Why Governments Always Underestimate the Cost of Wars

In August 1914, the German Chancellor Theobald von Bethmann Hollweg predicted that the war would be over by Christmas. His British counterpart Herbert Asquith thought the same. So did the French general staff, the Austrian foreign minister, and nearly every senior official in every belligerent government. The military planners were not alone: leading economists, including John Maynard Keynes, predicted that the financial strain of modern industrial war would make any conflict lasting more than a few months economically impossible. The war lasted four years and two months. It killed twenty million people. It consumed, in real terms adjusted for inflation, more than four trillion dollars of economic output. Every single prediction about its cost and duration was wrong in the same direction: it was longer, more expensive, and more destructive than anyone in authority had been willing to say in advance.

This directional consistency is not coincidence. The systematic underestimation of war costs is one of the most reliable patterns in political and economic history, and it is not primarily a story about intelligence failures or planning incompetence. The officials who predicted a short, cheap war in 1914 were not stupid men. They were responding to a set of structural incentives that consistently push governments toward optimistic forecasting when they decide to fight. Understanding those incentives explains why the pattern has persisted across radically different state types, military technologies, and historical contexts — and why it will almost certainly persist in the future.

The Commitment Problem in War Planning

Every government that chooses war faces a fundamental commitment problem: the decision to fight requires public support, and public support is easier to secure if the costs are minimized. A leader who tells the population that the coming war will last four years, kill a million of their countrymen, and consume half the national budget will face enormous political resistance. A leader who says the war will be short and the enemy will fold quickly generates enthusiasm and consent. The incentive to understate costs is built into the political structure of war-making in any system where leadership requires public legitimacy.

This is not a modern democratic problem. It is an ancient one. Thucydides’ account of the Athenian decision to invade Sicily in 415 BC is one of the most precise records of war cost optimism ever written. The proponents of the expedition — particularly the ambitious young general Alcibiades — systematically overstated the strategic benefits, understated the logistical requirements, and dismissed the objections of the more cautious general Nicias. The Assembly voted for the invasion on the basis of projections that turned out to be wrong in every significant respect. The expedition was destroyed. Athens never fully recovered. Thucydides tells us that the Assembly’s willingness to believe optimistic projections was itself a form of political corruption: the desire for the war’s supposed benefits overwhelmed the willingness to hear honest assessments of its costs.

The commitment problem interacts badly with the institutional structure of military planning. Military establishments have consistent incentives to request more resources than they need for any given operation, because budget requests are typically cut in the political process and because militaries are judged on their ability to achieve objectives rather than their fiscal efficiency. But military establishments also have incentives to minimize cost estimates when making the case for a specific operation, because minimizing costs makes the operation more likely to be approved. The result is a systematic pattern in which the military simultaneously over-requests budget resources in peacetime and under-predicts operational costs when war is being decided. Both behaviors serve institutional interests. Both distort decision-making.

The Duration Fallacy

The most consistent error in war cost forecasting is not underestimating the daily cost of fighting. It is underestimating the duration of conflict. A war that lasts twice as long as predicted will cost roughly twice as much, even if the planners correctly estimated the cost per month. And wars have almost universally lasted longer than predicted, for a structural reason that is obvious in retrospect but systematically ignored in advance.

Wars are symmetric interactions: you cannot know how long a war will last without knowing how long your opponent is willing to fight, and your opponent’s willingness to fight depends on how the war is going, which neither side can predict accurately before it begins. Any prediction of short duration is implicitly a prediction about the enemy’s psychology — a prediction that the enemy will recognize their situation as hopeless and sue for peace before committing their full resources. These predictions fail routinely because they reflect the predictor’s own assessment of the enemy’s position, not the enemy’s assessment of their own position.

The American planning for the Iraq War in 2003 is a textbook illustration. Deputy Defense Secretary Paul Wolfowitz famously dismissed General Eric Shinseki’s estimate that the occupation would require several hundred thousand troops, calling it “wildly off the mark” and predicting that Iraq’s oil revenues would cover reconstruction costs within two to three years. The actual cost of the Iraq and Afghanistan wars to the United States, calculated through 2021 by the Costs of War project at Brown University, exceeded eight trillion dollars when long-term veteran care and interest on war debt were included. Wolfowitz’s error was not a failure of intelligence analysis. It was a failure to account for the enemy’s agency — their willingness and ability to contest the occupation in ways that extended its duration and cost far beyond what any initial planning scenario had modeled.

The duration fallacy is compounded by what economists call the sunk cost effect. Once a war is underway and significant resources have been committed, the political cost of acknowledging that the original projections were wrong becomes extremely high. A government that admits its war is lasting longer and costing more than predicted faces accountability for having made optimistic promises that proved false. The rational political response is often to continue the operation — to “stay the course” — in the hope that eventual victory will retrospectively validate the decision to fight. This dynamic extends conflicts well past the point where rational cost-benefit analysis would recommend withdrawal, adding costs that the original war planners could not have anticipated because they did not anticipate fighting a domestic political battle over sunk costs simultaneously with the military campaign.

