Bang for the Buck? Reassessing NATO’s 5 Percent Benchmark Through Comparative Defense Accounting

Abstract

The renewed emphasis on defense spending following Russia’s full-scale invasion of Ukraine has reshaped transatlantic security debates. At the Hague Summit of June 2025, NATO allies committed to invest 5 percent of gross domestic product (GDP) annually by 2035, with at least 3.5 percent allocated for core defense requirements and up to 1.5 percent for defense- and security-related spending. The benchmark is simultaneously a political objective and a proxy for allied commitment. Yet comparing defense expenditure across countries is methodologically demanding because national accounting systems differ in their treatment of defense-related expenditure. This issue brief argues that headline expenditure figures are an incomplete measure of national defense effort. Drawing on official data, it examines how institutional differences shape international comparison. It focuses particularly on defense research and development, where the United States sustains an innovation ecosystem characterized by high technological ambition, tolerance of program failure, and correspondingly higher costs. The Zumwalt-class destroyer program illustrates both the opportunities and pitfalls of this approach. The issue brief demonstrates that meaningful comparison requires attention to the broader public-finance context. It concludes that assessments of allied burden sharing should distinguish between headline defense expenditure, defense-related public expenditure, and the military capability generated per unit of spending—referred to here as “bang for the buck.”

Introduction

Defense expenditure has long served as an indicator of political commitment within NATO. Since the Wales Summit September 2014, allies agreed to move towards spending at least 2 percent of GDP on defense and at least 20 percent of defense expenditure on major equipment including associated research and development, member states have increasingly been judged by quantitative spending targets rather than by qualitative measures of military capability. Following Russia’s full-scale invasion of Ukraine in February 2022, the political debate shifted towards a substantially more ambitious benchmark, formalized at The Hague in June 2025.

The Hague commitment is not a single figure but a two-tier construction. Allies agreed to allocate at least 3.5 percent of GDP to core defense requirements and to meeting NATO Capability Targets, assessed against NATO’s agreed definition of defense expenditure, and up to 1.5 percent of GDP to protecting critical infrastructure, defending networks, ensuring civil preparedness and resilience, unleashing innovation, and strengthening the defense industrial base. Allies submit annual plans showing a credible, incremental path; the trajectory and balance of spending will be reviewed in 2029, and the target date is 2035. Critically for the argument developed here, the second tier has no agreed accounting definition, which means that the headline 5 percent figure aggregates one measured quantity with one unmeasured one.

Although attractive because of its simplicity, percentage-of-GDP comparison obscures substantial differences in national accounting systems and in the allocation of defense-related costs across ministries and public budgets. Two countries reporting identical defense expenditure as a share of GDP may devote markedly different total public resources to maintaining military capability and may convert those resources into military output at markedly different rates.

The United States presents a particularly complex case. While the Department of Defense constitutes the core of American military spending, significant defense-related expenditure is financed through the Department of Veterans Affairs, the Department of Energy’s National Nuclear Security Administration (NNSA), the Department of Homeland Security, the intelligence agencies, and Treasury-funded military retirement obligations. Some of these fall outside NATO’s reporting framework; others are partially included under specific accounting rules. The U.S. federal budget itself distinguishes budget function 050 (national defense), which comprises subfunction 051 (Department of Defense), subfunction 053 (atomic energy defense activities) and subfunction 054 (other defense-related activities)—a structure that already demonstrates that “defense spending” is not a single administrative object.

European countries face a different institutional landscape: Universal healthcare systems, national pension schemes, and varying arrangements for military retirement means that expenditure associated with armed forces personnel is often financed outside defense ministries. Defense budgets may therefore understate the broader public cost of sustaining military establishments—although, as section 5 shows, NATO’s reporting rules correct for part of this, and the asymmetry is smaller than it is often assumed to be.

The objective of this issue brief is not to question NATO’s accounting methodology, but to identify how differences in public-sector organization affect the interpretation of defense spending: specifically, to what extent differences in public-sector accounting, military personnel systems, veterans’ benefits, healthcare financing, defense R&D, foreign assistance, and the organization of the state affect the comparability of NATO defense expenditure and burden-sharing metrics.

The analysis asks three research questions:

1. To what extent do institutional accounting differences affect comparisons between NATO defense expenditure?
2. How do differences in defense research and development influence military capability relative to expenditure?
3. What alternative indicators better capture the effectiveness, the “bang for the buck,” of defense spending?

The issue brief adopts a comparative public-finance perspective, integrating insights from defense economics, public administration, and security studies. It is a conceptual and documentary study: it triangulates published official accounting data rather than generating new estimates, and its claims about capability output are therefore illustrative rather than measured.