Available via license: CC BY 4.0
Content may be subject to copyright.
ARTICLE
Threats and the Public Constraint on Military Spending
Matthew DiGiuseppe1, Alessia Aspide1and Jordan Becker2,3,4*
1
Faculty of Social and Behavioural Sciences, Institute of Political Science, Leiden, The Netherlands,
2
Department of Social
Sciences, United States Military Academy at West Point, USA,
3
Brussels School of Governance, Brussels, Belgium and
4
Chaire
Économie de défense - Institut des hautes études de défense nationale, Paris, France
Corresponding author: Matthew DiGiuseppe; Email: mdigiuseppe@gmail.com
(Received 4 October 2022; revised 25 February 2023; accepted 23 July 2023)
Abstract
The public places an important constraint on funding security in Europe, and austerity risks making the
constraint tighter. Several recent studies show that curtailing military spending is a popular way to reduce
debt in Europe. Yet it remains unclear if military spending aversion persists when threats are salient. We
fielded an original survey experiment in Italy weeks before the Russian invasion of Ukraine to examine
how information about security threats influences military spending preferences and fiscal trade-offs.
We found that information about threats increases support for military spending. To validate the survey
experiment, we recontacted and remeasured our respondent’s preferences three weeks after Russia’s inva-
sion and find evidence consistent with our initial experiment. Our findings suggest that, while public
opposition to military spending remains high in Italy, external threats dampen the public’s opposition
to military spending, even under high debt burdens.
Keywords: NATO; military spending; survey experiment; fiscal policy; public opinion
NATO allies faced two key changes as they emerged from the coronavirus pandemic. The first is
the high debt burdens driven higher by the economic response to the health emergency in which
governments spent unprecedented sums to support sidelined businesses and workers. While ini-
tially generous, many states are now grappling with the prospect of repaying these debts as mon-
etary policy support is curtailed and fiscal space risks being further curtailed by rising inflation,
with the prospect of more stringent enforcement of fiscal commitments (IMF 2022). This is espe-
cially true in countries like Italy, which entered the pandemic with already high debt burdens,
and are likely to face strong pressure from creditors to credibly address their debt burdens.
The second challenge is bearing the costs of adjustment to increasing geostrategic instability.
The Russo-Ukrainian war has stimulated discussions of increasing defense spending in many
countries, but burden-sharing concerns have been a constant in NATO’s history (Sloan 2020).
While US President Donald Trump became known for his harsh criticism of allied investments
in defense (Becker et al. 2019), presidents from Kennedy (1963) to Obama (Gates 2011) also
expressed ‘aggravation’(Goldberg 2016) about European allies under-investing in defense.
Geopolitical events have led many allies –including Italy (Lanzavecchia 2022)–to recommit
to meeting the defense spending guidelines they agreed on at their 2014 Wales Summit
(Belgian MOD 2022; Scholz 2022; van Defensie 2022). However, public support for increased
spending is not guaranteed, not least because of the budgetary trade-offs they imply. For example,
*Jordan Becker is an active duty officer in the United States Army. This research is the authors’alone and is not official
U.S. government policy.
© The Author(s), 2023. Published by Cambridge University Press. This is an Open Access article, distributed under the terms of the Creative
Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction,
provided the original article is properly cited.
British Journal of Political Science (2023), page 1 of 18
doi:10.1017/S0007123423000443
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
in the days since Italy pledged to increase military spending in line with the 2 per cent of GDP
Wales Pledge, public sector workers have protested against the allocation of scarce resources
towards the military (Rathbone 2022).
Recent events provide an opportunity to probe two open questions regarding the relationship
between economics and security in Europe. First, what role do threats (or their absence) play in
support of military spending? Scholars have an incomplete understanding of how emergent
threats do or do not translate into new security spending. While Nordhaus, Oneal, and
Russett (2012) found that external threats correlate with aggregate military spending, operation-
alizing threats is challenging (Becker 2019; Haesebrouck 2019), as is identifying a causal link
between them and spending. Beyond looking at threats and public support for military spending
(Eichenberg and Stoll 2003), it has been difficult to establish if public support for military spend-
ing grows with external threats or if collective action problems (Olson and Zeckhauser 1966;
Schelling 1963) persist even in heightened threat environments.
Second, it is unclear how public opinion over military spending responds to increases in public
debt and calls for consolidation in the context of deteriorating international security environ-
ments in which threats appear to be increasing (Schilde 2017). While previous research has
demonstrated that fiscal constraints limit military spending in cross-national country-year obser-
vational studies (Christie 2019; DiGiuseppe 2015) and that cutting back discretionary spending
such as military spending entails lower political costs than in the case of entitlement spending
(Breunig and Busemeyer 2012), the theorized public constraint has yet to be convincingly estab-
lished. A stream of recent studies, conducted pre-pandemic, suggests that voters confronted with
policy options to reduce debt burdens find cutting military spending the most attractive (Bansak,
Bechtel, and Margalit 2021; Huebscher, Sattler, and Truchlewski 2021). However, these studies
were conducted in periods and countries with relative geostrategic calm and within surveys
that prioritize economic policy and neglect other government activities. Will such results hold
up in surveys asking respondents to consider the threat environment in addition to material eco-
nomic considerations?
We address these questions with a survey experiment fielded in Italy as Russian troops massed
on Ukraine’s border pre-invasion in January of 2022. To examine the effect of a heightened threat
environment, we randomly assigned half of our respondents to read a news article with actual
quotes about the potential consequences of a Russian invasion on Italian national security and
then measured respondents’fiscal policy preferences. We found that information about national
security threats increases military spending support by approximately 20 per cent of a standard
deviation and, in a multidimensional fiscal trade-off, decreases support for military spending cuts
to reduce public debt by 11–16 per cent on average.
To demonstrate the validity of our results, we recontacted respondents who received no infor-
mation about the threat, our control group, three weeks after the February 2022 invasion of
Ukraine and again measured their fiscal policy preferences. Responses were consistent with
our original survey experiment. Italians were more supportive of military spending post-invasion
and less willing to cut military spending to reduce government debt than they were pre-invasion.
Our findings have implications for several literatures. First, this is the only study we are aware
of that uses random assignment of information about security threats to assess public opinion
towards military spending. There have been several experimental studies on public opinion
toward the use of force (Kertzer 2023), but few studies have leveraged the randomization of
experiments to probe how people think about longer-term fiscal commitment towards security
beyond an isolated incident like an intervention (Tomz and Weeks 2013), and none have exam-
ined how such choices occur within the context of more realistic multi-dimensional policy trade-
offs. This is important because most of our knowledge about the relationship between public
opinion and military spending comes from observational studies (Williams 2015) and most of
these focus on the United States (Hartley and Russett 1992; Higgs and Kilduff 1993; Bartels
1994). While some studies address threats directly (Ostrom and Marra 1986), and others address
2 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
threat perceptions in Europe (Béraud-Sudreau and Giegerich 2018; Eichenberg and Stoll 2017),
no study to date, as far as we know, has used experimental data to analyze the relationship
between threat perceptions and public support for defense spending in Europe. By experimentally
priming threat, as we do here, we can have greater confidence in our estimates of the causal effect
of public threat perceptions on defense spending.
