Four years have now passed since the final American military forces left Afghanistan on 30 August 2021, completing a withdrawal that ended twenty years of US-led military involvement and allowed the Taliban to consolidate control over the country. The images from Kabul airport remain some of the defining images of the post-9/11 era: desperate civilians attempting to leave, an American evacuation conducted under extraordinary pressure and a Taliban movement returning to the capital twenty years after the United States had removed it. Four years later, the predictions made during those chaotic weeks can be evaluated more clearly. Afghanistan has not returned to the scale of nationwide warfare that characterised much of the previous two decades, and the Taliban has established a stronger monopoly over national territory than the previous republic often possessed. At the same time, Afghan women and girls face one of the world’s most restrictive systems of state-imposed gender discrimination, with girls generally prohibited from secondary and university education and women excluded from large parts of employment and public life. Afghanistan therefore presents an uncomfortable political reality: a government can increase one form of state capacity, particularly control over violence, while simultaneously destroying freedoms and human capital that determine whether the state can develop over the long term.
The distinction between order and good governance is essential because they are often treated as though they are the same thing. The Taliban inherited a country that had experienced decades of insurgency, foreign intervention, militia warfare and terrorist violence. The previous Afghan republic possessed internationally recognised institutions, elections and constitutional rights, but its authority was uneven geographically and depended enormously on foreign military and financial support. The Taliban now controls most territory through a much more centralised coercive structure. Checkpoints between provinces have declined compared with wartime conditions, Taliban insurgents are no longer fighting the government because they are the government, and ordinary commercial movement in some areas has consequently become safer. This is a real change. A farmer who can transport produce without crossing an active front line experiences a genuine economic benefit. A merchant who no longer expects a road to close because of fighting can trade more predictably. Security is an economic input even when the government providing it is authoritarian.
This creates a problem for simplistic theories connecting democracy automatically with state strength. The Afghan republic formally possessed democratic institutions while often struggling to enforce rules outside major cities. The Taliban possesses almost no meaningful electoral legitimacy while exercising much stronger territorial control. This does not demonstrate that authoritarianism is superior. It demonstrates that state capacity and regime type are different variables. A democracy can be weak. A dictatorship can be administratively strong. The normative argument for democracy concerns political rights, accountability and the peaceful replacement of governments; the capacity argument concerns whether institutions can implement decisions. The best political systems need both legitimacy and capacity. Afghanistan currently has much more of the second in selected coercive areas than the first.
The economic benefits of reduced conflict, however, are constrained severely by the Taliban’s own policies. The exclusion of girls from secondary schools and universities is not simply a human-rights issue, although that would be sufficient reason to oppose it. It is also a massive negative human-capital shock. Education increases productivity partly by developing literacy, numeracy, technical knowledge and professional skills. Preventing roughly half the population from accumulating advanced education therefore reduces the future supply of doctors, teachers, engineers, administrators and entrepreneurs. The cost compounds because today’s schoolgirl becomes tomorrow’s missing professional. A policy imposed in 2022 can therefore reduce national output in 2042 even if the rule is repealed long before then.
Human-capital losses behave somewhat like lost compound interest. Suppose education raises an individual’s lifetime productivity substantially and the economy loses several years of schooling across millions of girls. The direct cost is not simply the school expenditure that did not occur. It is the discounted value of decades of future output that education would have generated. Add lower labour-force participation and the economic loss becomes much larger. Afghanistan is one of the poorest countries in the world and therefore has exceptionally little productive capacity to waste. Excluding women from education and employment is effectively choosing to operate the national economy with a large portion of its potential workforce deliberately constrained.
The healthcare consequences demonstrate how ideological restrictions can collide with their own social objectives. Conservative rules limiting interaction between unrelated men and women increase the importance of female doctors and nurses because many Afghan women need female healthcare providers. But preventing girls from completing secondary school and university eventually reduces the number of women capable of becoming doctors. The policy therefore produces an internal contradiction: restrictions justified partly through gender segregation destroy the female professional workforce required to make segregation compatible with functioning healthcare. Ideological systems often encounter this problem when purity rules interact with administrative reality.
The same applies to education itself. If women cannot train as teachers, maintaining separate girls’ education becomes increasingly difficult even if the Taliban eventually permits some forms of it. Every year of exclusion therefore damages the institutional capacity required for later reversal. Policy creates path dependence. Reopening a school is easy physically. Replacing a generation of missing teachers is not.
