Defense and foreign affairs, the evidence

The evidence on defense and foreign affairs in the United States is strongest on spending and cost and weakest on results. Public sources agree that the United States spent about $954 billion on its military in 2025 by the SIPRI measure, that the Department of Defense has failed to obtain an audit opinion eight years running, and that major weapons programs have grown more expensive; they cannot show directly how many conflicts deterrence prevented or how much influence aid and alliances buy. This page sets out what is measured, where the measures conflict, and what is unknown; see the subject overview and who decides.

In plain words

It is easy to count how much money goes into defense. It is hard to count what it buys.

Even counting the money is tricky. “Defense spending” can mean the Pentagon budget, all national security spending, or an international measure used to compare countries. These give different numbers.

The cost of wars depends on what you include. Do you count only the money spent while fighting, or also the care veterans will need for decades? One estimate is about $1.6 trillion; another is about $8 trillion. Both are real numbers. They measure different things.

The hardest question, whether a policy kept the country safer, often has no clean answer, because nobody can see the war that did not happen.

Defined terms

SIPRI military expenditure. The Stockholm International Peace Research Institute’s internationally comparable measure of military spending, republished by the World Bank.

Military burden. Military expenditure as a share of gross domestic product.

Disclaimer of opinion. An auditor’s statement that it could not obtain enough reliable evidence to give any opinion on financial statements.

Sustainment cost. The cost of operating and maintaining a weapon system over its service life, as distinct from buying it.

Overseas contingency operations (OCO). A budget category formerly used for war-related appropriations.

Burden sharing. The distribution of defense costs among allies.

What the data show: spending

Defense and foreign affairs spending can be measured several ways, and each answers a different question.

International comparison. SIPRI figures republished by the World Bank put U.S. military expenditure at about $954 billion in 2025, 3.1 percent of GDP, down from 3.4 percent in 2024. The world total was about $2.86 trillion, a world average burden of 2.5 percent. China spent about $336 billion (1.7 percent of GDP) and Russia about $190 billion (7.5 percent of GDP).⁠¹ By arithmetic on those figures, the United States accounted for about a third of world military spending.

The budget request. The Congressional Research Service reports that the Defense Department’s fiscal 2027 request totals $1.45 trillion: about $1.1 trillion discretionary and $350 billion in mandatory funding that assumes Congress will pass a reconciliation bill, $440.9 billion (44 percent) above the fiscal 2026 enacted level of $1.01 trillion. The request includes $413.1 billion for procurement and $343.7 billion for research, development, test, and evaluation.⁠²

Share of the federal budget. Treasury’s public guide states that mandatory spending “represents nearly two-thirds of annual federal spending” and that Congress allocates “over half of the discretionary budget towards national defense”.⁠³ Defense is therefore the largest item in the part of the budget Congress votes on each year, but a minority of total federal spending.

Allies. NATO reports that European Allies and Canada together spent 2.3 percent of GDP on defense in 2025, up from 1.4 percent in 2014, with a real increase of nearly 20 percent from 2024 to 2025.⁠⁴

What the data show: costs and management

Defense and foreign affairs management data come mostly from GAO and the Defense Department’s own audits.

Audits. The Congressional Research Service reports that the fiscal 2025 audit, released December 18, 2025, was the department’s eighth annual financial statement audit and its eighth consecutive disclaimer of opinion; entities receiving disclaimers accounted for at least 64 percent of the department’s total budgetary resources. Section 1005 of the fiscal 2024 NDAA (P.L. 118-31) requires a clean opinion by December 31, 2028.⁠⁵

High-risk areas. GAO’s 2025 High-Risk List includes five Defense Department areas: business transformation, business systems modernization, contract management, financial management, and weapon systems acquisition.⁠⁶

Weapons programs. GAO reported in April 2024 that F-35 “sustainment cost estimates have increased 44 percent, from $1.1 trillion in 2018 to $1.58 trillion in 2023,” while the Air Force, Navy, and Marine Corps “project they will fly the F-35 less than originally estimated on an annual basis”.⁠⁷

What the data show: foreign aid and sanctions

Defense and foreign affairs tools beyond the military are smaller in dollars and harder to evaluate. The Congressional Research Service estimated U.S. foreign assistance at $48.18 billion in fiscal 2019 under a broad definition, about 1 percent of total federal budget authority.⁠⁸ On sanctions, GAO found in 2019 that the Treasury, State, and Commerce departments assess the impacts of specific sanctions on their targets but have not assessed sanctions’ overall effectiveness and are not required to; GAO’s review of academic studies found sanctions more effective when implemented through an international organization and when the target had existing dependency on or a relationship with the United States.⁠⁹

