Polycrisis Resilience: A General Definition and the H Heuristics Framework

How this edition works. The page follows the flagship's two parts. Part I (sections 1 to 7) develops the general definition: the condition of entangled crises, the anatomy of the definition, its six capacities, the scales at which resilience is held, neighbouring concepts, assessment and a 166-country scorecard. Part II (sections 8 to 14) presents the framework: its drivers, the evidence that development reduces vulnerability, the five systems, the development cycle, the multi-dividend screen, implications and ten propositions.

What is new here. Four interactive pieces go beyond the PDF: a recovery-window model that shows why bouncing back is not enough (section 3), the 166-country atlas and sortable scorecard with a test of whether weaknesses cluster beyond income (section 7), a disaster-mortality explorer by income group and country (section 9) and a multi-dividend screen you can re-cut by the vulnerabilities you care about (section 12). Everything else is the flagship's argument in web form.

Status of the evidence. The disaster-mortality series are public data (EM-DAT as published by Our World in Data). The scorecard is the H Heuristics Polycrisis Resilience 2026 panel built from World Bank indicators; the clustering test is new to this edition. The recovery model, the capacity matrix and the multi-dividend screen are conceptual illustrations and judgements, labelled as such where they appear. The full report is available as a PDF (HH-2026-55, DOI 10.6084/m9.figshare.34938445).

Climate change, environmental degradation, economic instability, geopolitical conflict, demographic transitions, public health threats, technological disruption and infrastructure vulnerabilities increasingly interact. These challenges do not operate independently: their interactions can amplify existing vulnerabilities, create cascading disruptions and make societies more susceptible to multiple simultaneous or sequential shocks.

The 2020s made the pattern hard to ignore. COVID-19 caused an estimated 14.9 million excess deaths in 2020 and 2021 (Msemburi et al. 2023) and the deepest global recession since the Second World War. Before recovery was complete, the war in Ukraine and tight supply chains sent food and energy prices to multi-decade highs and central banks raised interest rates. By 2025, 3.4 billion people lived in countries that spend more on interest than on health or education (UNCTAD 2025); forced displacement reached 123 million (UNHCR 2025); 295 million people faced acute food insecurity (FSIN and GNAFC 2025). Harm was greatest where several stresses converged on the same people and places.

The word polycrisis was introduced by Edgar Morin and Anne Brigitte Kern in the 1990s (Morin and Kern 1999) and returned to wide use in 2022 through Adam Tooze (Tooze 2022). The Cascade Institute gives the precise version: crises in multiple global systems become causally entangled in ways that significantly degrade humanity's prospects (Lawrence et al. 2022). The key word is entangled. Two unrelated events that happen to coincide are a coincidence; a polycrisis exists when the events alter one another's course.

Three mechanisms of entanglement. Lawrence and colleagues identify three ways crises become entangled (Lawrence et al. 2024). A common cause is one stress, such as climate change, driving crises in several systems at once. A domino effect is a crisis in one system triggering a crisis in another: drought raises food prices, subsidies strain public debt, austerity provokes unrest. An inter-systemic feedback is a loop: conflict causes hunger, hunger causes displacement, displacement fuels further conflict. The older systemic-risk literature describes the same channels (Centeno et al. 2015; Helbing 2013; Perrow 1984).

Why entanglement is more frequent. Three structural changes. Tighter coupling: in July 2024 a faulty software update disabled about 8.5 million Windows devices within hours, grounding flights and disrupting hospitals and banks (Weston 2024). Eroding planetary buffers: 3.3 to 3.6 billion people live in contexts highly vulnerable to climate change (IPCC 2023) and compound events are becoming more likely (Zscheischler et al. 2018). Depleted policy buffers: higher debt and weaker growth prospects in developing economies (Kose and Ohnsorge 2024) mean a fiscal squeeze from one crisis reduces the capacity to respond to all the others.

What this changes. If crises were independent, a society could prepare for each separately and add up the results. In a polycrisis that fails three ways: harm is non-additive (a flood in a currency crisis does more damage because reconstruction costs more to finance); recovery windows shrink (each shock strikes a system that has not recovered from the last); and vulnerabilities cluster in the same households, regions and countries. Resilience must be judged against combinations and sequences of crises.