This is why international pressure has focused so heavily on women’s rights as a condition for deeper diplomatic recognition. The Taliban wants international legitimacy, access to frozen assets, economic relationships and reduced sanctions. Western governments want leverage over Taliban behaviour. Recognition therefore becomes a bargaining asset. But four years of pressure reveal the limitations of conditionality when the target values ideology more than the offered economic benefit. If Taliban leadership places extremely high value on its interpretation of social order, sanctions relief may not be sufficient to change policy. External governments often assume actors maximise economic welfare similarly to themselves. They do not.
This is the same utility-function problem encountered with the Houthis and Iran. Economic punishment works best when the target government places high political value on economic performance. A regime capable of tolerating substantial poverty while preserving ideological control can absorb sanctions differently. That does not make sanctions useless. It changes the expected elasticity of behaviour.
The humanitarian cost then creates another dilemma. Sanctions intended to pressure the Taliban can reduce economic activity affecting ordinary Afghans who possess almost no influence over Taliban policy. International aid can alleviate suffering but may indirectly free Taliban resources for other priorities. Cutting aid punishes civilians. Continuing aid can stabilise a regime donors oppose. This is the humanitarian substitution problem: when outsiders finance food, healthcare or basic services, the governing authority can devote more of its own limited resources elsewhere.
The answer cannot be abandoning humanitarian assistance. Millions of Afghans should not be used as instruments for coercing their rulers. Assistance instead needs to be designed to minimise regime capture while maintaining essential services. Direct support through international organisations, local healthcare networks and carefully monitored programmes can separate humanitarian objectives from political recognition as much as possible. Perfect separation is impossible.
Afghanistan’s fiscal position makes this especially difficult because the previous republic depended heavily on foreign assistance. Before 2021, grants financed an enormous portion of public expenditure. When foreign support disappeared suddenly after the Taliban takeover, the economy experienced a massive contraction. That collapse demonstrated a central weakness of the previous state: external financing had created institutions whose expenditure commitments exceeded the domestic economy’s ability to sustain them. A government can look administratively substantial while foreign donors finance much of the payroll.
This is a broader lesson for development policy. Aid can build schools, ministries and security forces rapidly. Sustainability requires eventually creating enough domestic tax capacity and economic growth to finance them. Otherwise institutions become contingent on donor politics. When the external patron leaves, the state experiences a fiscal cliff.
The Afghan security forces demonstrated an even more extreme version. The United States and allies spent enormous sums training and equipping Afghan forces, yet many units collapsed rapidly as Western support disappeared and the Taliban advanced in 2021. The problem was not simply cowardice or training. Logistics, air support, maintenance, salaries, political confidence and leadership depended heavily on the external system. Imported capacity is not necessarily institutionalised capacity.
This should remain one of the central lessons of the entire Afghanistan intervention. The United States could purchase enormous quantities of military equipment. It could not purchase indefinitely the political institutions required to make Afghans believe the republic would survive without American backing.
This distinction between resources and legitimacy explains why the Taliban could defeat a government possessing vastly more expensive military equipment. The Taliban’s fighters believed their organisation would continue. Afghan government soldiers increasingly believed theirs might not. Expectations therefore became self-fulfilling. Once enough officials and commanders concluded collapse was likely, fighting became personally irrational. Units surrendered, local deals multiplied and collapse accelerated.
The final weeks of the Afghan republic therefore resembled a bank run. A bank can possess substantial assets and still collapse if everyone expects others to withdraw first. A government can possess soldiers and ministries and still collapse when everyone expects others to defect first. Political orders contain confidence effects.
The Taliban now benefits from the opposite expectation. Most domestic actors believe it will remain in power, reducing incentives to challenge it openly. This does not mean the regime is popular. It means opposition coordination is difficult when each potential opponent expects others to remain passive. Authoritarian durability often rests less on majority support than on collective-action failure among opponents.
The absence of a viable organised alternative reinforces this. The anti-Taliban resistance is fragmented and militarily weak. Former republic elites are divided and many live abroad. Foreign governments have little appetite for restarting a proxy war. The Taliban therefore faces limited immediate threat to national control despite severe international criticism.
This demonstrates why regime change is much easier to demand than to produce. A government can be deeply repressive while remaining the only actor capable of administering the state. Removing it without a credible replacement can recreate civil war. External policy therefore needs to distinguish opposition to Taliban policies from pretending the Taliban does not govern Afghanistan.
Diplomatic engagement is consequently unavoidable. Governments can speak to Taliban officials without recognising the regime formally or endorsing its social system. Engagement allows negotiation over humanitarian access, terrorism, migration and detained citizens. Refusing contact would reduce leverage rather than increase it.
