War makes states and states make war is a catchy aphorism that for all of its appeal remains contested. Scholars who have studied European history and/or the trajectories of major powers have found strong evidence in support of the relationship. Scholars who study less developed regions tend to argue that the relationship does not work in their selected patch of the world. Still others argue that the relationship is more complicated. Its operation is contingent on the presence or absence of additional, critical variables. It depends, according to different authors, on what types of war are waged, how the wars are fought and financed, what types of regimes are at war, where the wars are fought, or even who wins and loses.
Oddly enough, the proposition that war tends to expand state capacity (revenue extraction) in an upward ratcheting way has never been tested all that comprehensively. Part of the problem has been missing data on revenues and systematic ways to control for inflation. By default, much of the information that we have is based on major powers because they are prominent and because their fiscal records are more accessible. However, more data are now available. Hence, it is possible to examine the empirical relationship between war and revenues more broadly.
In a new effort to address the contested nature of war making-state making processes, we advance a more generalized argument and expand the number of cases and geographical scope for investigation beyond earlier analyses. In the end, we put forward a theoretical and empirical resolution of some of the analytical debates on this topic. We maintain that it is not war per se that makes the state. Rather, it is intensive warfare that leads to extensive mobilization and, consequently, more state expansion. Other factors play some role as well, but it is the extent of warfare, not its mere occurrence, that is most critical for state expansion.
We believe that claims for the lack of applicability of the war making-state making relationship – although fairly well established in European history – to various parts of the non-European world should consider the types of wars fought in these parts, and not so much the utility of the “bellicist” theory alone. In other words, we are modifying the traditional aphorism to “intensive war makes states, and states make war at varying intensity.”
The War-State Building Relationship
The basic contention in the war making-state making model is that states are inherently organizations that make war and collect tribute in order to make more war (Pollack, 2009: 291):
The state is a political organization that claims a monopoly on violence within a given territory. To the extent that a particular state actually possesses such a monopoly (or something close to it), it is ideally situated to perform two vital functions: make war and collect tribute from those living under its jurisdiction. A good deal of the tribute collected is typically dedicated to the military apparatus of the state, which in turn, is used to forcibly extract revenue from society and sometimes, other nations. This is the basis of the powerful combination of fiscal and military powers behind the state. In essence, all states are fiscal-military organizations – some more efficient and successful than others.
The survivors of the intensely escalatory warfare in Europe (and some of its leading former colonies) between the sixteenth and mid-twentieth centuries developed into relatively strong states. To survive the warfare, more and more revenue needed to be extracted from wherever they could be found. Meanwhile, more demanding wars required an expanded bureaucracy to collect and manage state revenues. They also entailed larger armed forces and more expensive weaponry technology. In turn, stronger states could then fight more total wars, which lead to an increasing threshold of future resources that states would need to participate in the warfare system. It is in this sense that war made the state and states made war.1
Yet there are at least four possible problems with the war-state building relationship. One is that it is too simple. A number of variables could intervene between the war stimulus and the state building outcome.2 A second possible problem is that the relationship is outdated.3 It may have worked in European history but, for various reasons, it no longer accurately describes the impact of intensive conflict episodes today – in large part, perhaps, because the wars are not as sufficiently intense as they once were. But, a third possibility is that some types of wars build states while others do not.4 Or, the fourth possibility is that the relationship is entirely spurious. Wars only seem to build state capacity sometimes but the processes that seem most overt only mask more subtle interactions.5
In many respects, these four problems may actually represent one more complicated and encompassing problem: war per se simply does not have a generic, generalizable effect. Sometimes wars build states into stronger organizations than had existed before the war. Other times, they destroy the states. In between these two extremes are perhaps many cases in which war participation seems to have little effect. In other words, wars come and go but the states that fight them remain virtually untouched before and after the combat.
If the war making-state making relationship is too simple, then we need to move beyond this bivariate relationship. This is especially the case if different types of war occur in diverse parts of the world, leading to differential war-state building effects in various regions. Alternatively, since war making activity has shifted gradually away from Europe, perhaps, the problem is that non-European wars fail to have some ingredient(s) that once characterized European warfare. If so, we need to capture what distinguishes these regions and/or wars theoretically – which is still yet another argument for transforming the bivariate war making-state making relationship into a multivariate one.