Next, our study has implications for those interested in public support for fiscal policies and
especially those concerned with the composition of austerity policies. We demonstrate that atti-
tudes towards fiscal trade-offs are malleable –they change based on exposure to information
about, in the case of our survey experiment, external military threats. As such, evidence that mili-
tary spending is not a priority for respondents (Bansak, Bechtel, and Margalit 2021; Huebscher,
Sattler, and Truchlewski 2021) is potentially overstated or is at least partially a feature of an
economic-oriented survey environment. When security is made salient and threats are identified,
citizens’appetite for military spending cuts is substantively moderated.
Third, our findings contribute to the political economy of security work disentangling the
complex foundations of budget composition preferences among electors. In particular, we iden-
tify micro-foundations for arguments that ‘guns versus butter’preferences are both conditional
(Hauenstein, Smith, and Souva 2021; Whitten and Williams 2011) and malleable (Williams
2019). Access to additional information affects respondents’attitudes toward budgetary tradeoffs
involving defense spending.
Lastly, our paper has clear policy implications for the current crisis and beyond. First, it shows
that worries about the commitment of European allies to NATO were potentially overstated.
Consistent with work by Schilde, Anderson, and Garner (2019), European allies are ready to
commit to defense when confronted with a threat, even when under fiscal stress. Second, it sug-
gests that a clearer understanding of an external threat among electorates may moderate a more
general tendency to free- or easy-ride (Cornes and Sandler 1984) on the investments of allies.
Since 2014, threat perceptions moderating free- and easy-riding has been apparent –at least anec-
dotally: Douglas Lute, US Permanent Representative to NATO from 2013 to 2017, argued that
‘the president most responsible for NATO’s increased defense spending was neither Obama
nor Trump, but Putin’(Lute 2021). This argument was made in reference to Russia’s illegal
annexation of Crimea in 2014 –the Russian attack on Ukraine in 2022 has yielded further evi-
dence of European willingness to invest in defense when threats are apparent to voters. Moreover,
responsiveness to the threat of public opinion on budget composition suggests that subsequent
multilateral and institutional pressure to bear a fair share of the responsibility for collective
defense may also bear fruit. Not only did states geographically vulnerable to Russian aggression
decide to further increase defense spending (Janes 2022; Rolander 2022; Vanhanen 2022), but
geographically distant states like the Netherlands (van de Broek 2022) and traditional defense
spending laggards like Belgium and Germany (Belgian MOD 2022; Scholz 2022) did as well.
Our study identifies some of the micro-foundations of these choices and tests them empirically.
Does the Public Constrain Defense Allocation?
States’requirements to invest in military capabilities to secure their territory against external threats
and meet international obligations (like NATO’s Wales Pledge on Defense Investment) compete
with the need to ensure electoral competitiveness by providing public or private goods at home.
This ‘guns versus butter’trade-off often implies that electorates’desire for ‘butter’constrains demo-
cratic governments’provision of security (Powell 1993; Whitten and Williams 2011).
Anecdotal evidence from Italy provides a key illustration of how public constraints impact
military spending. Following NATO’s 2014 Wales Pledge on Defense Investment, American offi-
cials traveled to allied capitals to reinforce the importance of the pledge to Washington and check
on progress. When the US Permanent Representative to NATO discussed the topic with senior
Italian officials, he learned that Italy had emerging military shipbuilding projects that it did
British Journal of Political Science 3
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
not plan to include the costs of in its reports on military equipment expenditures to NATO allies
because, in a time of fiscal austerity, such expenditures would be difficult to justify domestically if
they were made public (Lute 2022).
This tension between domestic politics and defense spending is not limited to Italy. Although
many allies and thought leaders (Chollet, Keil, and Skaluba 2020) have suggested a focus on
defense ‘outputs’rather than ‘inputs’, states have been reluctant to discuss such outputs publicly.
The concept of defense outputs is in itself challenging –economists generally argue that there are
no concrete measures for defense outputs the way there are in private markets (Hartley 2012).
While inputs can be clearly quantified in the form of defense expenditures (and might be con-
sidered ‘costs’by economists), outputs (which may be considered ‘benefits’) can come in the form
of non-quantifiable security benefits, or economic spillover effects, both of which differ among
individual allies. ‘The value of defense items can only be ascertained by considering counterfac-
tual consequences: What would the world look like absent those expenses?’(Cooley et al. 2022).
NATO, however, has agreed on ‘output metrics’, which it cites specifically in the Wales Pledge on
Defense Investment (NATO 2014)–it is to these output metrics that commentators like those
cited above are referring. In economic terms, they are ‘measures of intermediate, rather than,
final outputs such as protection, security, safety, peace, and stability’(Hartley and Solomon
2015). In practice, allies can argue that even when their ‘inputs’are lagging, they may be so effi-
cient that their ‘outputs’are not. However, only Denmark (NATO 2013) has elected to publicly
release the outputs report NATO Ministers of Defence agreed upon, and outputs tracked by the
European Defence Agency correlate strongly with NATO’s defense spending metrics (Becker
2017). This further suggests that allies are reluctant to engage in detailed public discussion of
the politics of defense spending and capabilities acquisition.
While the public has often been theorized to be a constraint on military spending (Becker
2019; Carter, Ondercin, and Palmer 2021; DiGiuseppe 2015; Fordham and Walker 2005), few
studies have attempted to probe this mechanism directly. However, in line with anecdotal evi-
dence, they generally find that military spending correlates strongly with public preferences for
military spending (Eichenberg and Stoll 2003; Hartley and Russett 1992, 199). Relatedly, policy-
makers do seem to be responsive to the public’s foreign policy preferences. Two recent studies
show that public opinion has a causal effect on policy makers’stated policy preferences by ran-
domly providing information about public opinion concerning the use of force in Israel (Tomz,
Weeks, and Yarhi-Milo 2020) and UK power projection in the South China Sea to legislators
(Chu and Recchia 2022). Consequently, there is a reasonable expectation that the public’s
voice on defense concerns, which is often considered divorced from domestic politics, is
heard. The underlying mechanisms of this expectation merit further exploration.
Does the Public Prefer Butter to Guns?
If the public is an important constraint on military spending, it is worth asking whether the pub-
lic is generally averse to military spending when public preferences change. Here, again, there is
limited research within the security literature that addresses the question at the microfounda-
tional level by exploring public opinion. Existing studies all rely on observational data overwhelm-
ingly drawn from US samples (Knopf 1998) and have limited temporal coverage. One exception is
Fay (2020) who examines the correlates of support for military spending among thirteen NATO
countries. He finds that attitudes towards NATO and a strong economy are predictive of unidi-
mensional support for military spending. However, all studies deviate from the ‘guns versus but-
ter’theoretical framework and only ask about unidimensional support for military spending that
does not explicitly invoke budgetary trade-offs that are at the heart of fiscal policy decisions and
preferences. In other words, these studies examine questions that pose an increase in military
spending without the acknowledgement of its fiscal consequences or closed answer choices
that allow respondents to choose among various options.
4 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Researchers focused on fiscal policy preferences have found evidence pointing toward popular
distaste for military spending. Recent studies in this research area have probed multidimensional
preferences for budget trade-offs. For example, Bonica (2015) and Barnes, Blumenau, and
Lauderdale (2022) allow respondents to select their preferences for spending packages along mul-
tiple dimensions in the US and UK, respectively. Both find little appetite for military spending,
especially among left-wing voters. Notably, several studies have examined support for budget
consolidation or austerity policies in the context of debt reduction. Huebscher, Sattler, and
Truchlewski (2021) assessed preferences for austerity policies in the UK, Spain, and Germany.
