $$

and

(6) $$ {\displaystyle \begin{array}{cc}\hskip-0.7em h\left({s}_t-{x}_t\right)\hskip0.21em \le \hskip0.21em {c}_s^{\hbox{'}}\left({\dot{s}}_t+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_s-{s}_t\right\}& \hskip-0.5em \mathrm{if}\;{\dot{s}}_t\hskip0.35em =\hskip0.35em -\delta\;\mathrm{or}\;{s}_t\hskip0.35em =\hskip0.35em 0,\\ {}h\left({s}_t-{x}_t\right)\hskip0.35em \ge \hskip0.35em {c}_s^{\hbox{'}}\left({\dot{s}}_t+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_s-{s}_t\right\}& \mathrm{if}\;{s}_t\hskip0.35em =\hskip0.35em 1,\\ {}h\left({s}_t-{x}_t\right)\hskip0.35em =\hskip0.35em {c}_s^{\hbox{'}}\left({\dot{s}}_t+\delta \right)\hskip-0.2em +\hskip-0.2em \max \left\{0;{\gamma}_s-{s}_t\right\}& \mathrm{otherwise}.\end{array}} $$

In what follows, we work directly with these continuous-time first-order optimality conditions. Moreover, it is straightforward to see that in continuous time, away from the boundaries of [0,1]², these first-order optimality conditions will hold as equality, and thus the dynamics of state and society capacity can be represented by the following two differential equations:

(7) $$ {\displaystyle \begin{array}{l}\dot{x}\hskip0.35em =\hskip0.35em \max \left\{{\left({c}_x^{\hbox{'}}\right)}^{-1}\left(h\left(x-s\right)-\max \left\{{\gamma}_x-x,0\right\}\right);0\right\}-\hskip0.1em \delta \\ {}\dot{s}\hskip0.35em =\hskip0.35em \max \left\{{\left({c}_s^{\hbox{'}}\right)}^{-1}\left(h\left(s-x\right)-\max \left\{{\gamma}_s-s,0\right\}\right);0\right\}-\hskip0.1em \delta .\end{array}} $$

The roles of the two main assumptions highlighted above—the single-peakedness of h and the increasing returns aspect of the cost function—are evident from Equation 7. First, when x and s are close to each other, $ h\left(x-s\right) $ is large, and thus both of these variables will tend to grow further. Conversely, when x and s are far apart, $ h\left(x-s\right) $ is small, and investment by both parties is discouraged. This observation captures the primary force that will lead to the emergence of different dynamics of elite-society relations and different types of states in our setup. ¹⁴ Second, the presence of the max term implies that once the capacity of a party falls below a critical threshold ( $ {\gamma}_x $ or $ {\gamma}_s $ ), there is an additional force pushing toward further reduction in this capacity.

Dynamics of the Capacity of Society and the State

Our main result in this section is summarized in the next proposition.

Proposition 1 Suppose Assumptions 1, 2, and 3 hold. Then there are three (locally) asymptotically stable (Nash) equilibria:

1. 1. $ {x}^{\ast}\hskip0.35em =\hskip0.35em {s}^{\ast}\hskip0.35em =\hskip0.35em 1 $ .

2. 2. $ {x}^{\ast}\hskip0.35em =\hskip0.35em 0 $ and $ {s}^{\ast}\hskip0.1em \in \left({\gamma}_s,1\right) $ .

3. 3. $ {x}^{\ast}\in \left({\gamma}_x,1\right) $ and $ {s}^{\ast}\hskip0.35em =\hskip0.35em 0 $ .