Before, not only after. Polycrisis Resilience is not exclusively concerned with responding to crises after they emerge. It emphasises reducing vulnerability before disruptions occur, which makes prevention, development, adaptation and systemic improvement central. The emphasis has a practical basis: harm from shocks depends far more on pre-existing conditions than on the response. A day's notice of a storm or heatwave can cut damage by 30% (Global Commission on Adaptation 2019); resilient infrastructure in low- and middle-income countries returns about $4 per $1 (Hallegatte et al. 2019); disaster finance arranged in advance arrives faster and is better spent than appeals launched afterwards (Clarke and Dercon 2016).

Answering a common criticism. Resilience thinking is often charged with asking the vulnerable to bounce back from harms that could have been avoided (Béné et al. 2012). Polycrisis Resilience does not treat endurance as a virtue in itself. Its aim is to reduce the frequency and severity of the crises that people must endure, which is why reducing vulnerability is one of the five verbs in the definition, and why the objective is judged by whether people emerge from crises less exposed.

Vulnerability, as used here. The product of exposure (being in harm's way), susceptibility (how much harm a given shock causes) and coping capacity (the ability to limit and recover from harm), following the IPCC and UNDRR glossaries (IPCC 2022; UNDRR 2017). A crisis threatens essential functions and demands urgent response; a shock is a sudden event that triggers one; a stress is a slow pressure, such as debt accumulation or land degradation, that weakens a system over time.

What the model shows. A low-resilience system loses part of its welfare permanently with each shock, so losses ratchet down. A system that merely bounces back returns toward baseline between shocks but is hit just as hard each time, and when shocks arrive before it has recovered the gaps accumulate. A system with Polycrisis Resilience recovers and uses recovery to reduce its vulnerability, so each shock is smaller than the last. The difference comes entirely from the two assumptions the framework emphasises: recovery windows and vulnerability reduction. Drop the interval to three or four years and the bounce-back line starts each shock from a lower level.

What it leaves out. The model does not price the vulnerability reduction. In practice it requires investment, and report 2026-79 found that interest payments exceed health spending in 34 countries with 2.7 billion people, so the fiscal space to buy it is itself unevenly distributed. It also treats shocks as regular and equal in size; real shocks are lumpy: report 2026-80 found that 58% of recorded disaster damage and 76% of the death-rate total fall in the worst 1% of country-years, so realised returns to prevention are concentrated in a few extreme events. The model is a way to think with, not a forecast.

How the capacities interact. The six are distinct but not independent. They draw on the same assets: education, health, savings, infrastructure, institutions and trust. A strong health system contributes to prevention, absorption, recovery and adaptation alike, so investments in shared foundations strengthen several capacities at once, and where the foundations are weak all six are weak together, which is one reason vulnerabilities cluster. The capacities also involve trade-offs, such as buffers against transformation or efficiency against redundancy, which should be made explicitly with the full range of possible crises in view (Linkov et al. 2014).

How the scales interact. Resilience at different scales can be complementary (household savings work better when banks are sound), substitutive (where states are weak, households and communities carry more of the load through remittances and mutual aid) or contradictory (a food-export ban protects one country's consumers and raises world prices for everyone else). A shock that overwhelms one scale is passed up to the next, and a polycrisis is dangerous partly because it can overwhelm several at once. The strongest systems link the scales, as adaptive social protection does by using national delivery systems to channel support to affected households (Bowen et al. 2020).

Resilience for whom? Resilience is not distributed evenly. Informal workers, women, migrants, displaced people and the urban poor absorbed a disproportionate share of income losses in the early 2020s. The World Bank estimates that natural disasters cause consumption losses equivalent to US$520 billion a year, far more than the value of assets destroyed, because poor households cannot smooth consumption or rebuild (Hallegatte et al. 2017). The framework answers the question "resilience of what, to what?" (Carpenter et al. 2001) by including vulnerability reduction in the objective and by counting the functions the poorest depend on, such as food markets and primary care, among the essential ones.

Poverty traps. Barrett and Constas propose a theory of resilience built around thresholds: a household is resilient if it can avoid falling below a level of well-being from which it cannot recover on its own (Barrett and Constas 2014). Each shock that pushes a household below the threshold, by forcing it to sell livestock, take children out of school or skip medical care, raises its vulnerability to the next. A polycrisis, which delivers shocks in quick succession, is especially likely to push households into traps, so reducing vulnerability beforehand and protecting assets during the crisis are two sides of the same task.