Where the evidence is disputed

What did the post-9/11 wars cost? The answer depends on scope. A Defense Department estimate cited in the Costs of War project put emergency and OCO spending for Iraq, Syria, and Afghanistan at $1.596 trillion through fiscal 2020, excluding non-Defense classified programs. Neta Crawford of the Costs of War project estimated about $5.8 trillion spent and requested through fiscal 2022, counting war zone spending, related base budget increases, homeland security, and interest on borrowing, and about $8 trillion including future veterans’ medical and disability costs, which she noted “will likely exceed $2.2 trillion” over coming decades based on Linda Bilmes’s research.⁠¹⁰ Critics of the larger figure argue that some homeland security and base budget spending would have occurred anyway; defenders argue that narrow figures hide costs taxpayers will still pay. Both figures are documented; they answer different questions.

Is U.S. spending high or low? In dollars, the United States spends nearly three times China’s figure; as a share of GDP, Russia’s burden is more than twice the U.S. burden.⁠¹ Advocates of higher spending also note that dollar comparisons at market exchange rates may understate what other countries’ spending buys, while advocates of lower spending note that U.S. allies add substantially to collective capacity. Which comparison is right depends on what the spending is for.

How will burden sharing be counted? NATO’s June 2025 commitment sets a 5 percent of GDP target for 2035, of which at least 3.5 percent is core defense under NATO’s agreed definition and up to 1.5 percent covers related spending such as protecting critical infrastructure, civil preparedness, and the defense industrial base.⁠¹³ Interpretation: because the new target counts categories that older defense-only figures did not, comparisons between allies’ pre-2025 burdens and their future reported totals will need care, and claims that a given ally “met the target” will depend on which half of the target is meant.

Is management failure a reason to cut or to reform? The audit record and cost growth are documented.⁠^5,7 Some read them as evidence that more money will be wasted; others argue that audit failures reflect old accounting systems more than lost funds, and that cutting budgets would harm readiness without fixing accounting. The audit results themselves cannot settle which reading is right, because a disclaimer means the evidence was insufficient, not that money was proven missing.

What do Americans want? Pew’s April 2024 survey shows broad agreement on preventing terrorist attacks (73 percent top priority) and narrower majorities for maintaining military advantage (53 percent),⁠¹¹ while NATO views split by party: 75 percent of Democrats and 42 percent of Republicans favorable in March 2026.⁠¹² The data show stable consensus on goals and partisan division on commitments.

What is unknown

Defense and foreign affairs outcomes are hard to know from public data.

• Deterrence success. There is no public measurement of attacks or wars that did not happen because of U.S. posture.
• Classified spending and capability. Some spending and most operational capability are not public, so outside analysts work with partial information.
• Long-run effects of aid and alliances. Whether aid and alliances buy durable influence is debated, and GAO found that even the agencies administering sanctions do not assess overall effectiveness.⁠⁹
• Future costs. Sustainment, veterans’ care, and interest on borrowing extend decades beyond decisions made now.

Evidence

Defense and foreign affairs evidence on this page comes from SIPRI data republished by the World Bank, Congressional Research Service reports, Treasury, NATO, GAO, a Costs of War project paper, and Pew surveys. Each figure carries its year and definition; mixing definitions is the most common source of error in this subject.

Competing interpretations

Defense and foreign affairs evidence supports at least four readings.

• Underinvestment. Rival spending, especially Russia’s burden and China’s growth, shows the need for more capacity, and the fiscal 2027 request responds to that.
• Overextension. War costs, cost growth, and failed audits show a system that spends more than it can manage and commits more than the public supports.
• Reform without cuts. Spending levels are a strategic choice; the evidence most clearly supports better accounting, acquisition, and evaluation regardless of level.
• Limited evidence. Public data cannot measure deterrence, so strong claims in any direction rest on judgment more than measurement.

Application to this framework

Defense and foreign affairs evidence shows why My Vote Forward insists that every figure carry a definition and a date, and why tradeoffs are presented separately from priorities. If participants were asked whether defense spending is “too high,” the answer would depend on whether they were shown the dollar figure, the GDP share, or the share of discretionary spending. The framework’s question design rules require naming the measure and baseline in any such question, and its published releases rules require stating that participants are not a representative sample.

Unresolved questions

• What public measure of defense outcomes, as opposed to inputs, could be published without compromising security?
• Should war cost estimates include future veterans’ care and interest by default, and who should produce an official figure?
• How should a participation tool present spending comparisons when dollar figures and GDP shares point in different directions?