This proposition shows that there exist three relevant (asymptotically stable) equilibria: one corresponding to an inclusive state, one corresponding to a despotic state, and one corresponding to a weak state. The intuition, as already anticipated, is that when we are in the neighborhood of the steady state $ {x}^{\ast}\hskip0.35em =\hskip0.35em {s}^{\ast}\hskip0.35em =\hskip0.35em 1 $ , $ h\left(x-s\right) $ is large, encouraging both parties to move further toward $ {x}^{\ast}\hskip0.35em =\hskip0.35em {s}^{\ast}\hskip0.35em =\hskip0.35em 1 $ . In contrast, in the neighborhood of $ {x}^{\ast}\hskip0.35em =\hskip0.35em 0 $ or $ {s}^{\ast}\hskip0.35em =\hskip0.35em 0 $ , $ h\left(x-s\right) $ is small, and neither party has as strong incentives to invest, and in fact, one of them ends up with zero capacity. ¹⁵

The equilibria presented in Proposition 1 and their local dynamics are exactly the same as those shown in Figure 1 in the Introduction. This can also be seen from the numerically constructed Figure 3 presented here. ¹⁶

Figure 3. The Direction of Change of the Capacities of State and Society in a Simulated Example

Conditional Comparative Statics

One of the most important results of our framework is that comparative statics—which show how the equilibrium changes when perturbed by exogenous factors—are conditional and in particular depend on the prevailing balance between elites and society (where we are in the diagram in Figure 1). The easiest way of seeing this is to consider an increase in $ {s}_0 $ to $ {s}_0+\overline{s} $ (this may be driven, for example, by changes in military technology or international factors, which necessitate greater building of state capacity for military reasons, as in Tilly’s [1990] discussion). As Figure 2 in the Introduction illustrates, such a change can leave a polity in the same region as before, in which case the equilibrium trajectory will be shifted uniformly up, but the long-run outcome will remain unchanged (this will correspond to the Montenegrin case we discuss in the historical perspective section). Alternatively, this increase can shift us from, say, Region III to Region II, in which case not only the equilibrium trajectory but also the long-run outcome will change, and in fact it will involve greater state capacity (corresponding to the Swiss case below). However, depending on the exact value of $ \left({x}_0,{s}_0\right) $ , the same increase of $ \overline{s} $ in state capacity could also shift us from Region II to Region I, in which case the effect on the long-run state capacity will be negative instead of positive as in the Prussian case (also discussed below). This discussion thus establishes

Proposition 2 The effects of changes in the initial conditions $ \left({x}_0,{s}_0\right) $ on equilibrium dynamics and the long-run outcome for state capacity are conditional in the sense that these depend on which region we move out of and into.

A more rigorous discussion of conditional comparative statics is provided in the Appendix, where we generalize the model so that the f function depends on state and societal capacity and we consider variations in the importance of these capacities for income. In that case, even for given initial conditions, dynamics can change fundamentally as the boundaries between regions shift, and once again exactly where we are in terms of initial conditions will determine how elites and society respond to the same change in structural factors.

Equilibrium with Forward-Looking Players

In this section, we briefly outline how the results generalize to a setting with long-lived, forward-looking players.

The technology of investment and conflict are the same as in the previous section. The only difference is that now both society and elite are long-lived and forward-looking. To maximize the parallel with the model with short-lived players, we assume that both players again correspond to sequences of nonoverlapping generations, but each generation has an exponentially distributed lifetime or equivalently, a Poisson end date with parameter, $ 1-\beta $ , where $ \beta \hskip0.35em =\hskip0.35em {e}^{-\rho \Delta} $ . We assume that this random end date is the only source of discounting. Clearly, this specification guarantees that as the period length ∆ shrinks, discounting between periods will also decline (and the discount factor will approach 1). Again to maximize the parallel with our myopic model, we assume that in expectation, there is one instance of conflict between the two players during the lifetime of each generation. Because with this Poisson specification, the expected lifetime of this generation is $ 1/\left(1-\beta \right) $ , this implies that a conflict arrives at the rate $ 1-\beta $ . ¹⁷

The main result of this section is contained in the next proposition.