Here they found that in all three countries, citizens, when given the choice to determine the com-
position of spending cuts, strongly prefer that most spending cuts come from defense expendi-
tures. Experimental evidence also supports this notion. Bansak, Bechtel, and Margalit (2021)
asked respondents in the UK, Spain, Italy, Greece, and France to compare pairs of austerity pol-
icies that had randomly distributed elements like spending cuts on different government pro-
grammes or tax increases on different segments of society. They found that hypothetical
programs that included military spending cuts had the most support compared to the average
proposal. Lastly, in a study of US voters, Roth, Settele, and Wohlfart (2021) find that simply pro-
viding the correct information about the size of the debt-to-GDP ratio decreased support for
military spending in a unidimensional framework.
While security scholars generally acknowledge such a public constraint on military spending,
they continue to debate its specific nature and effects. Bueno de Mesquita et al.’s(1999) formal,
game theoretical analysis of the underpinnings of democratic peace theory found that such a
peace rested on two factors: that elected officials with large winning coalitions are more dedi-
cated to victory once war has broken out and that they are more careful about starting wars in
the first place –not that public opinion is systematically opposed to defense spending. Fordham
and Walker (2005, 142) find, however, ‘that, although other considerations may outweigh the
effects of regime type, liberal states tend to devote proportionally less to their militaries than do
autocratic states’. Subsequent work added nuance and conditionality to these findings, however.
Whitten and Williams (2011) found that at the national level, parties’approaches to defense
spending are driven simultaneously by two factors: first, attitudes toward internationalism
and, second, attitudes toward welfare spending. Internationalists who favour welfare spending
are most likely to support higher defense spending. Williams (2019, 1,194) finds micro-level
confirmation for this argument using non-experimental survey data: ‘voters who favor
increased welfare spending subsequently also support increased defense spending’, particularly
those voters who are interested in the government’s role in job creation. Becker (2021)found
that governments in Europe and North America seemed to understand these preferences –
increasing spending on military personnel at the expense of equipment modernization when
faced with rising unemployment. In another observational study, Hauenstein, Smith, and
Souva (2021) find that the apparent propensity of democracies to spend less on defense is,
in fact, driven by threats. Our analysis seeks to subject these theories to experimental analysis
at the micro-level.
So, while some researchers have found that the public has a distaste for military spending and
is willing to sacrifice security when confronted with options to reduce government debt and bud-
get deficits, other work points toward scepticism regarding these claims. In addition to this dis-
sensus in the literature, there are at least three other reasons for such scepticism. First, these
studies were conducted in a period when respondents’countries were experiencing relative
peace and concerns about security were minimal among the public. Second, these survey envir-
onments primed respondents to think largely about economic priorities and consequences.
Notably, the surveys asked respondents to consider austerity options and the consequences for
both personal and national economic priorities.
Furthermore, the finding that citizens neglect national defense to preserve their material
welfare is at odds with literature suggesting the public can be hawkish and are willing to
British Journal of Political Science 5
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
make sacrifices in the name of national defense. For example, Knopf (1998)showsthatUS
public opinion on military spending moved with Soviet actions during the Cold War. In the
context of Europe, Schilde, Anderson, and Garner (2019) suggest that European citizens are
more hawkish than conventional wisdom assumes. For evidence, they cite a Transatlantic
Trends survey devoted to studying foreign policy views, which suggests Europeans were on
average more likely to support budget cuts to domestic spending areas than defense when
asked about debt reduction (Everts et al. 2012). Notably, Europeans were found to be less sup-
portive of defense cuts than the traditionally more hawkish American public. Together, this
suggests that the systematic evidence that defense is a low priority among citizens may be a
product of survey environments or lack of validity outside the peaceful period in which the
surveys were conducted. When foreign policy and threats are salient, citizens are potentially
more permissive of military spending even in times of economic duress. As such, concerns
that fiscal consolidation will further reduce the public’s appetite to contribute to collective
defense may be overstated.
Threats and Public Opinion
The idea that individuals are responsive to information about national security threats is
consistent with those studying support for military interventions. Several experimental
studies show this to be the case. For example, Malhotra and Popp (2012) measure attitudes
towards counterterrorism measures among respondents that were randomly distributed a hypo-
thetical probability of a terrorist attack in the next five years. Those that were provided with a
higher threat probability indicated more hawkish foreign policy views. Gadarian (2010) shows
that people express more hawkish views when presented with more threatening compared to non-
threatening contexts. Similarly, Myrick (2021) shows that non-partisan priming of threats can
influence foreign policy attitudes toward China among US respondents, while Kertzer and
Zietzoff (2017) show that partisan and social cues can have a great effect on an individual’s sup-
port for foreign intervention. Lastly, within the context of terrorism, Conrad et al. (2018) show
that support for torture increases when perceptions of threat levels are increased.
These and other studies suggest that attitudes toward foreign and domestic security policy are mal-
leable. Consequently, the nature of public attitudes towards military spending is also likely to shift
depending on both the survey context and real-world events. This also corresponds with research
that shows framing and elite cues have the potential to shift opinion on a variety of economic issues
including budgetary preferences (Ardanaz, Murillo, and Pinto 2013; Bansak, Bechtel, and Margalit
2021; Bisgaard and Slothuus 2018;FerraginaandZola2022; Slothuus and Bisgaard 2021).
The previous literature leaves open several questions. First, we have a poor understanding of
the malleability of citizens’preferences toward military spending and fiscal trade-offs. Evidence
indicates that in times of relative peace and within economic policy surveys, citizens are averse
to military spending. However, it remains to be seen if this will change as threats are made salient.
As discussed above, there is good reason to believe citizens’attitudes in this area are subject to
change through increasing levels of threat perception (Gadarian 2010; Malhotra and Popp
2012), and that national policy is sensitive to threats as well (Hauenstein, Smith, and Souva
2021). Consequently, we expect that citizens will be more accepting of military spending when
provided with information about national security threats.
Hypothesis: Information about threats will increase support for military spending
We also pre-registered another main hypothesis and four ancillary hypotheses that returned null
results. First, we hypothesized that information about Italy’s debt would decrease support for
military spending, conditional on individuals’prior estimation of debt-to-GDP size, left-right
orientation, and employment status. Next, we hypothesized that information about debt would
6 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
condition the effect of threats on military spending. Notably, the debt information treatment does
not have a significant interaction effect with the threat treatment on the outcomes we report
below.
1
As such, we are justified in ignoring this treatment in the analysis that follows.
Methodological Approach
Typically, studies of military spending rely on observational country-year panel analysis. While
this research is useful for observing general correlations in military effort variation and how insti-
tutions shape spending (Digiuseppe and Poast 2018; Fordham and Walker 2005), it is less useful
for understanding the impact of more dynamic political constraints such as public opinion.
Further, relying on observational (non-experimental) studies of public opinion towards military
spending allows for comparisons across time but has a limited ability to identify causal effects
given the scarcity of appropriate instruments, exogenous shocks, or data on confounding factors.
In light of these limitations, we use randomization via survey experiments to help study the effect
of the threat environment on the relationship between external threats and military spending. To do
so, we recruit a nationally representative sample in Italy using a pre-existing internet panel (N=
1,544). We chose Italy because it is a hard case. Italy entered the pandemic with a high and long-
standing debt burden and thus faces a greater urgency for debt stabilization than other NATO allies.
