The Global Centre for Risk and Innovation (GCRI)
Systemic risk arises when systems depend on one another, so that a failure in one travels into others: a power outage stops water pumping, a port closure stalls manufacturing, a cyber incident halts payments. Cascades can amplify a modest shock into a large loss, or be absorbed by buffers and substitutes. This page sets out how cascades develop and propagate, how dependencies are mapped, and what the evidence serves for those responsible for connected systems.
GCRI · In development · service level R0
Cascading failures are failures that spread from one system to others that depend on it, as when a power outage stops water pumping or a port closure stalls manufacturing. Buffers set their timing, concentration creates single points of failure, and feedback can amplify losses. Dependency maps and scenarios show where cascades are likely to travel.
For cascades and interdependence, in GCRI's spatial-intelligence reading, the map opens on night lights, true-colour imagery and population density with relief in 3D, framed on Asia; every other NASA layer, live feed and forecast waits in the layer list.
The interactive map needs JavaScript; the layer list and sources are below.
Layers 3
Hazards and exposure, for this page
Keyboard, with the map focused
Hold Ctrl or ⌘ and scroll to zoom
Latest available
NASA GIBS / EOSDIS · ASTER GDEM (METI, NASA) · Blue Marble (NASA) · © OpenStreetMap contributors (ODbL) · Natural Earth
Loading the latest NASA imagery.
Drag to move. Ctrl or ⌘ and scroll, pinch or the buttons to zoom; hover for the values under the pointer; reset returns to the region.
For cascades and interdependence, in GCRI's spatial-intelligence reading, the map opens on night lights, true-colour imagery and population density with relief in 3D, framed on Asia; every other NASA layer, live feed and forecast waits in the layer list.Satellite and model products as NASA publishes them, each with its own date; indicators, not assessments of any place; claim ceiling INTEGRATED.
Layers and dates
Network, common-cause and correlated failures. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
When one system supplies another with power, data, materials or services, its failure removes an input the second system needs. Each link adds a path, and paths combine into chains that cross sectors and borders.
Stocks, backup generation, redundant routes and spare staff delay the effect of a lost input. The cascade arrives when the first buffer runs out, which can be hours or months after the trigger.
Where many systems rely on the same supplier, hub, cloud region or chokepoint, one failure reaches all of them at once. Concentration often grows unseen because each dependent made a reasonable choice alone.
Losses can feed back: an outage reduces revenue, which cuts maintenance, which raises the chance of the next outage. Reinforcing loops turn a contained event into a longer and wider one.
Each term as this page uses it. Where a standard-setting body defines the term, the wording follows that body.
| Term | Definition |
|---|---|
| Interdependency | A two-way relationship in which the state of each system affects the state of the other. |
| Single point of failure | A component or supplier whose loss stops a function because no substitute exists. |
| Buffer | A stock, reserve or redundant capacity that delays or absorbs the effect of a lost input. |
Each indicator carries its unit, its cadence, the kind of source behind it and the evidence label a value would carry. No values are shown: a programme supplies them under its own authority.
| Indicator | Unit | Cadence | Source type | Label |
|---|---|---|---|---|
| Mapped critical dependencies | count of supplier and service dependencies recorded | annual | Dependency inventories | DECLARED |
| Single points of failure | count of dependencies without a substitute | annual | Dependency analysis | ESTIMATED |
| Buffer endurance | hours or days of operation after loss of an input | per asset | Continuity plans and asset records | DECLARED |
| Supplier concentration | Herfindahl-Hirschman index of spend or volume | annual | Procurement records | OBSERVED |
| Cascade depth in scenarios | number of systems affected downstream | per scenario | Interdependency models | MODELLED |
| Restoration sequence time | hours to restore dependent services | per event | Incident and restoration records | OBSERVED |
The topic as a computational graph: named sources feed the methods, an evidence review gates the outputs, and the decisions stay with the authorities that hold them.
Fig. 02Nodes carry their type, two metric rows and a state; connectors show what feeds what; dashed frames name the owners; the timeline is the cadence. Design of the pipeline, not a record of a run.
Named public datasets and institutional records, each with its licence class. Sources by agreement are used only under their terms and never republished.
| Source | What it gives | Licence class |
|---|---|---|
| Incident reports from infrastructure operators and regulators | Documented failures with their causes and downstream effects | Public records, terms per publisher |
| UNDRR Global Assessment Report | Analysis of systemic and cascading risk across sectors and scales | Public, UN terms of use |
| OECD national supply and use tables and Inter-Country Input-Output tables | Economic dependencies between industries and countries | Open, OECD terms of use |
| Grid and network operator outage databases | Records of outages, duration and customers affected | Public summaries; full data by agreement (to check) |
| Academic interdependency case libraries | Reconstructed cascades from past events, with methods described | Licence per publication |
Systemic Risk Management prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| Who decides | The question | Evidence this page serves |
|---|---|---|
| Infrastructure operators | Which upstream dependencies need backup, diversification or contractual protection | Dependency maps; single points of failure; buffer endurance |
| Emergency management agencies | Which services must be restored first after a widespread outage | Restoration sequences; cascade scenarios with timing |
| Sector regulators and competition authorities | Whether concentration in a shared provider needs oversight | Concentration analysis with its data gaps stated |
| Treasury and planning ministries | Where public investment in redundancy reduces the most shared exposure | Cascade depth in scenarios; restoration evidence |
In a network, each node carries load or provides service to others. When a node fails, its load shifts to neighbours. If they are close to capacity, they fail too, and the cascade continues until the remaining system can carry the load or the network fragments. Highly connected hubs make networks robust to random failures but vulnerable to targeted or hub failures.
Common-cause failure defeats redundancy. Two backup systems that share a power supply, a software version, a supplier or a location may fail together. Correlation often rises under stress: assets, markets and infrastructures that move independently in normal times move together in a crisis. Feedback loops, such as forced selling or panic buying, can turn a moderate shock into a severe one.
Network science and contagion models simulate how failures spread across a mapped network. They depend on knowing the network, which is often partial.
Fault trees and Bayesian networks represent how combinations of failures lead to outcomes. They capture common causes when those causes are known.
Statistical models of joint extremes, from probability and statistics, capture tail dependence. Data on joint extremes is scarce by nature.
Simulate behaviour of many interacting actors. They are useful for insight but hard to validate.
The current service level is R0: information and enquiries. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.