This essay is an attempt to establish the relationship between military funding and a level of democracy by analyzing 115 states in a cross-sectional study. The first part of the essay is dedicated to a summary of related literature and the development of testable theories. Literature on dynamics of the relationship between military expenditure and democratization is split. One school of thought claims that governments should increase military expenditure to provide a stable environment for further development of democracy, also known as the Increased Resources theory. A second theory of Decreased Resources states the opposite. State leaders should decrease military expenditure in order to limit the military’s strength and thus its ability to intervene. In the second part of the essay method, research design and the reliability of the data are assessed in more detail. After regressing the variables against each other a significant support for Decreased Resources theory is observed.


This essay will examine the dynamics of the way the military and the ruling group interact with each other and what outcome is more likely to be achieved. The 20 th and 21 st centuries showed that the military is a too important factor to be ignored in the formation of foreign policy or even domestic policies. This can be seen in almost any group of states, whether developed or not. Furthermore, during governmental transitions, it would seem, the importance of the military is ever increased because this transformation of the government may present an opportunity for intervention and increase the popularity of the army (Finer, 1988, pp. 65-76).

The primary focus on the relationship between military and democracy is based on conclusions from Beatriz (2008); she claims her findings show that the biggest probability of transition occurs between militaristic and democratic governments. While the probability of a military government turning into democracy is 0.0455, the second most likely transition is from a hegemonic party system with a probability of 0.0188, which clearly points to the intricacy of the dynamics of the former relationship (Beatriz, 2008, p. 737). However, is increased military funding during transitions more likely to invoke interventionist intentions, or otherwise? There is no consensus among literature; however, the two different views are well summarized and developed by Clardie (2010).

Increase Resource Theory

There are two main reasons identified in the literature showing that increasing army resources is beneficial to democratic transition. The first is simply to appease the military. The logic of this argument springs from the notion that democratic transitions result from bargaining between rivaling political groups and it is very important that the military should not perceive itself a loser in the transition (Clardie, 2010, pp. 3-5) . In other words, should officer corps see potential damage (e.g. budget cuts), they may oppose democratization and push their own agenda. To avoid this, a government should give the army plenty of new toys to play with while politics are resolved by politicians. It is in the interest of a new government to increase resources and keep the army satisfied.

The second reason to increase resources is to ensure domestic stability in a post-transitional polity. The basis of this argument recognizes that often transitions are marked with domestic turmoil and conflict between rivaling political forces. It may be a conflict between old groups that found a place and time for their competition during this fragile period, or it may be a result from an inclusion of new players in policy making. In any case, a new government faces an insecure and unstable environment (Clardie, 2010, pp. 6-7). To ensure a government’s survival a strong military is required so that it may protect the work of legislatures and provide order for a further development of democracy. Thus, the military may require an increase in resources to cope with domestic tensions. However, this might be a double-edged sword; the military may grow dissatisfied with a new democratic government and use these newly acquired resources as means to perform a coup d'état . This particular paradox leads into the argument for a decrease in resources.

Decrease Resource Theory

Similarly to the Increased Resources Theory, this school of thought also provides two reasons for why a decrease in resources is more likely to be beneficial for democracy. The first reason is that by increasing military strength, a government increases the chance of military intervention because the army will have greater abilities to ensure a successful coup (Clardie, 2010, p. 8). Publics and leaders can be suspicious of the weaponry the military has at their disposal and the many recruits who can be equipped. For this reason, it is best to decrease military expenditure thus limiting its scope of influence.

The second reason to decrease resources focuses on the negative impact an increased military expenditure may have on society. It is important to a newly constructed government to focus on providing better education and other social benefits. Developing rational welfare, taxation and budgetary systems is incredibly important for a healthy advancement of a political culture. In this case, an increase in military funding can result in a decrease in social expenditure and therefore a limited ability to provide public benefits associated with democracy (Clardie, 2010, p. 9). If a government does not deliver these benefits the public can grow discontent with a new government, thus creating a possible support base for opposition movements.

