Keywords

Introduction

In 2022 worldwide military spending peaked largely due to the invasion of Russia in Ukraine. While the two direct partners involved in the conflict increased their spending significantly (Ukraine by 640% and Russia by 9.2%), also other countries augmented their defense expenditure resulting in a worldwide increase of 3.7% (in real terms) compared to the 2021 numbers (Sipri, 2023). While defense spending of the enemy is an obvious determinant of a country’s military expenditure, defense spending of an ally also has an important influence. Depending on the nature of the military spending (having more public or more private benefits) countries can opt to free ride on the spending of other countries. Burden sharing debates are hence nearly as old as military alliances themselves. In fact, since the Cold War, theoretical models on military spending have exactly focused on this burden sharing between allies. First, we will give an overview of military spending data over the last three decades. The second paragraph gives a synopsis of the theoretical literature on burden sharing, while the third paragraph relates the data on military spending to the Gross Domestic Product (GDP) in order to frame the burden sharing debate. In the discussion part we nevertheless highlight the caveats one should keep in mind when comparing military expenditure data and debating defense burdens.

Military Spending: An Overview

The Stockholm International Peace Research Institute (Sipri) yearly publishes a database covering data on military spending for 173 countries. While for some countries the database goes back to 1949, this is not the case for all countries. Given the fact that for the Soviet Union data is only available from 1992 onward, Fig. 1 shows the evolution of worldwide military spending since this date. The figure is based on expenditure per region expressed in constant (2021) US$ (billions).

Fig. 1
figure 1

Military spending (in constant 2021 billion US$) by region, 1992–2022. (Source: SIPRI Military Expenditure Database (consulted at June 1st 2023))

Full size image

Figure 1 immediately highlights some important findings. First, over the last 30 years worldwide military spending has increased by 69%, i.e. from $1287.9 billion in 1992 to $2181.9 billion in 2022. This increase was unequally divided over the four decades with the first period, from 1992 to 2000, military spending even decreased by 8.5%. This decline reflects the de-investment in military capabilities after the end of the Cold War which reached its nadir at the end of the nineties. Induced by the terrorist attacks on the United States of 11 September 2001, the beginning of the new century indeed saw a new era of increased military spending with yearly increases until 2011. While the growth numbers were largely fueled by increased US spending (war on terror, war in Afghanistan and Iraq), also other countries such as China and Russia started to invest more in their armed forces. The year 2011 was a turning point as total military spending decreased largely due to the effects of the 2008 financial crisis and the budgetary deficit problems in most Western countries. Total world spending continued to fall until 2015 from which it started growing non-stop until today. Again the rise was motivated by rising geopolitical tensions, in case the annexation of Crimea by Russia.

Second, the figure also underlines the uneven distribution in military spending by region. The bulk of the spending is done in America representing about 40% of worldwide military spending. The share of the Middle East always fluctuated between 6.4% and 8% just to surpass the 10% threshold in 2014 and falling back to 8.25% in 2022. Military spending in Asia and Oceania marks a remarkable increase starting with 13% in 1992 to 20% in 2012 to more than 27% in 2022. The 27.7% share of Europe in total spending declined steadily to less than 20% just to arrive back at 21.8% in 2022. Africa is the world’s region with the lowest military expenditure with only a few years where its share is more than 2% of total world spending. Off course, these averages by region are often guided by the numbers of some leading countries in these regions. Table 1 shows the ten countries having the highest military spending in 2022. As the table also highlights the share in worldwide spending for each country, it immediately becomes clear that the data on regional spending are strongly dependent on the data from some countries (e.g. US, China, Saudi Arabia, …).

Table 1 Top ten countries in terms of military spending in 2022
Full size table

As Table 1 shows, more than 60% of worldwide military spending can be attributed to the 5 biggest spenders where almost 40% is even attributable to one country, the United States. The US remains indeed by far the greatest military spender in the world although the difference with its followers is narrowing. Figure 2 plots the share of the 10 biggest spenders over time.