2
Italy also has relatively fewer security externalities than other NATO countries given they are quite
far from Russia and have a sizeable political constituency that is relatively friendly to Russia
(Baldolini 2022). For example, among other large NATO allies, the Italian public has more confi-
dence in Putin and is more likely to view Russia as a ‘friend’in public opinion surveys. Among
Italians, 36 per cent were confident that Vladimir Putin would do the ‘right thing’in world affairs
in 2021 compared to a median of 22 per cent in nine other polled European states (Silver and
Moncus 2022). In 2022, 22 per cent of Italians still viewed Russia as friendly compared to an average
of 14 per cent in six other polled countries. Italy’s longest serving prime minister since the Second
World War, Silvio Berlusconi, reported exchanging gifts and ‘sweet letters’with Russian President
Putin in 2022, and has blamed Ukraine for the most recent iteration of the Russo-Ukrainian
War (Castelletti 2023). As such, information about threats may have a harder time priming concern
amongst Italian voters compared to other European publics. Lastly, Italy is the fifth most populous
European country in NATO and, compared to other NATO allies, disproportionally hosts US
troops. As such, public opinion within Italy is more consequential for NATO security than members
that contribute less to collective defense.
Our survey experiment uses events at the time of the survey as the basis for a realistic threat
treatment. We fielded this survey between 19 January 2022 and 4 February 2022. This was a per-
iod of heightened and highly publicized tension between NATO and Russia, but before the inva-
sion that started a month later. Following several questions about their concern for the
seriousness of public debt in Italy, we randomly present half the respondents with a fictional,
short newspaper article and an image as seen in Fig. 1. The fictional news article draws on public
reporting and actual official statements about the events. The article mentions that an invasion is
possible and that, because of Italy’s affiliation with NATO, there is the potential for the conflict to
escalate and draw in Italy. Further, we drew on existing documents to elaborate on the particular
consequences of a conflict for Italy. To do so, we drew on the 2015 Italian Ministry of Defense
‘White Paper for International Security and Defense’(Ministero della Difesa 2015). We aimed to
not only ask respondents to think about European security but also how it might directly impact
1
In the Supplementary Appendix, we present the results of the tests of the additional hypotheses on the debt information
treatment and demonstrate that the interaction between the two treatments, debt information and threat information, does
not threaten the inferences we report in the remainder of the manuscript.
2
We recruited respondents (aged 18 and over) through an existing online panel provided by the firm Respondi/Bilendi. We
stratified sampling based on age group, sub-national region, sex, and educational attainment. We present a comparison of the
Italian census to our demographics in the Supplementary Appendix.
British Journal of Political Science 7
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Italy, which we did by citing passages about destabilization in the Mediterranean Basin. Lastly,
since the threat is conditional on existing power, we mention uncertainty over Italy’s preparation
to deal with these threats with current capabilities.
We compare the effect of our treatment against a ‘pure control’instead of a placebo.
3
We do
this for several reasons. First, the only purpose of a placebo would be to hold the delivery of the
Figure 1. Threat treatment.
3
We demonstrate that the Supplementary Appendix that the treatment and control groups are well balanced across age,
sex, left-right orientation, and region.
8 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
treatment constant across experimental conditions. As such, placebo use in survey experiments is
often poorly justified (Porter and Velez 2022). Second, we were hard-pressed to find a placebo
that would not have its own unintended effects, increasing the likelihood of a type I error.
Further, solutions to this problem, such as using AI-generated content (Porter and Velez
2022), were not possible at the time in the Italian language. In the trade-off between risking a
placebo effect or potential treatment condition effects (for example the effect of reading any
news article), we chose the latter. As such, our findings are conditional on the assumption
that the method of treatment delivery has zero effect on the outcome rather than being condi-
tional on the fact that a placebo has zero effect.
After the experimental treatments, we measure our dependent variables. Following Roth,
Settele, and Wohlfart (2021) we first ask respondents three questions about their preferences
for three types of government spending: defense and national security, infrastructure, and educa-
tion. They are worded as follows: ‘Do you think the amount of government spending on [defense
and national security, education, infrastructure] should be increased, decreased, or remain the
same?’Responses take on one of five values: It should be increased a lot, it should be increased
a little, it should remain as it is, it should be decreased a little, it should be decreased a lot. The
question on defense spending serves as the first dependent variable of interest. The other two out-
comes serve to test if the treatment is working to prime threat, but not other considerations, as
intended.
4
We expect the threat treatment to have a negligible effect on these outcomes.
This initial dependent variable is easily understood but fails to consider the multidimensional
reality of budget trade-offs that citizens confront (Bansak, Bechtel, and Margalit 2021; Bremer
and Bürgisser 2023; Huebscher, Sattler, and Truchlewski 2021). As such, it may overestimate sup-
port for military spending because nothing is being sacrificed to provide it. We thus follow up
with two additional dependent variables that attempt to get closer to actual budgetary trade-offs
that occur when governments are aiming to reduce expenditures to repay government debt. As
such, we give respondents the option to trade off certain policies as well or choose to enact
both (which implies neglecting the national debt, a tradeoff in itself). In two separate questions,
we ask respondents about their preference for cutting military spending in relation to increased
taxes or domestic spending cuts in two separate questions. They are worded as follows:
Given the high government debt of the Italian government, experts have recommended that
Italy decrease spending to reduce debts incurred during the pandemic. Do you think Italy
should decrease military spending, domestic spending, or neither? [military spending,
domestic spending, neither]
If Italy had to choose between reducing military spending and raising taxes to reduce its
debt, which option would you prefer?’[reducing military spending, raising taxes, neither]
Results
Figure 2 presents the coefficients for the threat and debt treatments for three models estimating uni-
dimensional support for military, infrastructure, and education spending.
5
Each model, estimated
using OLS with robust standard errors, regresses both of these treatments on a different outcome
absent any covariates. For each model, higher values of the dependent variable indicate support
for increasing spending in the category concerned. We see that the threat treatment has a significant
and positive effect on preferences for defense spending. Substantively, the effect represents a 19 per
cent of a standard deviation increase in support for defense spending.
6
While this may appear small,
4
We present the distribution of all three variables in the Supplementary Appendix.
5
See the Supplementary Appendix for the distribution of these three variables.
6
We also show in the Supplementary Appendix that the results are robust to an ordered-probit estimation that accounts for
the non-continuous nature of the outcome.
British Journal of Political Science 9
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
it is consistent with the effect of other informational treatments (Haaland, Roth, and Wohlfart 2020,
202).
Further, we see that the threat treatment has a minimal or insignificant impact on other spend-
ing categories, notably education spending. The threat treatment does not significantly increase
preferences for education spending. As the lack of statistical significance does not enable us to
conclude that an effect is substantively negligible, to strengthen our claim of a meaningless effect
of threat treatment on education spending, we use an equivalence test (Hartman and Hidalgo
2018; Rainey 2014). In particular, we test for equivalence using the 90 per cent confidence inter-
val, which we report in Fig. 2. The upper bound of the 90 per cent confidence interval around the
education spending estimate is 0.16. It means that we could not reject the null hypothesis, that is,
we could not claim that an effect is negligible, for effects equal to or greater than 16 per cent of a
standard deviation. On the other hand, the threat treatment significantly increases preferences for
infrastructure spending (p < 0.042). However, the effect is half the size at 9 per cent of a standard
deviation. Moreover, the order of the questions was not randomised, with the question on infra-
structure spending always following the question on military spending. This order may have
primed respondents to interpret infrastructure spending as closely related to military infrastruc-
ture spending, while perceiving education spending as a clearly distinct spending category.