This can be a motive for the military to oppose the government as well. Motives for the disposition of government deserve a separate discussion, which due to limitations of this essay had to be excluded. However, the military as a defender of the nation as opposed to the government – Finer named this phenomenon “the manifest destiny of the soldiers” – can denounce the government as being incapable of delivering public benefits and assume civil control by means of violence or blackmail (Finer, 1988, p. 28). It brings us yet to another paradox: if one increases social expenditure, then the military may grow dissatisfied because their needs are not being met; if one decreases social expenditure, then the military may assume that a government is not handling its job properly. However, logic of state development proposes that decreasing resources to the military may allow governments to spend more on economic development, health, education and administrative infrastructures, all of which should increase the survivability of democracy as concluded by Geddes (1990) and Londregan and Poole (1990).

The Corporate Interest

We have established the two theories this essay will test, but there is one point recurring in both that needs to be extrapolated, namely the apprehension officer corps hold for the military’s survival and efficiency. This corporate interest suggests a concern with the preservation of hierarchy, discipline, and cohesiveness within the army; independence from civilian intervention; and funding to attract new recruits and acquire advanced weaponry (Geddes, 1990, pp. 126-27). In countries where joining the military is one of the conventional career paths, acquisitive motives can be assumed to rank high in most officer corps’ preferences, if only because the existence of opportunities depend on the survival of the military.

Some officers are tempted to intervene during transitions, others have high legalist or moral values that prohibit intervention and most are floating somewhere in between – but almost all care about the survival and efficiency of the military. Therefore, given the opportunity for intervention, officers want to move in or out of politics as a cohesive whole, because once factions split apart and take up arms against each other, it is very difficult to restore previous order, trust and unity (Geddes, 1990, p. 128). Given this concern officer corps hold for military strength, it is reasonable to believe that resources in the form of advanced weapons (aircrafts, tanks, missile batteries, etc.), increased salaries, health benefits or better housing for high officials can make a difference in the military’s behaviour and, therefore, in the regime’s outlook, as well.

Although opposition from outside the ruling group and numerous exogenous shocks (e.g. global economic crisis) sometimes decisively affect a regime’s survival, by focusing on internal dynamics of a state, specifically on the relationship between military and government, I will try to assess the level of influence it has on a regime’s transition to democracy. Often a military decision to intervene is made after carefully investigating the opportunity, developing popularity and acquiring motives. The interest of this essay is to see on which side of the scale the military expenditures are situated. Does increased funding push the army to intervene or is it otherwise? It should not be expected that a theory of social science will explain every bit of variance in the outcome. Most, if not all, theories do not and this is not an exception. It is expected, however, to record the degree and velocity of influence military funding has upon the democracy level of a regime, rather than claim that this is the only factor connecting the military to interventions.

H1: An increase in military funding (as percentage of GDP) will generate an increase in the level of democracy (as a score on Polity Combined scale).

Research Design

This essay will analyze cross-sectional data from 115 states. Using simple Ordinary Least Squares (OLS), military expenditures will be regressed against Polity Combined scores. Ideally time series data of each state would be more compatible, however due to time and space limitations this had to be avoided. The sample used in this essay was determined by the Pippa Norris data set (Norris, 2009). The democracy level variable Polity categorizes governments from strongly autocratic, which have a score of -10, to strongly democratic, with a score of +10. A principle by which states have been selected for this study is defined in terms of readily available data, so a state with a recorded Polity Combined value entered a test sample. This provided us with a very vibrant data set which includes developed democracies such as the US, Australia, Scandinavian countries and most of the regimes in western Europe, but also developing regimes found in Latin America, Africa, and Southern Europe, followed by autocratic regimes from the Middle East, and parts of Africa. Our objective here is to see how different funding strategies towards the military affect the regime itself, and the sample of 115 different states will do just that. If nations with high scores on Polity variable will appear to be associated with high military expenditures then our hypothesis will be satisfied providing support for Increased Resource s theory; if the opposite is observed then Decreased Resources theory will hold true.