Fig. 2
figure 2

Share in world military spending of 10 biggest spenders from 2022, 1992–2022. (Source: SIPRI Military Expenditure Database (consulted at June 1st 2023))

Full size image

The most striking element in Fig. 2 relates to the share of China as these numbers show a very steep increase. Chinese data are always to be interpreted with care for the country yearly presents an official defense budget but it is not always transparent and leading think tanks estimate actual spending to be much higher than the official numbers. This leads to different numbers (while SIPRI estimates Chinese military spending to be $292 billion in 2022, the International Institute for Strategic Studies puts the estimate on $319 billion (Center for Strategic and International Studies, 2023)). The trend in Chinese military spending is however clear, i.e. whereas in 1992 China merely accounted for more than 2% of worldwide military spending, today the country represents about 13%. Since Xi Jinping came to power, this share has surpassed the 10% level. In fact, military spending by China can be expected to increase even further as President Xi wants to modernize the People Liberation Army (PLA) by 2035 and transform it into a world class military by 2049. In this way, strengthening the PLA forms an integer part of China’s national strategy to achieve “the great rejuvenation of the Chinese nation” by 2049. While more modest in absolute numbers, the share of India in worldwide military spending also tripled over the timespan under study. Due to the increased spending from its neighbor China but also induced by the hostile relationship with Pakistan, India has doubled its defense budget since 2013 (see table) (International Institute for Strategic Studies, 2023). The three European countries in the list share a similar pattern, i.e. France, Germany, as well as the United Kingdom loose share over the last 30 years. While the three together accounted for more almost 14% in 1992, their cumulative share dropped significantly to less than 10% in 2022. Hence, today the three biggest European spenders together spend less than China on defense activities. The share of Russia today is more or less the same as it was in 1992, about 3.3% of world spending. The evolution of Russia’s share follows the expected pattern, i.e. a yearly decrease from 1992 onward to a nadir of less than 1.5% in 1998. Since 1999 Russia’s share remounted reaching a peak in 2016 where Russia’s expenditure represented more than 4% of worldwide spending.

While Russia’s war against Ukraine moved Russia from the fifth to the third largest spender, the impact of this war surpasses the numbers shown in Table 1. In fact, Ukraine itself would enter Table 1 at place 11, whereas in 2021 the country was only on the 36th place. After the Russian invasion, Ukraine increased its military spending by 640% thereby surmounting large NATO member countries such as Italy and Canada in the list of biggest spenders. The war at the Eastern borders of Europe also affected spending of other countries. Fearing Russia’s aggression, other former USSR countries also announced increased military spending in order to ensure territorial integrity. Lithuania for example passed a State Budget just to allow for an increase of defense expenditure to 3% of GDP in 2023, while Poland even targets a 4% defense burden (defense spending/GDP). Countries such as Sweden and Finland requested NATO membership. But also other countries reacted by increasing military spending. In fact, 23 Central and Western European countries increased their spending in 2022 leading to a sharp increase of about 13% (Sipri, 2023).

The Public Good Character of Military Spending and the Burden Sharing Debate

Throughout history countries have formed military alliances in order to collaborate on defense issues against a common enemy. One of the most famous (and long-lived) military alliance is the North Atlantic Treaty Organization or NATO. NATO was established by a group of 12 countries in the aftermath of World War II. The countries agreed to defend each other against attacks by non-members (the Warshaw Pact countries). Based on a doctrine of Mutual Assured Destruction (MAD) deterrence implied the launch of nuclear missiles as an answer to enemy attacks. Deterrence provided by nuclear weapons is a textbook example of a pure public good, i.e. a good with nonexcludable and nonrival benefits for the different member states. After the Soviet Union acquired missiles for a second strike, one could question this non-excludability (Kim & Sandler, 2020). If defense of one additional nation does not diminish the security effects for the other nations, defense is nonrival. Once collective defense is provided, it is also difficult or even impossible to exclude a member state (Khanna et al., 1998). Given these characteristics of collective security, member countries can benefit from the agreement regardless of their own contribution. Hence, a whole literature emerged on the aspects of burden sharing in military spending starting with the seminal article of Olson and Zeckhauser published in 1966. Their Public Good model argues that countries have an incentive to free ride on the defense efforts of others. Providing an economic theory to explain burden sharing dynamics, they also empirically showed that larger allies choose higher defense burdens (i.e. defense spending/Gross Domestic Product) than smaller ones and hence that the latter take advantage from the efforts of the former. The underlying reasoning for this so-called exploitation hypothesis is that smaller countries value the defense provided by the alliance more highly than the larger countries while the latter typically contribute more (due to a relative smaller disutility caused by increased defense spending). A second implication of the public good character of defense spending is that total spending in the alliance will be suboptimal. As member countries do not take into account the spill-in effects (i.e. the positive security effects for the other members) of their spending on other members, total spending will be suboptimal.