7
Thus,
we have greater confidence that the treatment is working,
8
at least partially, through individuals’
concern for national security matters and that it does not reflect a general preference for increased
spending in response to threats or new information.
It is also worth noting, again, that the threat treatment is moving all three spending categories
in a positive direction. This further suggests that a unidimensional outcome fails to account for
the budgetary trade-offs. Consequently, we next examine the impact of the treatments on defense
spending preferences when respondents are confronted with multidimensional trade-offs.
Given the questions about trade-offs (see above), each of these dependent variables can take on
three possible categorical outcomes in each question. In model 1, the three outcomes are decrease
military spending, decrease domestic spending, or neither. In model 2, the three outcomes are
reducing military spending, raising taxes, or neither. We estimate each model with a multinomial
logit. This requires we set one value of the dependent variable as the baseline. In both models, we
set the baseline condition to ‘neither.’
Figure 2. Effect of treatments on spending prefer-
ences.
Note: The thick and thin lines indicate the 95 per cent and
90 per cent confidence intervals, respectively, around the
coefficients from three separate OLS models estimated
with robust standard errors. Each model estimates the
effect of each treatment on three separate spending out-
comes (N= 1,544). We present the full regression results in
the Supplementary Appendix.
7
Moreover, in the Supplementary Appendix we report disaggregated data for four different military spending categories.
The threat treatment increases support for military infrastructure spending.
8
To further support this claim, we show that the threat treatment did increase support for NATO cooperation compared to
the pure control. This indicates that the treatment is moving concerns about national security as NATO is largely divorced
from national budgetary concerns. We note that the treatment did not increase citizens’probability assessments of Italy going
to war in the next five years.
10 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Table 1 presents the coefficients from the multinomial logit regression.
9
Relative to the ‘nei-
ther’condition, individuals are significantly less likely to support military spending cuts under
the treatment condition than when no treatment was delivered. For a more intuitive display of
the results, the top panels of Fig. 3 present two histograms of the choices individuals made
under the pure control and threat treatment conditions for two questions regarding options to
reduce sovereign debt.
First, notice that military spending is a preferred choice to reduce sovereign debt among the
respondents in both trade-off scenarios. This is consistent with the findings of previous research
that we discussed above (Bansak, Bechtel, and Margalit 2021; Huebscher, Sattler, and Truchlewski
2021). Next, we see obvious differences in support for military spending cuts under the different
treatment conditions that correspond with Table 1’s estimates. We next plot these results in the
form of the effect of the threat treatment on the change in the average predicted probability our
respondents will select each of the three possible outcomes in each model. The bottom panels in
Fig. 3 confirm that the threat treatment has a significant and substantive effect on individuals’
constrained fiscal policy choices. The bottom-left panel presents these marginal effects for a
trade-off between taxes, military spending reductions, or neither. We see that respondents are
11 per cent less likely to select military spending as an option for reducing sovereign debt
under the threat condition than they are under the control condition. The bottom-right panel
substitutes tax cuts for domestic spending cuts. We see that when confronted with a choice
between domestic cuts, military spending cuts, or neither, respondents are 16 per cent less likely
to select military spending under the treatment condition.
It is also interesting that the respondents are less likely to move their support from military
cuts to either domestic spending cuts or tax increases. Instead, the threat condition increases sup-
port for doing “neither’. We chose to include this option as a way to provide respondents with a
way to express displeasure with both options and to preserve the status quo. We can only specu-
late but it may imply that respondents are willing to delay fiscal consolidation in the presence of a
threat. This would be consistent with the idea that governments tend to rely on credit markets to
fund military expansions in the presence of domestic opposition to changes in domestic fiscal
policy (Becker 2019; DiGiuseppe 2015). However, more research is needed to understand citizens’
motivation for selecting the ‘neither’option.
Validating the Experiment Post-Invasion
A common critique of survey experiments in political science is that the treatments are ‘short,
fleeting, and weak (Mutz 2021, 230).’As such, one might doubt that such effects replicate in
response to real-world stimuli. In this section, we attempt to demonstrate that our results mirror
responses to the actual invasion of Ukraine.
Table 1. Multinomial logit models of budgetary trade-offs
Model 1 Model 2
Increase taxes Military spending cuts Domestic spending cuts Military spending cuts
Threat 0.328 −0.427*** −0.215 −0.681***
(0.224) (0.123) (0.158) (0.114)
Constant −1.361*** 1.306*** −0.734*** 0.684***
(0.177) (0.0903) (0.118) (0.0821)
N1,544 1,544
Standard errors in parentheses,*** p < 0.001, ** p < 0.01, * p < 0.05.
9
We show the results of the test of the IIA assumption in the Supplementary Appendix
British Journal of Political Science 11
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
We fielded our study between 19 January and 4 February 2022. Twenty days later, on the 24
February, Russia invaded Ukraine. This unfortunate turn of events provided an opportunity to
examine if our experimental treatment corresponds with real-life events. In effect, we can exam-
ine if the onset of the war changed opinions regarding spending priorities within-subject. To do
so, we recontacted respondents in our control arm between 11 March and 20 March 2022 and
Figure 3. Changes in discrete outcomes.
Note: The top panels (a) show the percent support for each consolidation strategy in the treatment (Threat) and control (No Threat)
conditions. The bottom panels (b) present the estimated change in average predicted probabilities by moving the treatment variable
from zero to one based on two multinomial logit models. The dot indicates the change in predicted probability for each of the three
outcomes estimated in each model. The lines indicate the 95 per cent confidence intervals around the estimate.
12 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
had them complete the survey again. We were successful in recontacting 86 per cent of the
respondents in the control arm of the original sample.
10
We also point out that the follow-up survey was not pre-registered as we moved fast to cap-
italize on the unfolding events. However, we have only made a few necessary changes to the ori-
ginal pre-analysis plan to adjust to the within-subject data structure. Given that we have two
observations for each individual, one pre-invasion and one post-invasion, we cluster the standard
errors in our analysis by the respondent.
11
We also adjust our inference criteria to account for a
one-tailed test given our updated expectations about the direction of the effect in Wave 1. Yet, we
report both the 90 per cent and 95 per cent confidence intervals in our presentation of the results.
It is also worth pointing out that the findings are not causally identified given that many events
happened between the two waves of our survey, such as the substantive increase in energy and
wheat prices. When confronted with a trade-off between cutting military spending, increasing
taxes, or postponing consolidation, respondents could have preferred the first option driven by
concerns about their already reduced disposable income. However, this event would make our
findings even stronger because we still find decreased support for military spending cuts.
Moreover, we find the information useful as a validation of the original experiment. Besides,
given the prominence of the invasion and the news coverage it did receive, it would be unlikely
that changes in our dependent variables from one wave to the next were due to other events that
did not stem directly from the invasion.
Figure 4 presents the analysis. Preferences for an increase in spending increased within
respondent for military spending while there was no appreciable change in preferences for edu-
cation or infrastructure. In the fiscal trade-off scenarios, we also see that the results are similar
pre- and post-invasion to our survey treatment. Our respondents are less likely to support cuts
to military spending after the invasion relative to before the invasion. They are also significantly
more likely to accept spending cuts and tax increases.
