Methodology

The Optional Work Index estimates how close a country is to the point at which paid work becomes economically optional, on a scale from 0 to 100. A reading of 0 represents the pre-automation baseline; 100 represents a state in which people can meet their needs without paid employment. The index is not a forecast. It is a measurement of present conditions.

Work becomes optional only when two conditions hold at once, so the index measures both and keeps them distinct.

These can diverge sharply. An economy can automate heavily while capital captures the gains, leaving work technologically unnecessary but not optional for individuals. The two-sided structure is designed to make that gap visible rather than average it away.

Each side is a weighted average of its sub-indices on a 0 to 100 scale. The two sides are combined by geometric mean rather than a simple average. If people cannot afford to stop working, capable machines do not make work optional. A small momentum factor, derived from the pace of recent confirmed milestones, is applied to the combined value.

The same formula, weights, and scoring thresholds are applied to all 150 scoring entities. Comparability comes from holding the method fixed; only the input data changes by country. For each input the system resolves the best available source for that country, recording which source filled each slot.

• 1World Bank — per-country indicators by ISO3; the primary backbone for labour, abundance and distribution.
• 2ILOSTAT — labour, informality and social-protection coverage.
• 3IFR World Robotics — robot density per 10,000 manufacturing workers.
• 4AI capability frontier — a global benchmark applied equally to every country.
• 5FRED / BLS — higher-resolution United States series, where available.
• 6ITU — ICT price basket: broadband cost as a share of income, for the affordability part of basic services.

Alongside the OWI, the project publishes the National AI Compute Index (NCI), a separate index. The NCI measures a country's effective AI compute: the reliable, affordable, accessible, and strategically controlled processing capacity available to its startups, researchers, enterprises, and public institutions. It is constructed as Raw multiplied by Enablement. Raw is hardware capacity (accelerators and clusters, plus data-centre capacity); Enablement is the geometric mean of six conditions: accessible, affordable, reliable, strategically controlled, talent, and software. Per-country figures are on the compute access page.

Cross-country data is uneven. Where a country lacks a primary series, the input is filled by a modelled estimate and flagged. Demand-side inputs, especially housing and the affordability part of basic services, have the weakest global coverage and currently lean on proxies. Every country page reports component-level coverage so readers can judge how much data stands behind a figure. Treat thin-coverage figures as provisional.

Model validation → · how the index is checked against ground truth.

The index is organised around a widely discussed projection that AI and robotics could make paid work optional within roughly ten to twenty years, attributed to public statements by Elon Musk (2024). The index treats this as a hypothesis to be measured, not a forecast it endorses.