As the military doctrine shifted from nuclear deterrence to flexible response in 1967, so did the conceptual framework for military spending. The idea underlying this new doctrine was that one would react more proportionally to aggression from the enemy using only nuclear weapons in case of a severe attack. As this doctrine hence combined nuclear deterrence with the use of conventional warfare capabilities, defense was no longer a pure public good. Due to the exclusive character of conventional capabilities allies do become partial rivals since one can be excluded from the protection offered by this conventional weapon (Oneal, 1990). To account for this ‘new’ nature of military spending, a whole range of authors have complemented the original model of Olson and Zeckhauser. Most contributions refine and redefine the nature of defense spending. Sandler’s Joint Product model (1977) exactly describes defense spending as composed of both deterrence spending and protective spending. Since protection of a larger area (e.g. an additional member country) necessarily reduces the level of protection (for the existing member countries), this form of defense spending is rival. Pure protective weapons are exposed to a so-called thinning effect, i.e. their effect decreases in quantity and quality when a larger area is protected. Hence, the number of members in the ally now becomes an important variable and should be limited. Given that member countries also have to choose whether to invest in deterrence (public) or protective (private) measures, the joint product model moreover predicts that countries will undersupply collective deterrence compared to the protective efforts. While smaller member countries will still have an incentive to free ride on the effort of larger members, the differences also diminish with the relative weight of the private benefits. The relative share of private benefits in total output not only determines the degree of free riding but also the efficiency of total spending. If a country receives sufficient private benefits from defense spending, it will be more conducive to contribute thereby reducing the risk of suboptimal defense spending (Sandler & Hartley, 2001).

After the fall of the Berlin Wall and the collapse of the Soviet Union, NATO expanded with new member countries. As the new members were different in terms of defense preferences and risks, this also had an effect on burden sharing in the alliance. Moreover, as the threat of the Soviet Union as a direct enemy had disappeared, NATO started to focus more on peacekeeping missions (e.g. Bosnia and Kosovo) and on out-of-area operations (e.g. Afghanistan War). This change toward a crisis-management doctrine required for example more highly mobile forces which had an effect on defense spending and hence on burden sharing. Whereas military spending related to the war on terror clearly has higher private benefits for countries being attacked more, out-of-area missions are expected to also bring purely public benefits as all members benefit from a secure environment (Sandler & Shimizu, 2014).

With Russia’s military actions in Ukraine and the annexation of Crimea in 2014, collective security was back on the agenda and some European nations started to increase their defense spending. At the 2014 Wales Summit, the NATO member states committed themselves to spend at least 2% of GDP on defense while devoting at least 20% of this spending to the procurement of major new equipment and to related R&D. Hence, the change in the security landscape gave again rise to public benefit aspects of defense spending (Christie, 2019). While the aggressive actions of Russia in Ukraine only intensified leading to the invasion in 2022, recent years also witnessed a surge in the number of non-traditional security challenges. These challenges are often referred to as hybrid threats highlighting the combination of different, also non-violent, means. Hybrid warfare in the form of cyberattacks, disinformation campaigns, unconventional warfare, etc. again necessitate a different, i.e. non-conventional, answer. Hence, these new challenges also influence the composition and public nature of defense spending. While the diversity of the threats impede a generalization, one can rationally assume that defense spending as an answer to hybrid threats largely creates private benefits (e.g. the benefits of a well-established cyber-defense are clearly rival and excludable) (Balcaen et al., 2021). Hence, today military spending not only brings more public benefits (collective defense against Russia, antiterrorism campaigns), the advent of hybrid threats also increases the private benefits.

Burden Sharing: The Data

It should be no surprise that the majority of the countries listed in Table 1 are also the largest countries as expressed in terms of Gross Domestic Product (GDP). Indeed, as the GDP is an indicator for the economic wealth of a country, a larger GDP translates into more spending power of a country. Hence, in order to compare the economic burden of military spending, one mostly uses the defense burden as an indicator, i.e. defense spending/GDP. While it was former President Trump who explicitly threatened to withdraw from NATO due to unequal burden sharing, the debate dates way back almost to the birth of the military alliance. During the Wales Summit in 2014 all NATO member countries pledge to contribute 2% of their GDP to defense spending within the next decade (NATO, 2014). Even despite recent increases following the Russian invasion, not all member countries attain this minimum level to date.