Among all the dependent variables, the ‘effect’of the real-life treatment is smaller than that
observed in the initial survey experiment. While this may seem surprising at first, we note that
Figure 4. Pre- and post-invasion on spending priorities. (a) (b). Panel 1: The lines indicate the 90 per cent (thick) and 95
per cent (thin) confidence intervals around the coefficients from three separate OLS models estimated with standard
errors clustered by respondent. Each model estimates the difference in the DVs from the pre-invasion of Ukraine to the
post-invasion of Ukraine (N= 1,154). The confidence intervals are derived from standard errors clustered by respondent.
Panel 2: These panels plot the average difference in predicted probabilities pre-invasion to post-invasion estimated
from two multinomial logit models. The dot indicates the change in predicted probability for each of the three outcomes
estimated in each model. The lines indicate the 90 per cent (wide) and 95 per cent (thin) confidence intervals around the
estimate. The confidence intervals are derived from standard errors clustered by respondent.
10
The attrition rate doesn’t correlate strongly with any key demographics or delivery of the treatment. We present a com-
parison of the respondents that we did and did not recontact in the Supplementary Appendix.
11
We have also run models with respondent fixed-effects that produce similar findings.
British Journal of Political Science 13
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
we did nothing to prime respondents in our follow-up survey to consider the invasion of Ukraine
in their responses. Changes in their opinions are, we assume, due to actual events. Thus, we
speculate that the drop in magnitude is likely due to the absence of a prime to think about
the events in Ukraine directly. The only treatment that respondents brought to this new survey
was their exposure to the events in between survey waves.
Conclusion
While ‘the sinews of war’, Cicero is said to have remarked, may be ‘infinite money’, govern-
ments’supplies of money are generally finite. States that are near the limits of their fiscal cap-
acity and that cannot count on the favorable opinions of creditors must confront a
guns-versus-butter trade-off. In a free society, that trade-off will be subject to democratic pre-
ferences. As existing empirical work in comparative political economy has shown (Bansak,
Bechtel, and Margalit 2021; Huebscher, Sattler, and Truchlewski 2021), democratic preferences
are biased strongly towards butter rather than guns. In this study, we have tested the sensitivity
of this preference to contextual insights about the international security environment. We have
shown that the public will relax their opposition to military spending when provided with
information about realistic national security threats. Our experiment has shown the finding
to be robust to both unidimensional and multidimensional policy choice frameworks.
Further, we have shown that real-world preferences correspond with the experimental
treatment.
This finding contributes to debates about the political economy of security and alliance politics
that are central to conversations about the current international order. Most of the existing litera-
ture, as well as conventional wisdom, suggests a strong public preference for ‘butter’over ‘guns.’
But by fielding a survey experiment in Italy, a challenging case that appeared at first glance to be
very likely to conform to this conventional wisdom, we identified mutability at the microfounda-
tional level –public opinion is responsive to contextual information about threats.
As the European and transatlantic security architectures –as well as the larger international
system in which they operate –face the challenge of high-intensity war involving great powers
in Ukraine, the relationship between public threat perceptions and willingness to invest in
national defense takes on greater policy importance. For instance, will European public opinion
continue to support Ukraine in the face of rising energy costs? Our findings suggest that this may
depend on national leaders’ability to frame the Russo-Ukrainian war as a threat to their own
states’national security.
By testing the relationship between threat framing and public attitudes toward defense spend-
ing in Italy, we have laid the groundwork for future work across Europe, and across elite and
popular opinion. It is far from certain that commitments to European defense will be able to
hang together strategically in the context of the material costs of the Russo-Ukrainian war, but
our findings suggest it is more likely than previously imagined.
Finally, our findings point to the micro-foundations of a critical transatlantic policy issue. If
citizens are more likely to support defense spending when primed to consider threats, allies
may give more consideration to the public components of joint threat assessments. Balancing
regional concerns in threat assessment, scenario development, and operational and defense plan-
ning is an evergreen challenge for NATO as a political alliance. Given that different allies experi-
ence threats differently, ‘rediscovering the early principle that allies should concentrate on those
tasks for which they are the most geographically suited’(Mattelaer 2018) holds particular prom-
ise. If national public opinion is indeed responsive to threat perceptions in general, it is likely that
such opinion is even more inclined to support higher defense spending that is focused on more
proximate threats.
Supplementary material. The supplementary material for this article can be found at https://doi.org/10.1017/
S0007123423000443.
14 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Data availability statement. Replication Data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/
DVN/EOWDU0.
Acknowledgements. Previous versions of this manuscript were presented at West Point, EISS 2022 in Berlin and EPSA
2022 in Prague. We want to thank Fenella McGerty, Dominika Kunertova, Niccolo Petrelli, Vytautas Koukstis and Irene
Menendez for helpful feedback and comments.
Financial support. This work was supported by the HORIZON EUROPE European Research Council under grant #852334.
Competing interests. None.
Ethical standards. The research was conducted in accordance with the conditions and principles set by the Ethics Review
Committee of the Social Sciences at Leiden University.
References
Ardanaz M, Murillo MV and Pinto PM (2013) Sensitivity to issue framing on trade policy preferences: Evidence from a
survey experiment. International Organization 67(2), 411–37. doi: 10.1017/S0020818313000076.
Baldolini S (2022) Salvini e Russia Unita: l’accordo che Letta chiede di disdire. Fondi di Putin o meno. la Repubblica.
Avai lable from https://www.repubblica.it/politica/2022/09/14/news/fondi_putin_ai_partiti_salvini_e_russia_unita_laccordo_
che_letta_chiede_di_disdire-365607286/
Bansak K, Bechtel MM and Margalit Y (2021) Why austerity? The mass politics of a contested policy. American Political
Science Review 115(2), 486–505. doi: 10.1017/S0003055420001136.
Barnes L, Blumenau J, and Lauderdale BE (2022) Measuring attitudes toward public spending using a multivariate tax sum-
mary experiment. American Journal of Political Science 66(1), 205–21. doi: 10.1111/ajps.12643.
Bartels LM (1994) The American public’s defense spending preferences in the post-cold war era. Public Opinion Quarterly 58
(4), 479–508.
Becker J (2017) Transatlantic Burden-Sharing: Origins and Strategic Implications. Defence-In-Depth. Available from https://
defenceindepth.co/2017/11/15/transatlantic-burden-sharing-origins-and-strategic-implications/
Becker J (2019) Accidental rivals? EU fiscal rules, NATO, and transatlantic burden-sharing. Journal of Peace Research 56(5),
697–713. doi: 10.1177/0022343319829690.
Becker J (2021) Rusty guns and buttery soldiers: Unemployment and the domestic origins of defense spending. European
Political Science Review 13(3), 307–30. doi: 10.1017/S1755773921000102.
Becker J et al. (2019) Transatlantic Shakedown: Does Presidential ‘Naming and Shaming’Affect NATO Burden-Sharing? In
Alliance Management in the 21st Century: Strategy, Political Economy, and Defense Planning. Antwerp, Belgium.
Belgian MOD (2022) Le gouvernement approuve l’avant-projet de loi de programmation militaire issue du plan STAR.
Beldefnews. Available from https://beldefnews.mil.be/le-gouvernement-approuve-lavant-projet-de-loi-de-programmation-
militaire-issue-du-plan-star/?lang=fr
Béraud-Sudreau L and Giegerich B (2018) NATO Defence spending and European threat perceptions. Survival 60(4),
53–74. doi: 10.1080/00396338.2018.1495429.