We argue that the main feature that distinguishes contemporary third world warfare from the older European variety is its mobilization intensity. Wars in Europe once were largely ignored by most of the population. Unless armies happened to march through agrarian fields, much of the population might have little direct contact with soldiers. But, indirect pressures on the population gradually increased as desperate states sought revenues to pay for expanding and more expensive combat. As the combat grew in scope, it became more difficult for the local populace to evade the fighting. By the twentieth century, few city dwellers were safe from aerial bombardment. As wars became increasingly total, few members of a population at war could escape some contact with dead and injured relatives, women working in factories, or sundry sacrifices on the home front that insured food, chemicals, and gasoline were committed to the war effort. Total warfare, therefore, was distinguished by the high degree of involvement of the entire society at war.
In marked contrast, most contemporary interstate warfare is much more limited in effect and degree of involvement.6 States often make little effort to expand revenue collection for wars which are over relatively quickly. The short time frames make widespread conscription or even civilian shortages unlikely. The battle fronts are reasonably far way in most cases. Much of the population may be aware that war is ongoing but its degree of involvement is considerably less than total.
Thus, it was not European warfare per se or warfare of the 1494-1945 era that was different. Rather, it was the intensity and the growing intensifying impact of escalating warfare over a long period of time that made surviving political organizations stronger than they had been at the outset of this warring states period. Most European political organizations did not survive the destruction. If there were as many as 1000 actors in 1000 CE, only 500 still existed in 1500 CE. By 1945, the number had dwindled to about 30. But, many of the states that remained were much stronger political organizations than the states that had failed to experience a half-millennium of upward spiraling warfare.7 Our main hypothesis, therefore, is
H1: States that engage only in limited warfare are less likely to experience positive state making impacts than states that engage in less limited or total warfare.
Of course, there are many types of state making impacts. In this examination, we focus exclusively on revenue extraction which is customarily regarded as the primary avenue to expand state capacity. Peacock and Wiseman (1961) established, at least for the United Kingdom, that warfare tends to exhibit a ratchet-like effect on state revenues and spending. Both fiscal activities increase during wartime and then may decrease in the post-war era but the decrease does not return the extraction level to the pre-war level. Therefore, warfare provides an opportunity to overcome the normal resistance to paying more taxes and decision-makers are usually keen to exploit this window of opportunity. If they are involved in a life and death struggle, they have little choice but to mobilize more resources for the fight. In contrast, they face more domestic resistance and fewer options in struggles that are perceived to be less salient. Thus, we expect that step-like expansions of revenue collection should be associated with more intensive warfare.
There are other factors that might also conceivably influence this effect. Although we have some doubts as to whether these factors should have significant impacts for all of the dimensions of war effects, we include them in our empirical analysis. One such factor is the level of economic development. We expect that pre-war economic levels will act as constraints on post-war economic growth (Van Raemdonck and Diehl, 1989; Jaggers, 1992). For instance, less developed states are more likely to remain less developed after engaging in war. More developed states may be poorer after war, at least temporarily, but the level of economic complexity is unlikely to be diminished for long.8
War outcomes and regime types are other possible factors. War winners are more likely to experience positive impacts (or avoid negative impacts) than are losers. War losers may have to pay reparations and sustain losses of territory and population. Their governments may also suffer in terms of popularity, legitimacy, and survival (Stein and Russett, 1980; Bueno de Mesquita and Siverson, 1992; Bueno de Mesquita, Smith, and Siverson, 2003). Therefore, it should be more difficult to sustain wartime levels of resource extraction after losing.
As for regime type, Lake (1992) argues that democracies are more capable of extracting increased resources from their populations. If they should lose a war, democracies, in comparison to autocracies, are also more adaptable and unlikely to experience institutional collapse (Jaggers 1992). Moreover, since democracies would have more legitimacy than autocracies, they would be able to overcome negative post-war environments more successfully. On the other hand, democracies are likely to be more sensitive to voter dissatisfaction and therefore more likely to reduce wartime extraction levels quickly in a post-war era. But, this propensity may be offset by popular demands for more governmental services, especially in welfare – a trend since at least the last quarter of the nineteenth century. Certainly, one of the reasons why older European states have come to have high revenue extraction levels has been their commitment to welfare spending.
Still another factor to consider is the interaction between regime type and war victory. There has been considerable and ongoing debate as to whether and why democracies tend to win the wars in which they participate (Lake, 1992; Jaggers, 1992; Reiter and Stam, 2002, 2003; Desch, 2002, 2008; Choi, 2004; Rasler and Thompson, 2005; Downes, 2009; Keir and Krebs, 2010).