As discussed in the previous paragraph, the unequal level of burden sharing can be expected to relate to the level of publicness (i.e. non-excludability and non-rivalry) of the benefits caused by defense spending. The higher the public character of these benefits compared to the private ones, the more room for free riding on the efforts of other (typically larger) member countries. Numerous studies have empirically tested burden sharing between NATO member countries over time by examining the correlation between capacity (as measured by GDP) and defense burden (as measured by defense expenditure/GDP). While Olson and Zeckhauser in their original paper found empirical evidence for the exploitation hypothesis, after 1967 there was no significant relationship. Also during the flexible response regime, the share of private defense benefits increased and studies find little (except for 1985) support for free riding behavior by smaller member countries until 2010. The period from 2011 to 2017 again presents signs of disproportionate burden sharing. Following the theoretical literature a possible explanation for this pattern is that from 2011 onward defense spending was again characterized by more pure public benefits. As mentioned earlier, this could indeed be the case as the period represented a re-born Russian threat as well as increased transnational terrorism. Actions against Russia but also against terrorist groups such as Al Qaeda decline worldwide danger and hence have a highly public character (Kim & Sandler, 2020). As these threats did not disappear until today (on the contrary), one can expect that the public benefits of defense spending are still significant. On the other hand, as also highlighted in the previous paragraph and extensively discussed by Balcaen et al. (2021), the emergence of hybrid threats is likely to also positively impact the private benefits of defense spending. Hence, while the former is expected to lead to more free riding (i.e. Russian threat and transnational terrorism), the latter (i.e. increased hybrid warfare) is expected to curb free riding behavior.

Based on the data on real GDP and Defense burden for all NATO members in 2022 (see Table 2), we calculate the Pearson’s correlation coefficient to test for the exploitation hypothesis. With a coefficient of 0.50582 these data suggest that there exists free riding from the smaller countries in 2022. The coefficient is also statistically significant (p = 0.00512). As can be seen from Table 2 however, the United States is a large outlier in our table, both in terms of GDP and in terms of Defense burden. If we re-calculate the correlation coefficient leaving out the US, we find no significant relationship between the size of a country and its defense burden leading us to reject the exploitation hypothesis.

Table 2 GDP and defense burden for NATO countries in 2022
Full size table

Of course, only comparing these burden sharing efforts does not take into account the differences in benefits each ally receives from the collective effort. Hence, some authors have already tried to link the burden sharing issue to these benefits. Often used metrics here are typically population or area (the underlying reasoning being that a larger population/area benefits more from collective security) (e.g. Kollias, 2008) or exposed border but also number of terrorist attacks (referring to the potential risk for a country) (e.g. Sandler & Hartley, 2001; Kim & Sandler, 2020). Taking into account the increasing use of hybrid threats it would also be reasonable to account for the vulnerability of a country vis-à-vis these new threats such as the number of cyberattacks (Du Balcaen et al., 2021).

Discussion

While our previous calculation did not really confirm nor reject the exploitation hypothesis, the burden sharing debate still carries on today, not in the least often sparked by quotes coming from political leaders of the United States. Better and more equally divided burden sharing was also again one of the topics during the recent NATO summits of Madrid in June 2022 and Vilnius in July 2023. As discussed previously in this chapter, burden sharing in NATO is intrinsically linked to the 2% rule. Although having the advantage of being very simple, it is dangerous to compare defense spending between countries without taking into account some important caveats.

Firstly, when comparing the defense budget of 2 countries one needs to compare like with like. In light of the current war in Ukraine one often compares the budget of Russia with that of the NATO allies which support Ukraine. A simple comparison between Russia’s budget and for example that of the United States seems to indicate a very unbalanced conflict where Russia is clearly in a disadvantaged position. Of course, military expenditure is a typical flow variable which, by definition, does not take into account the stock of military equipment a country already has available. Taking into account this large difference (Russia still has a lot of Cold War material), the exercise is still far from simple as the data on expenditure can also mean different things. While pensions for military veterans are accounted for in the NATO definition of defense budget, in the Russian data they do not explicitly fall under defense expenditure but are rather labeled as social policy expenditures. Moreover, Russia often uses state loans or guarantees coming from e.g. the Ministry of Industry to finance new military equipment. Of course, all expenses related to private military militias are also not included in the official budget. Hence, the Russian data clearly underestimate real military spending of the country which impedes direct comparison with for example spending in the United States. SIPRI does not only base its data on the data of the Ministry of Defence but also includes spending on military pensions, on paramilitary forces, the National Guard and the Border Service of the Federal Security Service as well as other spending by the Ministry of Defence (including for example education and healthcare) as well as the subsidies for the State Agency on Atomic Energy and the Baikonur Space Centre (Wezeman, 2020).