Bisgaard M and Slothuus R (2018) Partisan elites as culprits? How party cues shape partisan perceptual gaps. American
Journal of Political Science 62(2), 456–69. doi: 10.1111/ajps.12349.
Bonica A (2015) Measuring public spending preferences using an interactive budgeting questionnaire. Research & Politics 2
(2), 2053168015586471. doi: 10.1177/2053168015586471.
Bremer B and Bürgisser R (2023) Do citizens care about government debt? Evidence from survey experiments on budgetary
priorities. European Journal of Political Research 62(1), 239–63. doi: 10.1111/1475-6765.12505.
Breunig C and Busemeyer MR (2012) Fiscal austerity and the trade-off between public investment and social spending.
Journal of European Public Policy 19(6), 921–38.
Bueno de Mesquita B, Morrow JD, Siverson RM and Smith A (1999) An institutional explanation of the democratic peace.
American Political Science Review 93(4), 791–807.
Carter J, Ondercin HL and Palmer G (2021) Guns, butter, and growth: The consequences of military spending reconsidered.
Political Research Quarterly 74(1), 148–65. doi: 10.1177/1065912919890417.
Castelletti R (2023) Mosca in soccorso di Berlusconi: ‘Da Zelensky rabbia impotente.’la Repubblica.https://www.repubblica.
it/politica/2023/02/23/news/berlusconi_contro_zelensky_putin-389054834/.
Chollet D, Keil S and Skaluba C (2020) Rethink and Replace Two Percent. Atlantic Council. Available from https://www.
atlanticcouncil.org/content-series/nato20-2020/rethink-and-replace-two-percent/
Christie EH (2019) The demand for military expenditure in Europe: The role of fiscal space in the context of a resurgent
Russia. Defence and Peace Economics 30(1), 72–84. doi: 10.1080/10242694.2017.1373542.
Chu JA and Recchia S (2022) Does public opinion affect the preferences of foreign policy leaders? Experimental evidence
from the UK parliament. The Journal of Politics 84(3), 1874–7. doi: 10.1086/719007.
British Journal of Political Science 15
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Conrad CR et al. (2018) Threat perception and American support for torture. Political Behavior 40(4), 989–1009. doi:
10.1007/s11109-017-9433-5.
Cooley A, Nexon DH, Poast P, Alley J and Fuhrmann M (2022) Estimating alliance costs: An exchange. Security Studies 31
(3), 510–32.
Cornes R and Sandler T (1984) Easy riders, joint production, and public goods. The Economic Journal 94(375), 580. doi:
10.2307/2232704.
DiGiuseppe M (2015) Guns, butter, and debt: Sovereign creditworthiness and military expenditure. Journal of Peace Research
52(5), 680–93.
Digiuseppe M and Poast P (2018) Arms versus democratic allies. British Journal of Political Science 48(4), 981–1003. doi:
10.1017/S0007123416000247.
Digiuseppe M, Aspide A and Becker J (2023) Replication Data for: Threats and the Public Constraint on Military Spending.
Available from https://doi.org/10.7910/DVN/EOWDU0, Harvard Dataverse, V1.
Eichenberg RC and Stoll R (2003) Representing defense: Democratic control of the defense budget in the United States and
Western Europe. Journal of Conflict Resolution 47(4), 399–422. doi: 10.1177/0022002703254477.
Eichenberg RC and Stoll RJ (2017) The acceptability of war and support for defense spending: Evidence from fourteen dem-
ocracies, 2004–2013. Journal of Conflict Resolution 61(4), 788–813.
Everts P et al. (2012) Transatlantic trends survey inter-university consortium for political and social research (2012).
Inter-University Consortium for Political and Social Research.
Fay EM (2020) Individual and contextual influences on public support for military spending in NATO. Defence and Peace
Economics 31(7), 762–85. doi: 10.1080/10242694.2019.1668236.
Ferragina E and Zola A (2022) The end of austerity as common sense? An experimental analysis of public opinion shifts and
class dynamics during the COVID-19 crisis. New Political Economy 27(2), 329–46. doi: 10.1080/13563467.2021.1952560.
Fordham BO and Walker TC (2005) Kantian liberalism, regime type, and military resource allocation: Do democracies
spend less? International Studies Quarterly 49(1), 141–57. doi: 10.1111/j.0020-8833.2005.00338.x.
Gadarian SK (2010) The politics of threat: How terrorism news shapes foreign policy attitudes. The Journal of Politics 72(2),
469–83. doi: 10.1017/S0022381609990910.
Gates R (2011) The Security and Defense Agenda (Future of NATO). Brussels: United States Department of Defense.
Goldberg J (2016) The Obama Doctrine. The Atlantic. Available from https://www.theatlantic.com/magazine/archive/2016/
04/the-obama-doctrine/471525/
Haaland I, Roth C and Wohlfart J (2020) Designing Information Provision Experiments. SSRN Scholarly Paper 3644820.
Rochester, NY: Social Science Research Network. doi: 10.2139/ssrn.3644820.
Haesebrouck T (2019) Who follows whom? A coincidence analysis of military action, public opinion and threats. Journal of
Peace Research 56(6), 753–66. doi: 10.1177/0022343319854787.
Hartley K (2012) The Economics of Defence Policy: A New Perspective, 1st edition. London, New York: Routledge.
Hartley T and Russett B (1992) Public opinion and the common defense: Who governs military spending in the United
States? American Political Science Review 86(4), 905–15. doi: 10.2307/1964343.
Hartley K and Solomon B (2015) Measuring defense output: An economics perspective. In Melese F, Richter A and Solomon B
(eds), Military Cost-Benefit Analysis: Theory and Practice. New York: Routledge, pp. 36–74.
Hartman E and Hidalgo FD (2018) An equivalence approach to balance and placebo tests. American Journal of Political
Science 62(4), 1000–13.
Hauenstein M, Smith M and Souva M (2021) Democracy, external threat, and military spending. Research and Politics 8(4),
1–13. doi: 10.1177/20531680211049660.
Higgs R and Kilduff A (1993) Public opinion: A powerful predictor of U.S. defense spending. Defence Economics 4(3), 227–
38. doi: 10.1080/10430719308404762.
Huebscher E, Sattler T and Truchlewski Z (2021) Three Worlds of Austerity: Voter Congruence over Fiscal Trade-Offs in
Germany,Spain and the UK. SSRN Scholarly Paper 3747664. Rochester, NY: Social Science Research Network. doi:
10.2139/ssrn.3747664.
IMF (2022) Fiscal Monitor: Fiscal Policy from Pandemic to War. Chapter 1: Fiscal Policy from Pandemic to War.
Janes (2022) Poland to Increase Defence Spending to 3 per cent of GDP from 2023. Janes.Com. Available from https://www.
janes.com/defence-news/news-detail/poland-to-increase-defence-spending-to-3-of-gdp-from-2023
Kennedy J (1963) Remarks of President Kennedy to the National Security Council Meeting. Foreign Relations of the United
States, 1961–1963, Volume XIII, Western Europe and Canada –Office of the Historian. Available from https://history.state.
gov/historicaldocuments/frus1961-63v13/d168
Kertzer JD (2023) Public opinion and foreign policy. In Huddy L, Sears D, Levy J and Jerit J (eds), Oxford Handbook of
Political Psychology, 3rd Edn. New York: Oxford University Press, pp. 447–85.