In sum, although we are less concerned with economic growth than the impacts on state capacity, we are not sure the extent to which economic development levels should constrain expanding state capacity, but we acknowledge that it could. Revenue extraction is certainly more difficult in agrarian and less affluent economies than it is in economies built around industry, trade and urbanization. Since past research tells us that war outcomes, regime types, and their interaction, could also have a significant influence on state capacity, we need to investigate their roles as well. Losing autocracies do have some propensity to experience political instability in a postwar era which we expect could influence the likelihood of sustaining expanded state power. If democracies are better at winning, extracting revenues, and responding to demands for increased governmental services, we might anticipate that they will be more likely to maintain expanded state size in the aftermath of warfare. It may also be that resistance to increased taxes is greater in democracies (in comparison to autocracies) and that democratic decision-makers therefore have a greater incentive to exploit the window of opportunity associated with periods of intensive warfare.
These questions of war intensity impacts have already been examined. While there are a number of small N, empirical quantitative studies and case studies, Rasler and Thompson (1985b) demonstrated that among major powers, global wars, because they were more intense and mobilized more people and resources, were far more likely to lead to the expansion of state revenues and spending than were non-global wars.9 Nonetheless, Jaggers (1992) argued quite correctly that we needed to go beyond a dichotomous measure of war type and an overly restricted major power group in order to generalize the argument. He introduced several measures of war mobilization that were found to have a significant and positive impact on state capacity (e.g., revenues per capita) – a finding that included other non-major powers. However, Jaggers’ sample (which included both civil and interstate wars) was restricted to 20 European and American participants in 20 interstate wars that were fought between 1815 and 1954.10 As a consequence, 61 percent of the cases involved major powers and exactly half of the participation cases were linked to World Wars I and II. Although Jaggers’ study reflects an effort to obtain a more comprehensive and representative sample than one exclusively oriented toward major powers, we believe that we can broaden his sample even more.11
However, our efforts to broaden the sample are hampered by real limitations on the availability of state revenue data and the necessity of calculating the proportion of state revenues to GDP in order to control for inflationary effects.12 Still, Sarkees and Wayman’s (2010) revised Correlates of War list of interstate wars counts 95 wars and 338 participants between 1816 and 2007. Most of these war participants are not major powers. So, there is potential to expand the spatial and temporal scope of the interstate wars and war participants significantly. Despite constraints of data availability, our empirical investigation below roughly doubles the number of cases and war participants relative to Jagger’s analysis. More precisely, we look at 40 wars between 1870 and 2007 which involve 39 participants (without geographic restrictions) – 85% of which are non-major powers. The WW I and WW II cases present less than 20% of our sample. Obviously, we are still missing a fair amount of data, but the sample is at least becoming more representative.
We estimate two OLS regression models in order to determine the extent to which involvement in interstate warfare influences increases in state capacity. We hypothesize that interstate war involvement, particularly those that are fought with the greatest intensity, will have a positive effect on increasing state capacity, and we expect this relationship to hold for small and developing states as well as those that are larger and advanced. In addition to estimating the influence of interstate wars, we introduce three additional variables that are expected to influence state capacity in post-war settings: regime type, war outcome and economic development. Since democracies are thought to be more responsive to demands for post-war reconstruction and to enjoy more legitimacy than are autocracies, we expect that these regimes will recover more quickly from war. Also, past research indicates democracies are likely to win wars that they enter because either their leaders are confident that they can win, or they have greater wealth that can be mobilized in comparison to autocracies, their citizens are more willing to cooperate with wartime extraction, or because they produce more competent soldiers than autocracies (Lake 1992; Reiter and Stam 2002; Bueno de Mesquita et.al. 2003. While past research also raises questions about these causal justifications, most scholars acknowledge some relationship between regime type and war outcome.
In addition to democracy, we expect that that those states that emerge as winners in the post-war environment will see a positive increase in state capacity in comparison to those that lose. Losers are expected to recover more slowly due to greater internal destruction that would make it harder for state leaders to mobilize internal revenues for rebuilding. Also, losers are likely to suffer additional burdens associated with war reparations and disrupted trade relationships – both of which would slow the process of mobilizing revenues. Meanwhile, energy consumption is introduced as a control for the level of economic development since earlier research indicates that poor states are likely to be affected most adversely by war involvement (Van Raemdonck and Diehl 1989).
The first OLS model estimates the influences of democracy, war outcome, the type of interstate war and energy consumption. In the second model, an interaction term between democracy and winner is estimated along with the direct additive influences of the remaining variables.