An integrating concept. Polycrisis Resilience does not replace disaster risk reduction, climate adaptation or macroeconomic policy; it asks how they fit together when crises interact. A finance ministry that arranges contingent credit for disasters, or an energy ministry that reduces import dependence to protect the balance of payments, is practising it whether or not it uses the term. Its roots are the ecological definition (absorb disturbance and maintain relationships among the parts (Holling 1973)), social–ecological adaptability and transformability (Folke 2006) and the UN and IPCC disaster and climate definitions (UNDRR 2017; IPCC 2022).

Three critiques, three answers. Resilience can be conservative, protecting unjust or unsustainable arrangements; the concept answers with transformative capacity and by defining success as protecting essential functions for people, not any institution. It can shift responsibility to the vulnerable; the answer is to place risk reduction and prevention at the centre and treat resilience as a property of systems and institutions as much as of households. And it can be too vague to measure; the next section proposes principles and indicators.

Neighbours in one line each. Disaster risk reduction asks how losses from hazards can fall; climate resilience how societies cope with a changing climate; economic resilience how fast output recovers; systemic risk how failures propagate; compound events how hazards combine physically (Zscheischler et al. 2018); human security whether people are free from want and fear (UNDP 2014); antifragility whether a system can gain from disorder (Taleb 2012). Polycrisis Resilience adds the interaction, the full capacity cycle and an explicit vulnerability objective, and takes from antifragility only a guarded version: emerge from crises less vulnerable, without welcoming them.

Two kinds of evidence. Capacity indicators measure the assets, buffers and institutions expected to confer resilience before a crisis arrives. Outcome evidence measures how systems performed in past shocks: how far they fell, how quickly they recovered and whether they emerged less vulnerable. Neither is sufficient alone. Capacity indicators can be wrong about what matters, and outcomes reflect luck as well as resilience.

Five principles. Assess all six capacities, not only buffers, which miss prevention, adaptation and transformation. Look for clustering: how many capacities are weak at once matters more than the average, because a polycrisis exploits the absence of any fallback. Disaggregate by region and social group, because resilience for whom is part of the definition. Test against outcomes from past crises and revise measures where they fail. Publish components and use public data, so that weights can be challenged and results replicated.

Capacity is not performance. Before 2020 the Global Health Security Index ranked countries on epidemic preparedness; a study of 177 countries found those scores were not associated with lower infection or death rates, while trust in government and between people was associated with lower infection rates (COVID-19 National Preparedness Collaborators 2022). The H Heuristics scorecard reached a similar conclusion: conditional on income, most capacities predicted outcomes in 2020 to 2024 only weakly, with the exception of institutional quality, which was associated with markedly smaller losses of life expectancy. Resources on paper have to be converted into action under pressure.

Construction. The H Heuristics Polycrisis Resilience 2026 scorecard combines sixteen indicators, chiefly from the World Bank's World Development Indicators and Worldwide Governance Indicators (World Bank 2026; World Bank 2025): fiscal and financial buffers (interest share of revenue, revenue, reserves, debt service), social protection and health (under-five mortality, protection coverage, account ownership, undernourishment), institutions (government effectiveness, rule of law, political stability), infrastructure and digital access (electricity, internet, basic water) and external exposure (imported food and fuel). Each indicator is converted to a percentile and averaged within its capacity; the composite is the mean of the five. Higher means more capacity.

Two findings. First, capacity is steeply sorted by income: income per person explains 81% of the variation in the composite, which averages 22 in low-income and 70 in high-income countries. This is the strongest single piece of evidence for the development-oriented approach of Part II: the countries with least capacity are those with the largest development gaps. Second, income is not destiny: countries at similar incomes differ by twenty points or more, so choices about institutions, social protection and infrastructure shape resilience independently of income.

Where the gaps are. Low-income countries score 9 on infrastructure and digital access against 74 in high-income countries, and 15 against 81 on social protection and health. Fiscal buffers are thin almost everywhere, with a gap of only 13 points. The weakest capacity in most countries is institutional (47 countries), fiscal (41), external (36) or infrastructure (35); social protection and health is weakest in only 7, mostly middle-income countries.

