How the IMF underplays its own assessment of the real cost of war

The International Monetary Fund’s April 2026 World Economic Outlook arrives with an unusually candid subtitle – Global Economy in the Shadow of War – in the same month that NATO has locked in its most ambitious military spending target in history.

The opening analytical chapter notes explicitly that NATO members committed in June 2025 to raise annual defence and security-related spending to 5% of GDP by 2035, “more than double the previous 2 percent guideline”.

Two substantial chapters of empirical research follow. They document, carefully and at length, what that kind of spending surge does to economies. Then, with equal care, they decline to say whether it is wise.

Military spending

Chapter 2 of the IMF report draws on data from 164 countries since 1946 to study what happens when governments rapidly scale up military spending. A typical defence spending boom, it finds, raises outlays by about 2.7 percentage points of GDP over two and a half years.

This is, as the IMF notes, “broadly similar” to what NATO members now need to do to reach their 5% target.

Roughly two-thirds of such buildups are financed by borrowing. Public debt rises about 7 percentage points within three years. When the buildup is paid for by cutting other government programmes rather than borrowing, what the IMF calls “the guns versus butter trade-off emerges” – spending on social protection, health, and education falls sharply.

In wartime, the debt costs nearly double: a 14 percentage point jump, with social spending cut regardless of how the buildup is financed.

Responding to the IMF findings, Poland’s finance minister admitted that his country’s 5% defence target “is a lot” – while justifying it by invoking Russia’s threat to neighbouring states.

That exchange captures the central tension the IMF report raises but cannot resolve: the economics of military spending and the security calculus are treated as separate rooms, and the door between them is locked.

The IMF report says that a military spending buildup that prioritises domestic production and research could “support long-term productivity growth”.

But the return on military spending depends heavily on where the money actually goes. When a country buys most of its weapons from foreign suppliers, as most developing countries do, the spending flows out of the domestic economy rather than generating local jobs, wages or industrial capacity. The money leaves as imports.

It is left for the reader to draw the implication of the IMF suggestion that who gets the defence contracts matters enormously to the economic outcome.

Chapter 3 of the April 2026 World Economic Outlook report completes the picture from the other side. Drawing on global conflict data since 1945, it finds that actual wars generate output losses exceeding those from financial crises or natural disasters – roughly 7% of GDP within five years of the onset of conflict, with scars persisting a decade later.

In 2024 alone, more than 35 countries experienced active conflict, roughly half classified as fragile and conflict-affected states.

But recovery, the IMF says, is “slow and uneven”, critically dependent on whether peace, once achieved, holds. Where it does not, “output fails to recover”. Notably, the chapter acknowledges that recovery outcomes depend heavily on political factors, including power-sharing arrangements and security guarantees.

But these possibilities, the IMF says, “lie beyond the scope of the chapter’s analysis” – a caveat in a bracket.

A dataset that changes shape

There is a methodological move in Chapter 2 that deserves attention. The chapter opens with a dataset of 164 countries – establishing that defence spending booms are now most frequent in emerging markets and developing economies. But the dataset narrows when the analysis turns to its policy-relevant core – what a well-designed military buildup looks like, how to maximise economic returns, what governments should do.

The detailed breakdown of what defence budgets are actually spent on covers only 35 NATO and EU member countries. The model simulation of a large defence buildup is explicitly “calibrated to the European experience”, with the assertion that results “can be generalised to other contexts”.

Imagine a study of road accidents across every city in India that then uses only data from Mumbai’s expressways to recommend national speed limits – and presents the conclusions as universally applicable.

The world’s most conflict-affected regions – West Asia and sub-Saharan Africa, which Chapter 3 identifies as bearing a disproportionate share of active global conflicts – provide the data that establishes how serious things are. These countries then exit the analysis at the point where conclusions about what to do are drawn. The non-Western world frames the problem while the Western world supplies the solution.

The IMF’s largest shareholder is the United States, which is also NATO’s founding military guarantor. The normative lens and the geopolitical architecture it reflects are, in this instance, the same thing.

The footnote that belongs in the text

Chapter 2 contains an important acknowledgment which is instead a footnote. After noting that higher defence spending could serve as a deterrent to conflict, the chapter immediately cites research showing that “deterrence effects are modest”.

Military buildups, it acknowledges, lead to only “a small and persistent decline in conflict over the long term, with no effect on short-term conflict risks”.

In other words, the core security justification for spending more is empirically weak, by the IMF’s own account. But this merits only a footnote while the chapter’s main text proceeds as if the security rationale is settled.

What a graph says without saying it

NATO was founded in 1949 as a US-led military alliance. Since the Cold War ended in 1991, its expansion eastward has been the defining geopolitical fault line in Europe’s relationship with Russia.

Today, European governments are simultaneously pursuing ceasefire negotiations to end the Russia-Ukraine war and locking in military spending commitments that would push their defence budgets to levels not seen since the Cold War.

A figure in Chapter 2 of the IMF report plots world average defence spending since 1946 against the NATO 2035 target of 5% of GDP. The gap between them is visually stark: the global GDP-weighted average has hovered between 2% and 3% for decades, peaking during the Cold War and declining sharply after 1991.

The NATO target would take member states back above those Cold War peaks – while, simultaneously, a fragile peace process in Ukraine is underway.

The IMF report does not comment on this juxtaposition. Its authors, if asked whether the 5% target is consistent with the chapter’s own findings, can only say that spending targets are set by security risk perceptions and that assessing those perceptions is “beyond the scope of the chapter’s analysis”.

Convenient choices, what counts

Consider what the IMF has actually established across these two chapters: that large buildups are fiscally dangerous and socially costly; that wars cause lasting economic damage disproportionately borne by the world's poorest and most fragile states; that the deterrence argument for heavy military spending is, by the chapter's own citation, empirically modest; and that the analytical framework used to draw policy conclusions is calibrated to Western European experience and presented as universal – while the countries bearing the greatest conflict burden are used to frame the problem, then exit the analysis.

That last point has a human face the report does not show. The dataset begins in 1946. The Partition of South Asia – the price at which freedom from colonial rule was attained in 1947, and which killed up to a million people in communal massacres and displacement alone – is invisible to that definition. So is much of the violence that has actually ended the lives of millions across Asia, Africa and Latin America in the eight decades since. What appears instead are output loss percentages.

The report’s Executive Summary notes its own findings – and then recommends fiscal prudence and structural reform. The distance between the data and the recommendation is a choice about whose experience of war counts as evidence, as context and who gets to set the norms to assess and decide.

That choice now has a very specific weight. The same World Economic Outlook that brackets geopolitics as beyond its analytical scope, models three scenarios for what happens to the global economy if the West Asia war spreads – including a scenario that the IMF associates with global recession.

The war it is modelling is being fought with US military support against a country whose defeat has been a stated US foreign policy objective for decades. The NATO spending target the IMF declines to evaluate was set in the same institutional moment. The dataset it uses to model optimal defence spending draws on the military experience of the NATO alliance whose expansion helped produce the conditions for the conflicts now driving the outlook dark.

A 17-year-old reading this in Delhi or Lahore or Nairobi is not wrong to feel that the world is poised at something dangerous – and that the institutions tasked with providing the global economic outlook are describing the fire while carefully avoiding the question of who lit it and why.

Veena Naregal is professor of sociology at the Institute of Economic Growth, Delhi.