Kertzer JD and Zeitzoff T (2017) A bottom-Up theory of public opinion about foreign policy. American Journal of Political
Science 61(3), 543–58. doi: 10.1111/ajps.12314.
Knopf JW (1998) How rational Is ‘The rational public’?: Evidence from U.S. Public opinion on military spending. Journal of
Conflict Resolution 42(5), 544–71. doi: 10.1177/0022002798042005002.
16 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Lanzavecchia O (2022) Italian Parliament Votes to Raise Defence Budget to 2 per cent of GDP. Decode39. Available from
https://decode39.com/3052/italy-defence-budget-2-gdp-vote/
Lute D (2021) Making the Wales Pledge - Email to Authors, March 2.
Lute D (2022) 27 January 2022 Interview with Former US Permanent Representative to NATO.
Malhotra N and Popp E (2012) Bridging partisan divisions over antiterrorism policies: The role of threat perceptions.
Political Research Quarterly 65(1), 34–47. doi: 10.1177/1065912910385251.
Mattelaer A (2018) Rediscovering geography in NATO defence planning. Defence Studies 18(3), 339–56. doi: 10.1080/
14702436.2018.1497446.
Ministero Della Difesa (2015) Libro Bianco per La Sicurezza Internazionale e La Difesa. Italian MOD.http://www.difesa.it/
Primo_Piano/Documents/2015/04_Aprile/LB_2015.pdf
Mutz DC (2021) Improving experimental treatments in political science. In Druckman JN and Green DP (eds), Advances in
Experimental Political Science. Cambridge: Cambridge University Press.
Myrick R (2021) Do external threats unite or divide? Security crises, rivalries, and polarization in American foreign policy.
International Organization 75(4), 921–58. doi: 10.1017/S0020818321000175.
NATO (2013) Metrics 2013 (Denmark). Available from https://fmn.dk/nyheder/Documents/Metrics-2013.pdf
NATO (2014) Wales Summit Declaration Issued by the Heads of State and Government Participating in the Meeting of the
North Atlantic Council in Wales. NATO. http://www.nato.int/cps/en/natohq/official_texts_112964.htm
Nordhaus W, Oneal JR and Russett B (2012) The effects of the international security environment on national military
expenditures: A multicountry study. International Organization 66(03), 491–513. doi: 10.1017/S0020818312000173.
Olson M and Zeckhauser R (1966) An economic theory of alliances. The Review of Economics and Statistics 48(3), 266–79.
doi: 10.2307/1927082.
Ostrom CW and Marra RF (1986) U.S. defense spending and the Soviet estimate. The American Political Science Review
80(3), 819–42. doi: 10.2307/1960540.
Porter E and Velez YR (2022) Placebo selection in survey experiments: An agnostic approach. Political Analysis 30(4), 481–
94. doi: 10.1017/pan.2021.16.
Powell R (1993) Guns, butter, and anarchy. American Political Science Review 87(1), 115–32. doi: 10.2307/2938960.
Rainey C (2014) Arguing for a negligible effect. American Journal of Political Science 58(4), 1083–91.
Rathbone JP (2022) Russian Menace Brings Abrupt End to the West’s‘Peace Dividend.’Financial Times, June 7.
Rolander N (2022) Sweden Aims to Boost Defense Spending to NATO Target. Bloomberg.Com, March 10. Available from
https://www.bloomberg.com/news/articles/2022-03-10/sweden-pm-aims-to-boost-defense-spending-to-2-of-gdp
Roth C, Settele S and Wohlfart J (2021) Beliefs about public debt and the demand for government spending. Journal of
Econometrics 231(2022), 165–87. doi: 10.1016/j.jeconom.2020.09.011.
Schelling TC (1963) Strategic problems of an international armed force. International Organization 17(2), 465–85. doi:
10.1017/S0020818300033853.
Schilde K (2017) European Military capabilities: Enablers and constraints on EU power? JCMS: Journal of Common Market
Studies 55(1), 37–53. doi: 10.1111/jcms.12444.
Schilde KE, Anderson SB and Garner AD (2019) A more martial Europe? Public opinion, permissive consensus, and EU
defence policy. European Security 28(2), 153–72. doi: 10.1080/09662839.2019.1617275.
Scholz O (2022) Policy Statement by Olaf Scholz, Chancellor of the Federal Republic of Germany and Member of the
German Bundestag, 27 February 2022 in Berlin. Website of the Federal Government. Available from https://www.
bundesregierung.de/breg-en/news/policy-statement-by-olaf-scholz-chancellor-of-the-federal-republic-of-germany-and-
member-of-the-german-bundestag-27-february-2022-in-berlin-2008378
Silver L and Moncus JJ (2022) Few across 17 Advanced Economies Have Confidence in Putin. Pew Research Center.
Accessed September 5. Available from https://www.pewresearch.org/fact-tank/2021/06/14/few-across-17-advanced-
economies-have-confidence-in-putin/
Sloan SR (2020) Defense of the West: Transatlantic Security from Truman to Trump, Second Edition, 1st Edn. Manchester:
Manchester University Press.
Slothuus R and Bisgaard M (2021) Party over pocketbook? How party cues influence opinion when citizens have a stake in
policy. American Political Science Review 115(3), 1090–6. doi: 10.1017/S0003055421000332.
Tomz MR and Weeks JLP (2013) Public opinion and the democratic peace. American Political Science Review 107(4),
849–65. doi: 10.1017/S0003055413000488.
Tomz M, Weeks JLP and Yarhi-Milo K (2020) Public opinion and decisions about military force in democracies.
International Organization 74(1), 119–43. doi: 10.1017/S0020818319000341.
van de Broek A (2022) The Netherlands Will Spend 2% of Its GDP on Defence in 2024. The Dutch Government Decided to
Substantially Increase Its Defence Budget with 40%. We Stick to the Wales @NATO Pledge. For Our Security. #priceless
Https://T.Co/1zBao2rXED. Tweet. Twitter. Available from https://twitter.com/AvandenBroek/status/1527753804657082369
van Defensie M (2022) Summary Defence White Paper 2022 –Publication –Defensie.Nl. Publicatie. Ministerie van Defensie.
Available from https://english.defensie.nl/downloads/publications/2022/06/01/summary-defence-white-paper-2022
British Journal of Political Science 17
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
Vanhanen H (2022) In 2023, Finland Will Raise Its Defence Budget by 20%, Taking It up to 2.25% of GDP. This Is a Signal
of Commitment to National Defence and Broader Regional Security. We Intend to Stick to NATO’s Goals. Tweet. Twitter.
Available from https://twitter.com/HenriVanhanen/status/1565727121544192004
Whitten GD and Williams LK (2011) Buttery guns and welfare hawks: The politics of defense spending in advanced indus-
trial democracies. American Journal of Political Science 55(1), 117–34. doi: 10.1111/j.1540-5907.2010.00479.x.
Williams LK (2015) Estimating the defense spending vote. Electoral Studies 39, 243–55.
Williams LK (2019) Guns yield butter? An exploration of defense spending preferences. Journal of Conflict Resolution 63(5),
1193–221. doi: 10.1177/0022002718785969.
Cite this article: DiGiuseppe M, Aspide A, Becker J (2023). Threats and the Public Constraint on Military Spending. British
Journal of Political Science 1–18. https://doi.org/10.1017/S0007123423000443
18 Matthew DiGiuseppe et al.
https://doi.org/10.1017/S0007123423000443 Published online by Cambridge University Press