Differences in allocation of ‘security’ spending also exist within NATO allies and hence also comparisons of their military spending need to take account of these differences. Countering cyberattacks perfectly illustrates this point. While some countries developed cybersecurity capacities in the lap of their armed forces (e.g. Sweden’s military budget increase for 2023 will largely be spent on its Cyber-Defence Unit) some countries have installed a separate Cybersecurity Centre with the support of other Ministries (e.g. Home Affairs). If the latter is financed by non-defense budget, these efforts do not appear in the defense burden metric although they do have the same effect as if they were financed by defense budget.

A third impeding factor relates to the composition of the defense burden metric. As burden sharing relates defense spending to GDP, these metrics not only fluctuate as a consequence of changes in defense spending but also due to economic growth. A negative evolution of the GDP (e.g. due to the COVID pandemic in 2020) can already increase a country’s defense burden even without increases in its defense budget. Moreover, looking at the complete defense budget doesn’t always provide a clear image. As NATO members promised in Wales in 2014, they wouldn’t only strive to a 2% defense burden but also to 20% of the defense budget spent on military equipment. The reasoning behind this decision was to install some kind of efficiency standard for military spending. As can be seen from Fig. 3, there is a large dispersion in the spending per category between the NATO member countries. While Hungary spends more than 48% of its defense budget on equipment, Portugal barely spends 18% on new equipment. This high number for Hungary clearly illustrates the country’s defense modernization and capability-building program launched in 2017 but nevertheless stands out in the group of the other allies. But also for other smaller or medium countries it is difficult to obtain a continuous investment level of 20% in new equipment and numbers typically tend to fluctuate with investment plans. For bigger countries it is easier to reach the 20% investment goal (Struys, 2022). For the majority of the countries the personnel category takes the lion’s share of the budget. The average percentage spent on personnel is 44% with Portugal (63%) and Italy (62%) as the 2 extremes on the upper limit and Luxemburg (24%) on the lower limit. For some countries (e.g. Belgium with 49.5% personnel costs) these high numbers of personnel costs reflect a pension burden. And although these expenditures hence contribute to the 2% goal, one can hardly argue that they contribute to the military capabilities of the country. Hence, when measuring defense efforts, it is however important to not only measure inputs but also outputs. While a discussion on how to measure Defense output goes beyond the scope of this chapter, a seminal work here is Hartley (2011) and also the publication of Bogers and Beeres (2013) measures the output by studying the contribution of different countries to the Afghanistan Mission. How do the monetary inputs translate in concrete results? Beeres and Bogers (2012) provide a clear framework to see how inputs in the ‘military production process’ in terms of defense budgets transform to outputs such as number of deployable units or number of detected explosives. While a thorough discussion of this framework goes beyond the goal of this chapter, these authors show that the ranking of performances of different armed forces also largely depends on the selection of the output measure (e.g. deployability versus investment in material).

Fig. 3
figure 3

Defense expenditure per category for NATO members in 2022

Full size image

And last but not least, a correct comparison between defense budgets can only be done based on purchasing power parity exchange rates. Again, taking the example of Russia versus the United States, it wouldn’t be correct to just compare the rough nominal numbers for 1 $ cannot buy as much in the United States as it can in Russia. Personnel for example is much cheaper in Russia as the country still relies on conscription. A highly ranked officer in the Russian army earns approximately 25–30% of his British homologue (Wezeman, 2020). Not only personnel is cheaper in Russia but also Military Equipment. As Russia also has a well-developed industrial base, the country doesn’t have to import much material from other countries. Connolly (2019) estimated the difference in military expenditure expressed in market exchange rates or expressed in Purchasing Power Parity (PPP) exchange rates for Russia. His estimates show that expressed in PPP Russian military expenditure was about 2.5 times higher than expressed in common market exchange rates. This means that the difference with the US military expenditure is also significantly lower as commonly assumed. For 2018 for example Russian military expenditure would be about 25% of US military expenditure instead of only 10%. Adjusting military budgets for purchasing power is hence important. As inflation for military equipment usually surpasses general inflation, one should probably even correct nominal numbers even more (Robertson, 2019).

Summary

The invasion of Russia in Ukraine not only massively increased worldwide military spending but also sparked the debate on burden sharing between NATO allies. This chapter gave an overview of the different theoretical models used in this debate. It is clear that the dominant military doctrine determines the public good character of defense spending and hence also the incentives for member countries to free ride. Given the current geopolitical climate and the advent of new (hybrid) forms of warfare, one can expect that this burden sharing debate is not to be settled any time soon. As this chapter however also largely emphasizes, the debate can only be constitutional if one takes into account the caveats which are related to the metrics used in the debate.

Cross-References