Dependent Variable

Democracy Level . The dependent variable is the 21 point Polity Combined scale recorded in 2002. The variable itself is constructed by conceptualizing qualities related to democratic and autocratic authority in governmental institutions, rather than by categorizing states according to restricted forms of state rule. It is this unique feature of the Polity Project that enables analysts to compare states through a spectrum of governing authority. The range of governments captured by the scale goes from institutionalized autocracies, through mixed regimes also known as ‘anocracies’ to fully consolidated democratic regimes (Marshal, 2011, p. 1). This variable records major qualities of executive recruitment, constraints on executive branch, and the level of political competition (Marshal, 2011, p. 1). Although one of the requirements of OLS is continuity of the dependent variable, a 21 point scale with a possible zero and values in between the points can be considered continuous. The structure of the dependent variable, although ordinal in its nature, does not violate any of the OLS assumptions.

Independent Variable

Military Expenditures. Military funding is operationalized as a part of general domestic product dedicated to the military in year 2000; the data is from the Pippa Norris dataset (Norris, 2009). However, there are two issues related to the measurement of the main independent variable that deserve discussion. The first problem concerns the reliability of data. Official reports of military expenditure do not tell the whole story about what is happening in the budget (Clardie, 2010, pp. 10-13). Officer corps can employ corrupt practices and this way gain resources outside the formal budgetary. It would be very comfortable if official reports included the amount of resources achieved this way, however, they do not; thus it is a reasonable case for some measurement error. Another issue which might increase the probability of a measurement error is that the measurement error associated with a possible engagement in corruptive practices (mentioned earlier) may be correlated with low scores on the Polity scale (Clardie, 2010, pp. 10-13). Put differently, the military that uses corrupt means to attract additional resources, has the potential to damage democratic development and lower the chances of a successful consolidation of democracy. This means there is reason to believe that the model might have some bias since the measurement error associated with military funding is correlated with the dependent variable.

In order to better grasp the nature of the data used in our model, a table of descriptive statistics is presented below. From Table 1 we can see that our test sample is relatively anocratic, the mean value of Polity Combined is 4.10, suggesting that this particular group of states is open to liberal institutions. The lowest value of -10 is observed only once and assigned to the Kingdom of Saudi Arabia, a regime with few, if any, opportunities for representative participation in state affairs. On the other hand the positive end of the scale is occupied by various consolidated democracies ranging from the US to Eastern European states such as Lithuania. Military expenditure as percentage of GDP, on the other hand, is a more dynamic measure with a mean value of 2.76 percent. This offers us an interesting perspective on the importance of the military in modern times, only a 30 th part of the budget is assigned to the military across the sampled states. For this variable the low end of zero is uniquely and solely the virtue of Costa Rica that abolished armed forces in 1949. In contrast, the highest sampled military expenditure belongs to Eritrea, a country that fought a 30 year long war of independence against successive Ethiopian governments since 1960.

Table 1: Descriptive Statistics

Min. Value Max. Value Mean Std. deviation
Polity Combined -10 10 4.10 6.23
Military Expenditure 0 23.5 2.76 2.88
Ethnolinguistic Fractionalization 0.002 0.93 0.44 0.24
Internal Conflict 0 1 0.17 0.38
Electoral Systems 0 1 0.24 0.43
GDP per capita 90 41974 7019 10039

Control Variables

GDP per capita. A first control included in the model is the wealth of the country measured as GDP per capita in the year 2002. The most prosperous nation of the sample in that particular year was Norway with a GDP of 41,974 US dollars, while its opposite Ethiopia had a GDP of only 90 US dollars. Zakaria ( 2007, pp. 96-100) , for example, claims that economic prosperity is one of the necessary conditions for consolidation of democracy and in this instance our two countries score a 10 and 1 on Polity Combined scale respectively . Therefore, it is expected that GDP per capita will be positively related with a democracy level.

Ethnic Fractionalization. A second variable the model controls is ethnic fractionalization present in 2002. It is measured as a ratio varying between 0.0 and 1.0 with low points associated with low levels of ethnic fractionalization. This variable suggests that the more diversity there is in a state, the more difficult it is for a democratic process to develop because governmental institutions will have more problems including all interests into policy making. Even though it is neither a sufficient nor a necessary condition for democratic development – there are many prosperous democracies, as for example the US itself, where the electorate can be divided across ethnic and religious lines – this study expects this variable to be negatively related with a level of democracy.

