In June 2014, an Islamic State force of perhaps 1,500 fighters took Mosul, Iraq's second-largest city. On paper, the defenders included two Iraqi army divisions, part of a security force built over the preceding decade with roughly $25 billion in American assistance. Seven years later, the Afghan National Defense and Security Forces, built at a cost of around $88 billion, disintegrated in a matter of weeks.

These were not isolated misfortunes. The South Vietnamese army collapsed in 1975 after two decades of American support. Forces trained across the Sahel have repeatedly underperformed or overthrown the governments they were built to protect. Genuine successes exist, South Korea most prominently. But they are rare and slow, and they depend on conditions that most assistance programmes cannot replicate.

The persistence of this pattern points to something structural rather than incidental. Military assistance usually fails for reasons other than a shortage of courage among recipients or resources among donors. Three things combine to produce forces that look capable and are not: the design of aid, the politics of recipient states, and the incentives of the assessment system. The failure is built into the system.

Military assistance gravitates towards equipment because equipment is easy to see and count. A helicopter can be photographed at a handover ceremony and listed in a report to a legislature. It can be delivered in months rather than decades, and it frequently benefits the donor's own defence industry. Each of these features makes hardware politically attractive to the provider, whether or not it addresses the recipient's actual deficiencies.

Those deficiencies rarely lie in hardware. Combat effectiveness rests on a less visible substructure. It needs logistics systems that move fuel and ammunition to the right place, and maintenance regimes that keep vehicles running. It needs personnel systems that pay soldiers on time and promote officers on merit, and command arrangements that let information flow upward honestly. As Martin van Creveld showed across centuries of warfare, what a force can do is bounded by what it can sustain.

Donors compound the problem by building partners in their own image. The Afghan Air Force received modern aircraft whose upkeep depended heavily on foreign contractors. When those contractors departed in 2021, aircraft availability fell sharply. The close air support that Afghan ground units had been trained to rely on disappeared with it. The force had been designed around a support system it did not own and could not reproduce.

The result is a gap in capability that looks, from the outside, like capability. A partner force equipped to donor standards may perform adequately while foreign advisors, intelligence, and logistics remain in place. Remove that scaffolding and the force reverts to what its own institutions can support, which is often far less than its inventory suggests. Simpler equipment matched to local maintenance capacity would make for less impressive ceremonies. So would patient investment in pay, supply, and personnel systems. Both would produce more durable forces.

Takeaway

An army's real capability lies in the institutions that keep its inventory working, not in the inventory itself. Aid that delivers equipment without those institutions lends capability rather than building it.

Donors usually assume that the recipient government shares their objective: a capable army that can defeat an external enemy or insurgency. This assumption is frequently wrong. For many governments, the most immediate threat to survival comes from their own officer corps rather than from across the border or from insurgents in the hinterland. Building a competent, cohesive, professional army can mean creating the institution best placed to remove them.

Regimes respond with coup-proofing. They appoint commanders for loyalty rather than competence. They fragment command so that no single officer controls too much force, rotate capable leaders before they build personal followings, and restrict the flow of information between units. Each measure protects the ruler, and each also degrades the military's ability to fight. Caitlin Talmadge's work on authoritarian militaries shows how such practices can leave well-equipped forces brittle in battle.

Patronage adds a second layer. Command positions become rewards to hand out and revenue streams to exploit. Officers draw salaries for soldiers who do not exist; in late 2014 Iraq's own government acknowledged some 50,000 such ghost soldiers. Fuel, rations, and ammunition leak into private markets. For the officers involved, aid is a resource to be captured rather than a means of building an army.

Donors in theory hold leverage through conditions attached to aid, but those conditions rarely bite. The more strategically important a partner is, the less credible any threat to withdraw support becomes, and recipients understand this well. Stephen Biddle and colleagues have described the dynamic as a principal-agent problem: the donor pays, but the recipient pursues its own interests. Aid then reinforces the political arrangements that undermine effectiveness. Donors believe they are funding an army while they are in fact funding a patronage system.

Takeaway

Before asking whether a partner can build an effective army, ask whether its leaders want one. An army strong enough to win wars may also be strong enough to choose new rulers.

Assistance programmes must justify themselves to the governments that fund them, and justification requires measurement. The difficulty is that the easiest things to measure are inputs: soldiers trained, battalions formed, vehicles delivered, courses completed. These numbers rise steadily and reliably. Whether the resulting force will hold its ground under pressure is far harder to quantify, so that question tends to drop out of the reporting.

The incentives of the people producing assessments push in the same direction. Advisors typically rotate after nine to twelve months, and each rotation has reason to show progress during its tour. Senior commanders who designed the programme have staked their reputations on its success. Officials requesting further funding need evidence that previous funding worked. No individual needs to act dishonestly for the overall picture to become systematically optimistic.

Afghanistan illustrates the pattern. Over the course of the war, readiness rating systems were revised repeatedly, and from 2015 key assessments of Afghan forces were classified. The Special Inspector General for Afghanistan Reconstruction later concluded that officials had consistently struggled to measure the capability of the forces they were building, and had frequently overstated it. Partner units also self-reported much of the readiness data, and they had their own reasons to inflate it.

The only assessment that cannot be gamed is combat without foreign support, and programmes tend to avoid that test until withdrawal forces it. At that point the gap between reported and actual capability is revealed all at once. This is why collapses so often appear sudden to outside observers: the decline happened gradually, but its measurement did not. Honest evaluation would require independent assessors, metrics based on outcomes, and deliberate tests of partner units operating alone. Each of these risks producing bad news that nobody wants to deliver.

Takeaway

When everyone in a system benefits from reporting progress, optimism stops being a judgement and becomes a product of the system. Sudden collapses are often just delayed measurement.

Geoffrey Parker and Charles Tilly traced how European states were forged by the demands of war. Armies required revenue, revenue required taxation, and taxation required bureaucracies and bargains with society. The military and the state grew together, each sustaining the other.

Military aid inverts this sequence. It supplies an army without requiring the fiscal, administrative, and political institutions that would otherwise have to sustain it. The resulting force answers to its donor and its ruler more than to its society, and it depends on external resources it cannot replace.

Where assistance has worked, as in South Korea, it reinforced a state-building project already underway and was sustained over decades. The lesson is uncomfortable: armies cannot be imported, only grown.