The next institutional crisis is already visible—not as a scheduled event, but in the structural conditions accumulating before one. Identifying those conditions before they harden into crisis is the purpose of structural analysis.
Prediction Is the Wrong Standard
Crises are rarely predictable in the strongest sense. Structural analysis usually cannot name the date, trigger, institutional casualty, or sequence of contagion. What it can identify is vulnerability: leverage without buffers, exposure without oversight, obligations without capacity, or power without effective restraint.
This distinction matters. To say that the 2008 financial crisis, the COVID-19 pandemic, or democratic backsliding was “predicted” is too broad. In each case, however, important components of the eventual failure had been documented before the consequences became impossible to ignore.
In June 2007, the Bank for International Settlements warned that securitisation had contributed to weaker credit standards and that mortgage risk was being underpriced. The later majority report of the Financial Crisis Inquiry Commission concluded that the shadow banking system had grown to rival conventional banking under inadequate supervision and had become fragile through leverage, short-term funding, risky assets, and insufficient liquidity. These analyses did not forecast the precise collapse that followed. They identified much of the machinery through which it could occur. ([bis.org](https://www.bis.org/publ/arpdf/ar2007e.pdf?utm_source=openai))
The pre-pandemic record is clearer still. A 2019 World Health Organization survey found that many countries still lacked important preparedness capacities or current pandemic plans. That September, the Global Preparedness Monitoring Board warned of an acute pandemic risk, including a rapidly spreading respiratory pathogen, and concluded that the world was not prepared. The warning did not identify SARS-CoV-2. It identified the vulnerability that SARS-CoV-2 would encounter. ([who.int](https://www.who.int/publications/i/item/9789241515962?utm_source=openai))
Democratic backsliding is different because it is usually not a single event. It is an institutional crisis unfolding by increments: weakened oversight, constrained opposition, attacks on independent media, politicised administration, and the gradual concentration of power. The V-Dem Democracy Report 2025 identified 45 ongoing episodes of autocratisation as of 2024 and stressed that contemporary deterioration is often slow enough for its beginnings to be mistaken for noise. Here the structural precursor and the crisis may be the same process viewed at different stages. ([v-dem.net](https://www.v-dem.net/documents/61/v-dem-dr__2025_lowres_v2.pdf?utm_source=openai))
The common lesson is not that analysts possess prophecy. It is that institutions often receive enough information to justify precaution long before they receive enough political pressure to act.
The Warning-to-Action Failure
Preparedness has a political handicap: its costs are immediate, while its benefits are delayed and partly counterfactual. A reserve fund, regulatory constraint, redundant supply chain, inspection regime, or monitoring system must be paid for now. If it succeeds, the disaster it helped prevent never supplies dramatic evidence of the investment’s value.
Responsibility is also distributed. The institution detecting a risk may lack the authority to regulate it. The institution with authority may lack technical capacity, appropriated funds, or political permission. Private actors may capture the benefits of risk-taking while transferring part of the eventual loss to the public. Each participant can therefore acknowledge the aggregate danger while waiting for someone else to absorb the cost of preparation.
Uncertainty reinforces the delay. Before a crisis, uncertainty is treated as a reason not to act: the event may not happen, the model may be wrong, the intervention may be excessive. After the crisis, the same uncertainty disappears from political memory, and the failure is redescribed as obvious. Structural analysis must resist both errors. Vulnerability is not certainty, but uncertainty is not safety.
The Conditions Accumulating Now
AI governance presents one version of this mismatch. Across 11 selected United States agencies, the Government Accountability Office found that reported AI use cases nearly doubled from 2023 to 2024, while reported generative-AI uses increased roughly ninefold. Agencies also reported difficulties keeping policies and technical resources aligned with the technology’s development. Meanwhile, the NIST AI Risk Management Framework was expressly designed as a voluntary framework. This does not establish that an AI crisis is imminent. It does show how deployment can advance faster than inventory, evaluation, appeal, accountability, and oversight capacity. ([gao.gov](https://www.gao.gov/products/gao-25-107653))
Climate governance presents another. The UN Environment Programme’s 2024 assessment found that existing policies and national commitments remained far from emissions pathways consistent with the Paris Agreement’s temperature goals. The IPCC has likewise concluded that deep emissions reductions require rapid and far-reaching transitions across sectors and systems. The structural question is therefore no longer whether the physical risk has been identified. It is whether institutions can finance, permit, coordinate, build, and distribute the costs of transition at the required speed. ([unep.org](https://www.unep.org/resources/emissions-gap-report-2024?utm_source=openai))
Democratic resilience is the third domain, but it should not be described only as a future emergency. In many countries, institutional erosion is already under way. The danger is that gradual losses remain politically tolerable until accountability mechanisms can no longer correct them. An institution may retain its name and formal procedure after losing the independence or capacity that made it effective.
What Structural Analysis Must Provide
A useful warning must do more than announce danger. It should identify what is accumulating, what could trigger a phase change, how damage would propagate, which institutions possess authority to intervene, which incentives prevent them from doing so, and which capacities must be built before emergency politics begins.
That final question is decisive. Once a crisis is visible to everyone, political attention and public money may become abundant, but time, trust, institutional competence, and reversible options become scarce. The work of preparedness is to exchange resources that are available before the crisis for capacities that will be indispensable during it.
The visibility of a structural risk does not ensure institutional response. Preparedness is weakest when prevention is costly now, its benefits are diffuse and counterfactual, authority is divided, and the consequences remain politically deferred.
Discussion