Internal Conflict. This control is coded as a dummy with ‘1’ for those states that experienced an armed conflict between a state and its population in the year 2000. Domestic turbulence in the form of violent encounters is very much a threat to democracy and may significantly affect its further development, especially if a regime has just experienced a transition. A mean value of the dummy variable shows only a binomial distribution and from this we make out that 17 countries have suffered from internal turmoil among those being some of the more autocratic regimes of Rwanda with a Polity Combined score -4, Chad -2, Uganda -4 and Algeria -3. Internal conflict, therefore, should be negatively related with a level of democracy.

Electoral Systems. This is another dummy control included in the model. It is coded as ‘1’ for those countries with legitimately working electoral systems in the year 2001. ‘Comparative Study of Electoral Systems’ is an international project that tries to record turnout levels, vote choice, party affiliations and the number of candidates. From Table 1 we see that 24 countries have been coded as having legitimate representative institutions and most of those are assigned to the developed world. Politically active publics tend to be very responsive in regard to changes in the administration; therefore the scope of military’s influence in such societies is limited by the process of constitutional legitimization of authority. It is expected that politically active societies will have a positive effect on the level of democracy.


The method employed to test the hypothesis is OLS. It is simple to use and is known for its wide range of applicability. With the types of variables selected for this model there should not be any issues regarding the mechanics of OLS. A test for multicollinearity did not show any significant correlations between independent variables, suggesting that the calculations of individual predictors do not influence each other. In other words, their effect is independent, which contributes to the model’s accuracy.

However, there are several problems concerning cross-sectional data and this model in particular. Firstly, cross-sectional models are likely to be heteroscedastic, this is usually solved by using robust standard errors and by accounting for clusters among certain Polity Combined points. Unfortunately the limits of this essay do not allow for such precise modeling, leaving this issue to be addressed by future research . Secondly, it would be false to assume that an increase or a decrease in military funding would immediately manifest itself in a change of the military posture. In time series data this is dealt with by lagging the variables by one or several orders. Here, however, to account for this, Polity Combined variable represents the year 2002 while military expenditures were recorded at the year 2000, giving us a lag of two years.


Regression results show significant support for the Decreased Resources theory. The model (Table 2) fails to satisfy the null hypothesis and a low value of p ratio (p=.000) indicates that the main independent variable, military expenditure as percentage of GDP, is significant. The relationship is a relatively strong one, R squared suggests that 40 percent of variation in the dependent variable can be explained using this model. Results of the OLS concluded that there is a negative relationship between the main independent variable, military funding, and a dependent variable, Polity Combined scale. In other words, as governments dedicate larger shares of their GDP, the democracy level in a state decreases or, to put it technically, a unit increase in military expenditure as a percentage of GDP will generate a 0.8 decrease in Polity variable.

Table 2: OLS Regression Table

The Effect of Military Expenditure upon Democratization (OLS)
Coefficient Standard Error t-Ratio
Military Expenditure -0.79*** 0.17 -4.84
Ethnolinguistic Fractionalization -5.98*** 2.15 -2.78
Internal Conflict 1.43 1.29 1.11
Electoral Systems 3.57*** 1.27 2.8
GDP per capita 9.16E-5 0.0 1.59
Constant 7.14 1.33 5.39
R^2=0.40, N= 115
*p<0.05, **p<0.03, ***p<0.1***

The model presented two significant control variables. The first is a presence of a politically active public, Electoral Systems . As expected, a politically educated culture is positively related to a level of democracy because for such a population, the government has to construct legitimate taxation and representation systems that put constraints on the ruling group, all of which increases the level of democracy. The second significant control is ethnic fractionalization; the negative coefficient implies that pluralistic societies will have more trouble building democratic regimes. It follows in political theory that generally a homogenous society is more likely to agree on policy decisions and find consensus regarding emerging problems (Zakaria, 2007, pp. 119-62). Therefore, the presence of conflicting interests may slow the process of democratization in the best case and cause a stalemate in the worst.