Part I set out a general definition that any discipline, institution or society can apply. Part II presents one application: the H Heuristics Polycrisis Resilience Framework, developed by Hunter Hughes. It applies the concept through development, focused on reducing human and systemic vulnerability to interconnected crises. A country with reliable electricity, universal schooling, functioning health care, diversified exports and capable institutions will suffer less from almost any combination of shocks than one without them, however good its emergency plans.

Why development and resilience are one objective. In practice the two are kept apart: development through long-term investment, resilience through disaster management and humanitarian response, with separate budgets that compete for scarce resources. The framework rejects the separation for three reasons. Most of what makes societies vulnerable is a development deficit: poverty, ill health, poor housing, unreliable infrastructure, weak institutions, undiversified economies. Most of what protects them is a development achievement: savings, education, health systems, reliable power and water, financial inclusion, capable states. And crises reverse development, so development not protected by resilience is easily lost.

How it relates to the general concept. The general definition describes what Polycrisis Resilience is; the framework describes how to build it where development gaps are the main source of vulnerability. It gives particular weight to the capacities that crisis management most often neglects: risk reduction, adaptive and transformative capacity. The table below shows how its development pathway strengthens each of the six.

Convergence seeks to close the gaps in income, health, education and access to services between poorer and richer countries and communities. It matters for resilience because vulnerability falls steeply as these gaps close (section 8). After decades without systematic catch-up, there is evidence that convergence in income and many policy and institutional measures strengthened from the 1990s (Kremer et al. 2022), although long-term growth prospects have weakened since (Kose and Ohnsorge 2024).

Technology diffusion. Many of the most powerful tools already exist: solar power and batteries, mobile money, vaccines, drought-tolerant seeds, weather forecasting, digital public infrastructure. The constraint is adoption. Historically the average lag between a technology's invention and its adoption in a typical country has been about 45 years (Comin and Hobijn 2010); shortening it is one of the fastest ways to reduce vulnerability.

Sustainable development. Development that depletes its own environmental base, by degrading soils, exhausting aquifers or raising emissions, adds to future risk even as it reduces present vulnerability. The framework emphasises development that improves environmental conditions as well as material ones, so that today's gains do not become tomorrow's hazards.

Coordinated improvement. A new hospital without reliable power, or a social protection programme without a payment system, delivers less than it should. The framework emphasises improvements across interconnected systems, sequenced so that each strengthens the others.

The framework's premise is an empirical claim and can be tested. If development reduces vulnerability, then as countries develop, the harm caused by hazards of a given size should fall, and at any moment richer, better-educated and better-governed countries should suffer less from similar hazards. The evidence for both is strong, with qualifications.

Studying 73 countries, Kahn showed that richer countries do not experience fewer natural disasters but suffer far fewer deaths from them, and so do countries with stronger institutions (Kahn 2005). Toya and Skidmore found that higher income, education, openness and financial development go with fewer losses in deaths and as a share of GDP (Toya and Skidmore 2007). Formetta and Feyen found that vulnerability to climate-related hazards has fallen over time and declines with wealth (Formetta and Feyen 2019); Jongman and colleagues found the same for river floods, with the gap between rich and poor countries narrowing (Jongman et al. 2015). Education, female education in particular, is among the strongest single drivers (Lutz et al. 2014).

The protection extends beyond hazards. In Kenya, households using the M-Pesa mobile-money service absorbed negative income shocks without reducing consumption, whereas non-users' consumption fell by about 7% (Jack and Suri 2014); over the longer run mobile money lifted an estimated 194,000 households, about 2% of the total, out of poverty (Suri and Jack 2016). Countries with institutions that manage social conflict well recover faster from external shocks (Rodrik 1999).

The world series. In the 1920s natural disasters killed on average 26.5 people per 100,000 a year worldwide; catastrophic droughts and famines, floods in China and earthquakes accounted for most of the toll. By the 2010s the rate was 0.64, a decline of about 98%, and the 2020s so far (0.66) are similar. This occurred while recorded events rose from 84 in 1970 to between 407 and 451 a year in 2019–2025. Forecasting, early warning and building codes contributed, but so did higher incomes, the end of famine as a regular occurrence outside conflict zones, better roads and communications and stronger states.