State capacity is measured as the difference between the 5 year pre-war mean and the 5 year post-war mean of a state’s revenues/GDP. The difference between the 5 year pre-war mean and the 10 year post- war mean of state’s revenues/GDP will also be estimated. The revenues and GDP data are obtained from Banks Cross-Polity Time Series dataset (Banks, 2011) and are supplemented by data from the Economy Watch Statistics and Indicator Data Base.13
Democracy is a binary variable. States that have a Polity II value of 7 or greater (after subtracting the autocratic score from the democratic score) are coded 1 (democracies) and the remaining states are coded 0 in the year preceding the war. This data can be found in the Polity IV dataset at http://www.systemicpeace.org/polity/polity4.htm.
War winner is also a binary variable. Those states that are winners are coded 1 while all other outcomes are coded 0. This variable is based on the Correlates of War Interstate War dataset (v.4).14
War type is an ordinal variable based on Jaggers (1992) measure of the type of impact that a war is likely to have. While Jaggers established that the values of 1 and 2 applied to civil wars with or without external intervention, he reserved the values of 3 through 6 for interstate wars. In this study, interstate wars with two party conflicts were coded 3; wars with multiparty conflicts were coded 4; global wars with non-major power participants in World Wars I and II were coded 5; and lastly, hegemonic wars with major-power participants in World Wars I and II were coded 6. The presumption is that hegemonic and global wars are likely to impact state power more dramatically given their scale, scope and duration and the higher levels of resource mobilization required to fight them (Jaggers, 1992: 43).
Energy is an interval variable and reflects the level of primary energy consumption in the year before the war started. The data are based on the Correlates of War National Capabilities (v.4) dataset (Singer, 1987) found at http://www.correlatesofwar.org/COW2%20Data/Capabilities/nmc4.htm.
Each country’s involvement in a war represents a single case. The list of the pooled cases is provided in Appendix A-1. Data constraints played a significant role in determining the sample size. Our sample includes only those cases in which revenues as a proportion of GDP could be obtained for each war participant. Although revenues by themselves were available for a wider number of war participants, the lack of data availability for gross national product severely constrained our sample size. Since we believe that the presence of war and post-war inflation is likely to overestimate the changes in state revenues in the post-war aftermath, we included only those cases in which we had values for both revenues and GDP in the time period under consideration. Therefore, we have 102 pooled cases for our first dependent variable: the difference between the five year pre-war mean and the five year post-war mean in revenues/GDP. We have 106 pooled cases for our second dependent variable: the difference between the five year pre-war mean and the ten year post-war mean in revenues/GDP.15
The distribution of the independent variables within each of the pooled samples for our dependent variables is provided in table 1. The table below shows that neither democracies, winners nor democracies and winners overly dominate the two pooled samples. Lastly, the average difference between the pre and five year post-war means for revenues/GDP (our first dependent variable) is .020 with a range between -.195 and .214 values for 102 pooled cases. The average value for the difference between the pre and ten year post-war means (our second dependent variable) is .023 with a range between -.187 and .218 values for 106 pooled cases.
/Table 1 about here/
Tables 2 and 3 report the estimated coefficients for the five and ten year post-war changes in revenues/GDP. Together, the results are quite similar. When democracy, winner and war type are treated as direct additive influences on five year post-war changes in state capacity (see table 2, Model I) war type emerges as the only statistically significant variable. Possible collinear influences are minimal: war type is weakly correlated with democracy at 0.15 and moderately correlated with a winning war outcome at 0.35. The control variable for economic development (energy consumption) is statistically insignificant. However, in Model II (table 2) when an interaction term that combines the influence of both democracy with war outcome is added to the variables in the first model, it is associated with a significant positive influence on increasing post-war state capacity. War type continues to have a positive significant effect on post-war state capacity. Post-estimation simulations for these results are displayed in the upper half of table 4 and indicate that while the interaction effect of democracy and winning has a 153% increase in the five year change in post-war capacity, the type of war is associated with an additional 18% increase at 171%.
/ Table 2 here /
The OLS results for the effects of the independent variables on the ten year post-war changes in state capacity are provided in table 3. Like the five year post-war changes in state capacity, the independent additive influences of democracy and war winners are statistically insignificant, while war type has a strong positive influence (Model I, table 3). Again, collinearity is ruled out as a possible influence on the results. In Model II (table 3), however, when the interaction term between democracy and war winner is introduced with the earlier variables, its estimate is associated with a very strong effect on ten year post-war changes in state capacity as does war type as well. The post-estimation simulation results displayed in the bottom half of table 4 show that both of these variables, in line with the five year results, have a strong influence on increasing post-war state capacity in comparison to the baseline model where the variables are held constant at their mean. More precisely, the interaction between democracy and war winner has a 119% increase on the change in post-war state capacity, while war type has a 143% increase. Like the results for the five-year change in state capacity, war type has the stronger effect.