Two other controls, the presence of an internal conflict and GDP per capita, did not reach significance levels. Although the wealth of the country is considered to be one of the prerequisites for democracy development, this model could not record its effect; however, the variable came close to significance with p ratio of 0.1. On the other hand, Przeworski and Limongi (1997) even developed a benchmark of required wealth for consolidation of democracy of GDP per capita equal to 6000 dollars. As mentioned earlier, a model might suffer from inconsistency because it did not account for possible heteroscedasticity. Modeling time series data could eradicate this problem because the gradual accumulation of wealth would be captured more precisely. Internal conflict, although assumed to have a negative impact on democracy, also did not achieve significance. Looking at present democracies in the world, it is not always the case that having a militant opposition will damage the democratic government. For example, violent encounters between Israel and the Palestinian population do not seem to deteriorate Israel’s democratic process. Alternatively, India, the largest democracy in the world, is still able to conduct democratic elections despite having to control many Muslim extremist groups.

Practical implications of the negative relationship between military spending and a level of democracy point to the importance of agency, in this case the army, during transitions. Modern statehood invited the military to stand beside the politics and contribute to foreign and even domestic policy making. It should not be surprising though that politicians have difficulty handling modern warfare and conflicts, because the complexity of contemporary combat tactics, the expertise required to control state-of-art weapons, and a strong sense of respect GIs hold for higher officers are exclusively associated with responsibilities of the army. Yet, governments emerging as newly democratic should be very cautious of the scope of a military’s influence. Although there are reasons to believe that a new regime should immediately establish good relations with the military by giving extra resources, empirical testing of this relationship shows the opposite. State leaders should regulate the strength of the military by decreasing its resources as to safeguard the government from a military intervention. Weaker military will have a higher probability of an unsuccessful deposition of the ruling group. For this reason officer corps will be reluctant to join a conspiracy and likely to return to the barracks. On the other hand, whilst rationalizing the budget, state leaders should concentrate on building a politically educated society, developing constitutional constraints on authority and increasing economic well-being, all of which should increase the level of democracy.


This essay empirically tested the relationship between military funding and democracy. After assessing the literature on transitions and the characteristics of military, two testable theories were drafted. The Increased Resources theory supports the view that increasing military funding will appease the military and provide extra resources, so that the military will be sufficiently equipped to control domestic turmoil. Decreased Resources, which springs from the logic that lowering the share of budget dedicated to the military will allow the government to spend more on public benefits associated with democracy and limit the strength of the military, so that it will refrain from intervention. Accordingly, two hypotheses were tested by means of a simple OLS model. After the data was regressed a negative relationship at 0 percent level was observed between the democracy level and military expenditure, supporting the Decreased Resources theory.

These findings also contribute to a broader understanding of the democratization process. Given that the model has not been able to explain the rest of the 60 percent of variance, it is safe to assume that the process itself is very complex and affected by even more factors some of which cannot even be computed. For example, corruption associated with misconduct of budget allocation, misappropriation of office and even single prolific characters that can shift the power play to one side or the other. Therefore, while we scratched the surface of the relationship between military institutions and the regime itself, democratization should be viewed as a struggle of many forces for a vibrant society.


Beatriz, M. (2008). Credible Power-Sharing and the Longevity of Authoritarian Rule. Comparative Political Studies , 41(4/5), pp. 715-41.

Clardie, J. (2010). The Impact of Military Spending on the Likelihood of Democratic Transition Failure: Testing Two Theories. Armed Forces and Society, pp. 1-18.

Finer, S .E. (1988). The Man on Horseback . 2nd ed. London: Pinter Publishers.

Geddes, B. (1999). What Do We Know about Democratization after Twenty Years? Annual Review of Political Science , 2, pp. 115-44.

Londregan, J. B. and Poole, K. (1990). Poverty, the Coup Trap, and the Seizure of Executive Power. World Politics , 42(2), pp. 151-83.

Marshal, M. G. (2012). Polity IV Project: Political Regime Characteristics and Transitions, 1800-2009 . Center for Systemic Peace . Center for Systemic Peace, [online] available at: <> [accessed 15 March 2011].

Norris, P. (2009). Democracy Cross national Data Release 3.0, Spring 2009. John F. Kennedy School of Government, Harvard University, [online] available at: <> [accessed 15 March 2011].

Przeworski, A. and Limongi, F. (1997). Modernization: Theories and Facts. World Politics , 49(2), pp. 155-83.

Zakaria, F. (2007). The Future of Freedom . 2nd ed. London: W. W. Norton & Company.

©Gediminas Blazys. This article is licensed under a Creative Commons Attribution 4.0 International Licence (CC BY).