This is the same distinction between engagement and legitimacy that appeared with HTS in Syria. Diplomacy is communication with actors possessing power, not a moral award for actors one approves of.
The terrorism question remains particularly important because the original American intervention began after al-Qaeda used Afghanistan as a base before the 11 September attacks. The Taliban pledged under the 2020 Doha agreement not to allow Afghan territory to threaten the United States and its allies. Yet al-Qaeda leader Ayman al-Zawahiri was killed by an American drone strike in Kabul in 2022 while staying at a house linked to a senior Taliban figure, raising obvious questions about Taliban relationships with transnational jihadists.
At the same time, the Taliban itself fights Islamic State Khorasan Province (ISIS-K), which regards the Taliban as insufficiently radical and has carried out major attacks inside Afghanistan and abroad. This creates a strange alignment in which Western governments and the Taliban share an interest in suppressing ISIS-K while remaining adversaries on almost every other political question.
International politics repeatedly produces these issue-specific alignments. Cooperation does not require friendship. The United States worked with Stalin against Nazi Germany while opposing Soviet communism. Governments can exchange intelligence indirectly against a terrorist organisation without recognising one another’s broader political legitimacy.
The Taliban’s counterterrorism performance therefore should be evaluated empirically. Does Afghan territory become a platform for international attacks? Are ISIS-K capabilities declining or expanding? Are al-Qaeda networks receiving sanctuary? These outcomes matter more than rhetorical assurances.
Regional states have even stronger incentives for pragmatic engagement because geography prevents disengagement. Pakistan, Iran, China, Russia and the Central Asian republics cannot simply decide Afghanistan is somebody else’s problem. Refugees, narcotics, terrorism and trade cross borders. Each therefore maintains some form of relationship with the Taliban despite varying levels of distrust.
Pakistan’s relationship is particularly ironic. Islamabad was long accused of providing sanctuary or support to Taliban networks during the American war, partly because Pakistani strategists viewed a friendly government in Kabul as useful against Indian influence. Yet Taliban-controlled Afghanistan has not produced the strategic stability Pakistan expected. The Tehrik-i-Taliban Pakistan (TTP) has intensified attacks inside Pakistan, and Islamabad accuses the Afghan Taliban of allowing TTP militants sanctuary. The Afghan Taliban denies these allegations or disputes their extent.
This is a classic proxy-control problem. States often support ideologically related armed groups assuming influence will continue after those groups gain power. But proxies possess their own objectives. The Taliban is Afghan nationalist as well as Islamist and does not automatically subordinate its interests to Pakistan. Islamabad therefore helped create an outcome it believed would increase strategic depth and now faces greater militant insecurity along the same frontier.
This is another example of the danger of evaluating foreign policy through first-order effects only. Supporting the Taliban weakened an Afghan government aligned closely with Western powers and India. That was the immediate Pakistani objective. The second-order effect was strengthening a militant ideological ecosystem from which anti-Pakistan groups also benefited.
China approaches Afghanistan differently. Beijing wants stability, mineral access and assurance that militant groups do not threaten Xinjiang or Chinese projects. Afghanistan possesses potentially significant deposits of copper, lithium and other minerals, creating repeated speculation that resource development could transform the economy.
Natural resources, however, do not become wealth automatically. Deposits underground possess limited economic value without roads, electricity, legal certainty, skilled workers and security. This is the resource-capacity complementarity problem. A country can possess billions of dollars of theoretical mineral wealth while remaining extremely poor because the institutional cost of extracting it is too high.
Afghanistan therefore is unlikely to become prosperous simply because Chinese companies sign mining agreements. Investors need confidence contracts will survive, infrastructure needs financing and export routes need stability. Mining can generate government revenue. It cannot substitute for broad human-capital development.
The exclusion of women consequently undermines even the Taliban’s potential resource strategy. Modern mining requires engineers, accountants, healthcare systems, education and administrative capacity. Reducing the educated workforce reduces the productivity of every other asset.
This is why human capital is the multiplier on natural capital.
Oil beneath Norway and oil beneath a failed state are physically similar resources.
Institutions determine the economic outcome.
Afghanistan’s opium economy creates another complication. The Taliban imposed a sweeping ban on poppy cultivation after taking power, and cultivation subsequently fell dramatically. This is one area where centralised coercive capacity produced a measurable policy outcome that previous governments struggled to achieve. But eliminating poppy does not automatically create alternative livelihoods. Farmers cultivated opium partly because it was profitable, drought-resistant relative to some alternatives and supported by established credit and trafficking networks.