The Hidden Fiscal Architecture of War

The costs that governments present publicly when they seek authorization for war are almost always significantly lower than the true economic costs, even setting aside duration errors. This is not primarily because governments lie, though some do. It is because the true costs of war are distributed across many budget categories, many time periods, and many institutional actors in ways that make comprehensive accounting politically inconvenient and administratively difficult.

The most significant hidden cost is future veterans’ care. The United States entered World War I with a veterans’ population of roughly half a million. The war added 4.7 million veterans, many of them with injuries — physical and psychological — that would require medical care for decades. The peak year of World War I veterans’ care costs in the United States was 1969: more than fifty years after the armistice. The cost of providing care to veterans of the Iraq and Afghanistan wars will peak, according to current projections, sometime in the 2040s. No appropriations process at the time of the original war vote captures these costs, because they will be borne by governments and taxpayers who had no voice in the decision to fight.

War debt creates a similar temporal displacement of costs. Every major war in modern history has been substantially debt-financed, which means that the full fiscal cost of the war is spread over the decades required to service and eventually repay the debt. Britain was still making payments on World War I debt in 2015 — ninety-seven years after the war ended. The nominal cost of a war, presented at the time of the decision to fight, is therefore a systematically incomplete representation of the true fiscal cost, because debt finance obscures the total sum behind a series of future payments that will be made by future governments.

The third hidden cost category is economic disruption — the output foregone because of wartime resource mobilization, labor conscription, capital destruction, and trade disruption. These costs are real but they do not appear on any government ledger. They are the factories that were not built, the trade that did not happen, the human capital formation that was interrupted. The economic historian Mark Harrison has estimated that the total economic cost of World War II — including output destroyed and foregone as well as direct military expenditure — was approximately four times the direct fiscal cost recorded in government accounts. The gap between fiscal cost and economic cost is the gap between what governments present to the public and what the conflict actually extracted from the society.

Why Incentives Beat Institutions

One might expect that the consistent historical pattern of war cost underestimation would have produced institutional corrections: independent cost estimates, mandatory long-term fiscal projections, legislative requirements for comprehensive cost accounting before war authorizations. In some democracies, these reforms have been attempted. They have been consistently ineffective.

The reason they fail is that the political incentives that drive optimistic forecasting are stronger than the institutional mechanisms designed to counteract them. An independent budget office can produce a rigorous cost estimate; a determined executive can dismiss it, ignore it, or simply decline to seek the legislative authority that would require its production. A legislative requirement for long-term cost accounting can be waived in an emergency; wars are routinely framed as emergencies in ways that justify bypassing normal oversight processes. And most fundamentally: voters who want to believe that a war will be cheap and short will punish officials who insist otherwise and reward officials who project confidence.

The Austrian economist Ludwig von Mises observed in the 1920s that governments resort to inflationary finance during wars precisely because taxation makes costs visible while inflation disperses them invisibly across the entire economy. This is a specific instance of a general principle: the political system consistently selects for mechanisms that obscure war costs because visible costs generate political resistance and invisible costs do not. The consequence is that the true cost of military decisions is systematically hidden from the democratic accountability mechanisms that are supposed to constrain those decisions.

This does not mean wars are never worth fighting. Some wars — clearly defensive, clearly necessary — are worth fighting at whatever cost they impose. The question is whether the decision to fight can be made rationally if the cost information provided to decision-makers and the public is systematically biased in one direction. The evidence of history suggests that it cannot: that the structural biases toward optimism generate a consistent pattern of under-investment in diplomacy, under-preparation for long conflicts, and over-extension in pursuit of objectives that were set when the war was expected to be short.

Toward Honest War Accounting

The implication of the historical pattern is not fatalism. It is a demand for different institutional design. If war decisions are consistently distorted by optimistic cost projections, the institutional response should make those projections more honest and more comprehensive — not by expecting individuals to overcome their incentives, but by changing the incentive structure.

Independent war cost commissions with genuine authority and a mandate to project total lifecycle costs — veterans’ care, debt service, opportunity cost — would not eliminate optimistic forecasting, but they would create a public record against which actual costs could be measured. Requiring long-term fiscal impact statements for war authorization, similar to environmental impact statements for major infrastructure projects, would force a conversation about costs that is currently avoided. Mandatory debt accounting that tracks war expenditures separately from general government spending would prevent the fiscal obscuring that allows war costs to be absorbed invisibly into long-term government liabilities.

None of these reforms would eliminate the commitment problem or the duration fallacy. They would make the systematic underestimation of war costs harder to sustain without explicit denial of evidence. That is not nothing. The governments that led their populations into the catastrophe of 1914 on the basis of promises that proved catastrophically wrong were not acting in a vacuum. They were responding to institutional structures that rewarded optimism and penalized honesty about costs. Changing those structures will not guarantee that future wars are avoided. But it might ensure that when they are chosen, they are chosen with open eyes.