Patrick BrothersTransformational Leadership FellowBlavatnik School of GovernmentUniversity of OxfordThe National Capability Framework is the focus of my fellowship at the Blavatnik School of Government, University of Oxford. I build and iterate on prototypes to understand complex problems: their key components and how they connect and depend on one another. The Framework is one such prototype, and it improves each time policy experts, economists and domain specialists test and challenge it.
Many thanks to the academic, industry and government colleagues and experts who have applied, challenged and tested the framework and, in doing so, have contributed to its ongoing development.
A changing world, and a measurement gap
What, for whom, why now
Policy is reorganising around national capability: the know-how a nation holds and can put to work. Global issues such as strategic competition, reconfigured global trade and the arrival of general-purpose AI have revived industrial policy and shifted the question governments ask from ‘how large is a nation?’ to ‘what can it actually do, and what can it do on its own?’. National agendas now include sovereign AI, rearmament, chip sovereignty and energy independence and, in several regions, strategic autonomy. For resource-constrained emerging economies the stakes are arguably higher. Know-how that diffuses through workers, firms and institutions can accelerate development far faster than resources alone. In every case, the priority is to build indigenous capability rather than buy it.
The resulting decisions go well beyond resource allocation and the choice to build, buy or partner. They extend to sequencing, alliances and blocs and cross-domain trade-offs, where capabilities in one domain may depend on or interact with those in another. Yet the main classifications that inform these decisions each cover a single domain, and few frameworks look across the full spectrum of national capability.
Data is not necessarily the constraint. There is an abundance of traditional measures of inputs and outputs, stocks and flows: GDP, reserves, fleets, population, trade, investment, energy, patents and publications. What is missing is a comprehensive and consistent way to measure know-how, the capacity to convert resources into action. Know-how is where capability lives, and existing measurement serves it least well.
Two nations with similar stocks and flows can differ sharply in what they can do with them. A nation with fewer resources but superior or novel know-how can outperform one that dominates it on paper, as Taiwan in chip fabrication, South Korea in shipbuilding, the Netherlands in lithography, Estonia in digital government, Israel in cyber and Ukraine in battlefield drones have shown. Know-how can also transform the value of resources. American advances in horizontal drilling and hydraulic fracturing turned long-known shale deposits into the largest oil and gas output in the world. China’s mastery of rare-earth separation and refining, rather than the deposits themselves, gives it control of roughly 90 per cent of global processing. India’s hyper-efficient space engineering made it the first nation to land a spacecraft near the Moon’s south pole, in 2023, for less than the budget of many Hollywood films.
What stakeholders will do without a shared framework
Existing tools, and their limits
The Framework is designed for three primary audiences. The first is governments and policy institutions, which increasingly make decisions on industrial strategy, alliances and strategic investment that span domains. The second is researchers and analysts working on national power, technology competition and state capacity, who currently reconcile difficult to align taxonomies with little visibility of how each was built. The third is the wider community of builders, including data providers, standards bodies and civil-society organisations, for whom an open, versioned classification is a precondition for work that fits together. What unites all three is a need for a common language for national capability, and that is what the Framework sets out to provide.
In the absence of a shared classification of national capability, stakeholders fall back on four traditional approaches. They reconcile single-domain taxonomies (critical technology, military balance, governance, research fields) informally, which works until a question spans them. They rely on composite indices of state capacity, which are typically low in resolution and do not expose their internal classifications for reuse; many reducing a complex system to a handful of country-level scores, which can show that one nation governs ‘better’ than another but not which specific capabilities it holds or lacks. They commission bespoke taxonomies that leave no durable public reference. Or they fall back on traditional stock and flow metrics, which are plentiful and comparable but measure inputs and outputs rather than the capacity to act.
The fourth approach is the default, and its limitation is the point. Traditional metrics are indispensable, but on their own they generally reduce national capability to size.
Decision-makers need two things these tools do not yet provide together: the granularity to see discrete capabilities, and a holistic structure that places those capabilities side by side across the full spectrum of national capability. Meeting both needs means balancing breadth and depth. A framework broad enough to span defence, technology, human capital, institutions and the economy risks becoming shallow; one deep enough to support measurement risks becoming unmanageable. A primary objective of the National Capability Framework is to strike that balance, so that the existing abundance of data can be read as capability.
Nations, capabilities and the classification
What is classified, what is observed, and how
The National Capability Framework distinguishes between what it classifies and what it observes. The Framework classifies capabilities, while nations are the unit of analysis, observed against each capability. Each observation is therefore a nation–capability pair: Taiwan’s capacity to fabricate leading-edge semiconductors, for example, or Estonia’s capacity to deliver public services digitally. The same split underpins established statistical standards. The International Standard Industrial Classification (ISIC), for example, classifies economic activities, while the establishment is the unit that is counted.
A capability is the capacity to act. It is distinct from resources (the inputs available), performance (observed results), power (capacity realised through will and opportunity) and outcomes (downstream effects). Capability is the conversion layer between resources and results, which is why know-how sits at its centre. To keep definitions consistent, every capability in the Classification is described as ‘the capacity to…’.
The Classification currently organises 288 capabilities into 73 capability groups, nine domains and three dimensions (hard, soft and economic) drawn from the classical analysis of national power. The levels run from broadest to most specific: dimension, domain, capability group and capability. Every capability carries an identifier so that, once the Classification is stable, indices, research and AI systems built on it can be compared and combined.
Measures sit in a separate indicator catalogue linked to each capability. It covers both traditional measures of inputs and outputs, stocks and flows, and new measures of know-how as they are developed. Keeping the two apart means measurement can evolve without changing the structure. It also means the Classification organises capabilities without evaluating them, much as a library classification organises books without reviewing them.
The current reference set covers 197 nations: the 193 UN member states, the two UN observer states (the Holy See and Palestine) and two states with limited recognition (Kosovo and Taiwan). ‘Nation’ is used deliberately rather than ‘state’. It admits polities that are not universally recognised as sovereign but whose capability is highly consequential. The set is drawn from the 249 ISO 3166-1 codes plus Kosovo, and excludes dependent territories, special administrative regions and the Antarctic Treaty area.
Although nations are the reference unit of analysis, the Classification does not depend on the unit. In principle, it can be applied equally to sub-national jurisdictions, economic zones and groupings of nations, which is essential for questions such as the combined capability of AUKUS, BRICS or the European Union.