Proposition 3 Suppose Assumptions 1, 2, and 3 hold. Then there exist discount rates $ \overline{\rho}\hskip0.35em \ge \hskip0.35em \underline{\rho}>0 $ such that for all $ \rho >\overline{\rho} $ , there are three (locally) asymptotically stable equilibria:

1. 1. $ {x}^{\ast}\hskip0.35em =\hskip0.35em {s}^{\ast}\hskip0.35em =\hskip0.35em 1 $ .

2. 2. $ {x}^{\ast}\hskip0.35em =\hskip0.35em 0 $ and $ {s}^{\ast}\in \left({\gamma}_s,1\right) $ .

3. 3. $ {x}^{\ast}\in \left({\gamma}_x,1\right) $ and $ {s}^{\ast}\hskip0.35em =\hskip0.35em 0 $ .

Moreover, for all $ \rho <\underline{\rho} $ , there exists a unique globally stable equilibrium $ {x}^{\ast}\hskip0.35em =\hskip0.35em {s}^{\ast}\hskip0.35em =\hskip0.35em 1 $ .

Thus, this result shows that the main insights from our analysis apply provided that players, though forward-looking, are sufficiently impatient. ¹⁸ We note that this result is not a simple consequence of the fact that as we consider larger and larger values of $ \rho $ , players are becoming closer to myopic. It necessitates establishing properties of the the relevant value functions and their derivatives in the limit. Details as well as numerical illustrations are provided in the Appendix. ¹⁹

European State Divergence in Historical Perspective

In the Introduction, we discussed the divergent paths of political development in Switzerland, Prussia, and Montenegro. In this section, we elaborate on these historical experiences, illustrating how our theory is useful for interpreting these divergent trajectories. This section has two other related goals. First, we illustrate the notion of conditional comparative statics using the salient example of interstate warfare. Finally, we provide evidence that state capacity is greatest when there is more balanced contestation between state and society rather than under the auspices of a despotic state.

Switzerland

Switzerland was historically on the periphery of the Holy Roman Empire. The Empire still had an emperor, but it had fragmented into many small and relatively independent polities. The component Swiss polities, the cantons, had their own systems of assemblies. These were legacies of two forces, first of the broader pattern of assembly politics inherited from German tribes, which had been institutionalized initially by the Merovingians, and second, from the political consequences of the communal economic organization of the medieval period (Wickham 2017). Swiss cantons did not just have assemblies; they also featured feudal, often monastic, elites.

The Swiss confederation started in 1291 with men from the cantons of Uri, Schwyz, and Unterwalden taking oaths in Rütli, a meadow above Lake Uri, and signing the Bundesbrief (Federal Charter). The charter was concerned with public order and lawlessness and committed the three cantons to come to each other’s aid and provided a framework for resolving disputes. ²⁰ The disputes were often with the feudal elites. In Schwyz, for example, cantonal authorities were in conflict with Einsiedeln Abbey over access to lands, which the Abbey claimed. In the conflict Schwyz was excommunicated by the Bishop of Constance, and in retaliation the Abbey was attacked and looted.

We see here two very concrete ways in which the power of society to act collectively was manifested. One was through the direct democracy of the assemblies; “The symbol of Swiss freedom has thus been the Landesgemeinden—the cantonal assemblies through which direct participation was assured, and on-going self-government guaranteed” (Barber 1974, 11). The other was the process of oath taking, which started in 1291. Oaths were taken, and continually reaffirmed, to coordinate and commit adult men to common goals and help solve the collective action problem. This was such a prevalent feature of Switzerland that when the Swiss Confederacy first appeared, it was known to contemporaries as the Eidgenossenschaft, the “oath comradeship.”