Bangladesh. The 1970 Bhola cyclone killed an estimated 300,000 or more, 435 per 100,000 in EM-DAT's extreme-weather series; a 1991 cyclone killed about 138,000 (122 per 100,000); Cyclone Sidr in 2007, a storm of comparable strength, killed 3,400 to 4,200 (2.9) (Haque et al. 2012). The decline of more than 99% reflects deliberate preparedness (the Cyclone Preparedness Programme, founded in 1972, built a network of trained community volunteers, and thousands of shelters were built) and broad development: literacy, child mortality, roads, mobile coverage and incomes rose together, so warnings reached more people and more people could act on them.

What the groups show, and do not. The decadal rates by income group are lumpy because one catastrophe can dominate a decade: the lower-middle-income group's 59 per 100,000 in the 1940s is almost entirely drought (57), and the low-income group's 22.8 in the 1980s is drought (22.7). By the 2010s all four groups are at or below 1.2, and the low-income and lower-middle-income rates in the 2020s (0.6 and 0.2) are a fraction of their worst decades. The high-income rate is the exception: it rises to 2.1 in the 2020s, of which 2.0 is extreme temperature. Protection from floods, storms and earthquakes has been extended; protection from heat has not kept pace, which supports the framework's point that development must be sustainable and adaptive, not growth alone.

Why coordination matters. The last column of the table makes the section's central point: every system depends on the others. A social protection programme needs institutions to target it, infrastructure to pay it and fiscal resources to fund it; a power grid needs institutions to regulate it and an economy that can pay for it. Improvements made in isolation are often less effective than expected, while coordinated improvements can be more effective than the sum of their parts. This is the positive counterpart of the polycrisis: just as crises can compound one another, so can improvements, a property systems thinkers call reinforcing feedback (Meadows 2008) and practitioners call multisolving (Sawin 2018).

Where the gaps still are. About 666 million people lacked electricity in 2023, most of them in Sub-Saharan Africa, and about 2.1 billion lacked clean cooking (IEA et al. 2025); household air pollution from solid fuels causes an estimated 3.2 million deaths a year (WHO 2024). Just over half the world's population was covered by at least one social protection benefit in 2023, leaving about 3.8 billion with none (ILO 2024). Solar electricity from new utility-scale plants fell about 90% in cost between 2010 and 2023 (IRENA 2024), and Togo used mobile-phone data and machine learning to target emergency cash transfers during the pandemic (Aiken et al. 2022): the tools exist, and the constraint is adoption.

Iterative, not linear. Improvements in one area create opportunities in others, while changing conditions may require priorities to be reassessed. Electrification may make digital payments feasible where they were not, opening new options for social protection in the next cycle; a new climate projection, a debt crisis or a conflict may require a sharp change in priorities. Each real crisis is treated as a test of the system, and its lessons are fed back into the identification of vulnerability.

Why the design suits a polycrisis. Because crises interact in ways that are hard to predict, no plan can anticipate them all. A society that repeatedly assesses its vulnerabilities, invests in the most valuable improvements and learns from experience will, over successive cycles, become more resilient to crises that no one foresaw. This is the practical meaning of the claim that development and resilience are one objective: each turn of the cycle advances both.

Five criteria. The framework screens candidate interventions on breadth (how many vulnerabilities and systems does it affect?), depth (how much does it reduce each?), durability (do the benefits last and survive shocks?), distribution (does it reach people and places with the fewest buffers?) and deliverability (can it be implemented, financed and maintained with existing capacity?). The matrix below applies breadth.

An illustrative judgement. The matrix is an H Heuristics assessment drawing on the literature cited in the report, not an estimate of effect sizes, and the pattern will differ between contexts. It shows structure: reliable clean electricity, adaptive social protection and resilient infrastructure have strong or moderate effects on five of seven vulnerabilities; universal schooling on four. Clean cooking and primary health care have deep effects on fewer vulnerabilities and are best combined with broader investments.

Foundations first. Some interventions are foundational: electricity, digital identification and payments, statistical and registry systems, basic public financial management. They multiply the value of later investments because they are the platforms through which health, education, social protection and emergency response are delivered. The framework gives them priority early in the cycle even where their direct benefits seem modest. Framing them as resilience infrastructure, with measurable reductions in future crisis costs, can open finance unavailable to ordinary development spending: contingent credit, disaster risk finance, climate funds (Clarke and Dercon 2016).