A ban therefore destroys an income source before necessarily creating another.
This is successful enforcement without successful substitution.
If rural households become poorer enough, illicit cultivation can eventually return or move geographically. Sustainable drug policy therefore requires alternative crops, market access, irrigation and credit alongside enforcement.
Again, state capacity needs more than coercion.
It needs the ability to create viable alternatives to prohibited behaviour.
Migration reveals the human consequences of Afghanistan’s stagnation. Millions of Afghans remain refugees or migrants in neighbouring Pakistan and Iran, while both countries have increasingly pushed Afghans to return. Large-scale returns place additional pressure on an economy already unable to generate enough employment.
A refugee returning to Afghanistan needs housing, work, healthcare and schools.
For a family with daughters, the education restrictions create an additional reason not to return voluntarily.
This demonstrates how rights policy becomes migration policy. The Taliban may want citizens to return and foreign governments may want Afghan refugees to leave. Families make their own calculations according to future opportunity.
If a sixteen-year-old girl cannot attend school in Afghanistan but can attend school abroad, the economic value of remaining abroad becomes enormous even if living conditions there are difficult.
The Taliban therefore cannot maximise both ideological restrictions and voluntary diaspora return.
There is a trade-off.
International recognition presents the same problem. The Taliban wants diplomatic legitimacy without accepting many of the conditions major Western governments attach to it. Western states want policy change without providing recognition first.
Both sides therefore wait.
This is a bargaining deadlock caused by sequencing.
The Taliban says recognition and sanctions relief would create conditions for normal government.
Western states say policy reform must precede recognition.
A possible solution is incremental reciprocity rather than one grand bargain. Permit particular categories of girls’ education; receive targeted economic relief. Expand women’s employment in healthcare and humanitarian sectors; receive additional banking access. Improve counterterrorism verification; receive diplomatic engagement.
Small exchanges require less trust.
The difficulty is whether Taliban leadership actually wants the offered economic benefits enough to compromise ideologically.
Four years of evidence suggests the answer may often be no.
That should force international policy towards realism. External governments cannot transform Afghan society simply through sanctions and statements. They can reduce suffering, preserve educational opportunities where possible, maintain pressure around rights and prevent Afghanistan becoming a major international terrorist sanctuary.
Those are narrower objectives than the nation-building ambitions of the 2000s.
They may also be more achievable.
The American intervention should be evaluated with similar realism. It achieved important things. Al-Qaeda’s original sanctuary was destroyed. Millions of Afghan girls attended school for two decades. Cities developed. Media and civil society expanded. A generation experienced political and social freedoms that the Taliban now restricts.
Those outcomes mattered even if some proved fragile.
But the intervention also cost enormous amounts of money and thousands of lives while failing to construct a political order capable of surviving the withdrawal of foreign military support.
Both facts need to coexist.
Calling the entire twenty-year period meaningless ignores Afghans whose lives changed.
Calling it a successful state-building project ignores the speed with which the state collapsed.
The central lesson is probably that external powers can create institutional opportunity much more easily than institutional legitimacy.
America could finance elections.
It could not make political elites trustworthy.
It could train soldiers.
It could not make the government worth dying for indefinitely.
It could construct schools.
It could not guarantee that the political order protecting those schools would survive after American troops left.
Institutions become durable when domestic actors possess enough incentives to defend them themselves.
That threshold was never reached sufficiently.
The Taliban faces the inverse problem now. It possesses strong coercive control but weak international legitimacy and policies that undermine long-term economic development. Its regime may therefore remain politically durable while the country remains economically poor.
This is an important distinction because authoritarian durability should not be mistaken for national success.
A government can survive for decades while wasting enormous quantities of human potential.
North Korea demonstrates the extreme version.
Regime survival is one objective.
National welfare is another.
The Taliban has shown substantial ability to achieve the first.
Its restrictions on women make the second dramatically harder.
Afghanistan therefore presents a particularly stark example of the difference between state strength and societal strength.
The Taliban can control a checkpoint.
That is state capacity.
A girl learning medicine and becoming a doctor twenty years later is societal capacity.
The first produces order immediately.
The second produces development cumulatively.
Successful countries need both.
Four years after the American withdrawal, Afghanistan has become more politically predictable than many expected in August 2021. The Taliban did not fragment immediately. Another nationwide civil war did not begin. The regime consolidated.
But consolidation should not be confused with progress.
The country’s future is being shaped quietly inside classrooms that girls are prohibited from entering, universities missing half their potential students and workplaces