Uri, Schwyz, and Unterwalden were in principle subservient not just to local elites but also to the Habsburg Duke of Austria. The charter stated, “We have further unanimously vowed and established that we in these valleys shall accept no judge who has gained his office for money or for any other price and who is not our resident or native,” clearly undercutting the authority of Habsburg judges. After Einsiedeln was sacked, Uri and Unterwalden were also excommunicated. The Habsburgs attempted to use force to help local elites gain control, but they were defeated in 1315 at the battle of Morgarten. The three cantons could invoke the Bundesbrief to field a collective army. More pacts (and oaths) followed, and what came to be known as the Old Swiss Confederacy spread. The Confederacy was continually threatened—for example, by Duke Leopold II of Austria, whose army was decisively defeated at the battle of Sempach in 1386 by the combined Swiss forces. The Treaty of Basel in 1499 established the Confederacy’s de facto autonomy.

Therefore, the Swiss state emerged out of a constant conflict between communes (and the cantons) and local elites backed up by the higher political institutions of the Holy Roman Empire. After 1291, it was the cantons that accumulated power and took on a collective identity. After 1415, an assembly made up of delegates from all the cantons in the Confederacy began to meet regularly, thus weakening the power of elites and building republican institutions. The independent Swiss state was in the making. Fribourg and Solothurn were admitted in 1481, Basel and Schaffhausen in 1501, and Appenzell in 1513. The independence of this state from the Holy Roman Empire was finally recognized by the Treaty of Westphalia in 1648, which ended the Thirty Years War, though the consolidation of a modern central state continued for another 200 years.

Prussia

Prussia was never part of the Holy Roman Empire, but in 1618 it merged through marriage with Brandenburg, which was. The ruling family of Brandenburg, the Hohenzollerns, became the ruling family of Brandenburg-Prussia, with the ruler known as the Elector. During the Thirty Years War, the newly united territories were devastated. Brandenburg might have lost as much as half of its population (Blanning 2016, 12).

In 1640, Frederick William I came to the throne as the new Elector. Known as the Great Elector, he ruled for 48 years. He charted a new course for Brandenburg-Prussia based on the despotic path of state building. One of his main goals was to build a much more effective military. To achieve this, Frederick William needed tax revenues. Taxes had to be negotiated with various representative bodies, such as the Estates of Kurmark in Brandenburg. He started out by trying to get permanent grants of taxation, which would free him from the need to endlessly negotiate them. In 1653 he negotiated the so-called Recess that gave him 530,000 thalers over a period of six years. Crucially, he, rather than the Estates, got to collect the taxes, which allowed him to start building a bureacracy. In exchange he gave the nobility, which made up a chamber in the Estates, tax-exempt status. He went on to extract similar concessions from the Prussian Estates. Frederick William then overrode the authority of the Estates and started to tax without their agreement. In 1655 he initiated the Kriegskommissariat (the “war commissary”), which covered both tax collection and military organization, and he also stripped Estates of their militias, which responded “with bitter protests” (Clark 2009, 56). By 1659, the Estates retreated to local issues. They attempted to combine their forces, but their resistance was broken by “coercion and force” and “Leading Estates activists were intimidated or arrested” (Clark 2009, 57, and see Carsten 1954).

With the Estates sidelined, despotic state building commenced. In 1733, the basis of military recruitment was reorganized. The country divided into cantons of 5,000 households, with a regiment assigned to each for recruitment. At least a quarter of the male population was included in the rolls, dramatically increasing the potential size of the army. Frederick the Great, assuming power in 1740, further expanded the tax base and strengthened the Prussian military machine. The French philosopher Voltaire (supposedly) summarized the Prussian situation as “other states possess an army; Prussia is an army which possesses a state.”

Our main point here is that this very stark difference from Switzerland evolved out of initial circumstances that were far more similar than different. Swiss local autonomy and democracy were not entirely unique; there were similar models all over Germany (Brady 1985), and the representative institutions of the Estates had considerable powers in many parts of the territories. Yet in Prussia, such institutions weakened in the conflict with the Hohenzollerns. This did allow for the accumulation of state capacity, including an expanded fiscal system and army. Yet, as we will see, the Prussian despotic state eventually ended up with less capacity than the Swiss Republic.