Reliable clean electricity keeps clinics, water pumps, cold chains and communications running, powers productive work and schooling, reduces exposure to oil-price shocks and, where it replaces diesel and kerosene, reduces pollution. Distributed solar and storage now make it possible to reach many of the 666 million people without power faster and more cheaply (IRENA 2024; IEA et al. 2025).

Adaptive social protection combines routine support for poor households with the ability to scale up when a shock strikes, using registries, payment systems and pre-agreed triggers. Ethiopia's Productive Safety Net Programme and Kenya's Hunger Safety Net Programme show how such systems can respond to drought (Bowen et al. 2020). Resilient infrastructure protects the networks everything else depends on: the net benefit in low- and middle-income countries is estimated at US$4.2 trillion over the lifetime of new assets (Hallegatte et al. 2019).

Early warning saves lives at very low cost (Global Commission on Adaptation 2019); digital identification and payments let support reach people in days rather than months (Jack and Suri 2014; Aiken et al. 2022); and universal schooling, especially for girls, raises incomes, improves health and strengthens the ability to understand and act on warnings (Lutz et al. 2014). In a polycrisis the real choice is usually between investing now and paying more later: every crisis that pushes households into poverty traps creates costs that persist for years (Hallegatte et al. 2017).

For governments. Most governments manage risk sector by sector, and a polycrisis exploits the gaps between their plans. Assess vulnerability across systems and give one function, often at the centre of government or in the finance ministry, responsibility for understanding how crises would interact. Because budgets connect every sector, protecting spending on health, education, social protection and maintenance during crises, and arranging contingent finance in advance, stops short-term emergencies becoming long-term setbacks.

For development finance and communities. Development finance institutions already finance the systems the framework seeks to strengthen. The change is to treat resilience as a development outcome and development as a resilience strategy: count reductions in vulnerability among results, prioritise foundational systems and offer instruments that provide liquidity when crises strike. Much of the resilience that matters in a crisis is held locally, as Bangladesh's volunteers show; national actors should strengthen these capacities rather than bypass them.

Limits of the framework. Development is slow, so emergency response and humanitarian action remain essential; development can create new risks if it is not sustainable; national action does not remove the need for global action on the common causes of the polycrisis, above all climate change, ecological decline, conflict and the architecture of international finance; and the multi-dividend screen is a qualitative judgement to be tested with better data.

The crises of the twenty-first century do not take turns. They act on one another, and the harm they cause together exceeds the sum of their parts. The general definition holds that Polycrisis Resilience is the capacity of individuals, communities, institutions, economies, societies and interconnected systems to anticipate, withstand, adapt to, recover from and reduce their vulnerability to multiple interacting crises. It comprises six capacities, is held at every scale, and asks not only whether societies bounce back but whether they emerge less vulnerable, and for whom.

The framework applies the concept through development. Its premise, that reducing vulnerability to the polycrisis requires strengthening the systems, capabilities and development conditions that allow societies to become less exposed to interconnected risks, is supported by a century of evidence, and bounded by three limits: development can create new risks, falling mortality is not falling losses and development is slow and reversible. The framework turns the evidence into a method, a five-stage cycle repeated, each time from a higher base.

Data and method. Disaster-mortality series: Our World in Data, from EM-DAT (CRED/UCLouvain), decadal and annual deaths per 100,000 from natural disasters and event counts, downloaded November 2026. Scorecard: H Heuristics Polycrisis Resilience 2026 panel of 166 countries (World Development Indicators and Worldwide Governance Indicators, sixteen indicators in five capacities, each a percentile and averaged); income explanation is the R² of composite on log GDP per person; "vs income" is the residual. The clustering test shuffles each capacity's bottom-third flags across countries (5,000 draws) and within income groups (2,000 draws) and compares the count of countries with three or more. The recovery-window model, capacity matrix and multi-dividend screen are conceptual illustrations. Country boundaries: Mike Bostock's world-atlas.

Cite. Hughes, H. (2026). Polycrisis Resilience: A General Definition and the H Heuristics Framework (H Heuristics Research Report HH-2026-55). https://doi.org/10.6084/m9.figshare.34938445. Web edition: H Heuristics Digital Reports, 2026-87. The PDF is available at researchreports.hheuristics.com/reports/polycrisis-resilience-general-definition-framework/.