Montenegro

Montenegro was made up of kinship groups, referred to as clans, and lacked the elements of centralization that the Swiss (and Prussians) had inherited from the Carolingians and the Holy Roman Empire. Such kin groups had a great ability to coordinate, and they persistently opposed the creation of a state because “continued attempts to impose centralized government were in conflict with tribal loyalty” (Simić 1967, 87). Prior to 1852, Montenegro was a theocracy but where the ruling Bishop, the Vladika, could exercise no coercive authority over the clans that dominated the society, and “it was only when their central leader attempted to institutionalize forcible means of controlling feuds that the tribesmen stood firm in their right to follow their ancient traditions. This was because they perceived in such interference a threat to their basic political autonomy” (Boehm 1986, 186). Here Boehm is referring to the attempts of Vladika Njegoš to develop a state in Montenegro in the 1840s. Djilas describes the situation: “It was a clash between two principles—the state and the clan. The former stood for order and a nation, and against chaos and treason; the latter stood for clan freedoms and against the arbitrary actions of an impersonal central authority—the Senate, the Guard, the captains” (Djilas 1966, 107). Djilas records that Njegoš’s reforms were immediately confronted by the revolt of the Piperi and Crmnica clans motivated by the fact that “the imposition of government and a state was putting an end to the independence and internal freedom of the clans” (Djilas 1966, 115).

Njegoš was succeeded by his nephew Danilo, who made himself the first secular Prince of Montenegro in 1851, but his efforts to centralize authority also ran into fierce opposition. An attempt to raise taxes in 1853 led the clans to revolt, with the Piperi, Kuči, and Bjelopavlići declaring themselves an independent state. Danilo’s attempt failed, and a member of the Bjelopavlići assassinated him in 1860.

The Nature of the Comparison

Methodologically, this section uses Mill’s (1872) most similar design for choosing case studies. Switzerland, Prussia, and Montenegro diverged in terms of the capacities of their states, and we relate this primarily to differences in initial conditions, in particular how much capacity their state had in the contest with society. Apart from these differences, there were many similarities.

Consider Switzerland and Prussia. We have emphasized the similarities in terms of culture, ethnicity, and religion and of the institutions of the Holy Roman Empire. The main difference is that the Elector of Brandenburg-Prussia had more control over his territories than did the Habsburg Dukes over the Swiss cantons. The Elector was able to override the power of the Estates, whereas the local Swiss nobility could not stop the cantons signing an agreement that, for example, forbade Austrian judges.

Turning to Switzerland and Montenegro, they both had been peripherally part of the Roman Empire and shared the same type of mountainous ecology and an economy based on herding. We also argued that there were similarities in social structures, in particular, clans. The important distinction is that the state was stronger in Switzerland where there was at least the legacy of Carolingian centralization, which never existed in Montenegro.

These small differences around 1600 ended up creating very different dynamics of state-building and societal capacity during the early modern period because, in terms of our model, they put the different polities into three different basins of attraction.

Conditional Comparative Statics in Action

Our case study illustrates our central notion of conditional comparative statics. We focus on the effects of warfare, as discussed in Proposition 2.

For the Swiss, the threat of warfare, in particular the persistent threat from the Habsburgs to reinstate the overlordship of the feudal local elites, seems to have been an important incentive for the otherwise individualistic cantons and cities to unite into a larger confederation and to accede to more centralized authority and decision making. This seems to have overcome what might otherwise have been strong centrifugal tendencies. Thus the centralization that started in 1291 was likely encouraged by the pressure of warfare. As Figure 2 in the Introduction highlighted, it is plausible that this impulse toward greater state capacity pushed the Swiss cantons into the basin of attraction of the inclusive state.

The outcome in Prussia was very different. Though Brandenberg had many of the same structural features of Switzerland and Prussia was run by its ruling house, the territories to the east did not have the same history of communes and quasi-democratic politics like the Swiss cantons. Society was more easily dominated. Nevertheless, at the start of the early modern period it is plausible to believe that Prussia was in, or at the very least in the vicinity of, the basin of attraction of the inclusive state. In this light, the Thirty Year War can be thought of as forcing Prussia out of the basin of attraction of the inclusive state and into that of the despotic path. This is how Frederick the Great himself viewed the situation: “So long as God gives me breath, I shall assert my rule like a despot ” (quoted in Blanning 2016, 127).

Finally, war did not make the state in Montenegro and the incentives it created were not powerful enough to move the country out of the basin of attraction of the weak state. Continuous warfare with the Ottomans did induce the clans to try to coordinate more and create more central structures (see Durham 1928). As we have seen, however, this impulse was not sufficiently powerful to create a state.

The Capacity of the State

One of the most surprising implications of our analysis is that it is not the despotic state that has more capacity but the inclusive one. This implies an ordering of Switzerland then Prussia and then Montenegro in terms of their levels of state capacity, exactly as in Figure 2.

This can be seen when we focus on the early modern period. We focus on a central metric of state capacity in the literature—the size of the fiscal state. ²¹ The earliest available evidence suggests that despite the development of the absolutist Prussian state, the tax take relative to national income was higher in Switzerland. This is in line with the predictions of our model. Data in Aidt and Jensen (2009) suggest that central government tax revenues compared with national income were around 1.8% during the period 1860–1880 in Switzerland, rising to 2.1% during 1881–1914. But the central government was the smallest part of Swiss government tax revenues and expenditures. Cantons had raised income taxes since 1840, whereas the federal government did not do so until the 1930s. Though we have not been able to identify historical estimates of cantonal taxes, the Organisation for Economic Co-operation and Development provides a comprehensive breakdown since 1965. ²² In 2018, cantonal and municipal tax revenues on individuals were five times the levels of federal income taxes, whereas the local corporate taxes were about the same. In 1965, these differences were far larger; local income and corporate taxes were over 11 and 4 times, respectively, as large as federal taxes. Because in 1860, for example, there were local income taxes but no federal income tax, the number of 1.8% of GDP is clearly a large underestimate of the actual extent of taxation in Switzerland (Bullock 1924 provides an extensive discussion of the numerous taxes levied by Swiss cantons in the nineteenth century). Spoerer’s (2010) estimates for Prussia are that tax revenues were 5% of national income in 1860. Therefore, we can conservatively estimate that in 1860 tax revenues relative to national income were twice as high in Switzerland, about 10%, as in Prussia, and thus Switzerland appears to have achieved greater fiscal capacity with an inclusive state.

Conclusion

There is a great deal of diversity in the nature of states and their capacity around the world today. But societies, not just states, differ enormously as well. Some are highly mobilized and organized collectively, with high levels of “social capital” and institutions that facilitate collective actions. In contrast, in others nonelites are weak and incapable of contesting for power against elites and the state. The capabilities of states and societies go together.

In this paper, we have developed a new theory for studying the variation in state capacity and state–society relations, arguing that states endogenously acquire capacity in a dynamic contest between elites and society. At the heart of our model is the notion that elites that control states must contest with society for control over political power and the distribution of resources. If the state accumulates capacity, then this helps elites win this contest. But in response society can also accumulate capacity, and this contestation from society in turn encourages the elite to build further capacity. In our model, this logic leads to three distinct equilibria with very different constellations of state and societal capacity. In one equilibrium, which we called despotic, the state acquires far more capacity than society, in a sense dominating it. In the reverse situation, where society accumulates more capacity than the state, we have a weak state. Finally, a rough balance of power between state and society leads to the emergence of an inclusive state. Our model clarifies how the competition between elites and society in this case is the engine behind the emergence of the greatest state capacity. Elites in despotic states, because they can easily dominate society, have less reason to accumulate as much power and capacity.