Forecasts - Scale Systems

Nothing has resolved yet. The earliest horizon on this page is the end of 2027, so the scoreboard above will stay empty for a while and the Brier score is undefined until it fills. That is the honest state of a ledger that starts today rather than one assembled from predictions already known to have worked. For reference once it does fill: a Brier score is the mean squared distance between the stated probability and what happened, so lower is better, and 0.25 is what you would score by answering "50%" to everything. Anything at or above that is worse than useless.

72% confidence

The Bank of Japan's policy rate is still at or above 0.5% at the end of 2027 - that is, Japan does not return to the zero bound.

by end-2027 stated Aug 2026 Japan and Korea

Resolves on: Bank of Japan published policy rate on 31 December 2027.
Why this number: Japan left negative rates in March 2024 and has raised in steps to 1%. The Japan and Korea page argues the exit may be structural rather than a passing energy shock, because a shrinking workforce tightens the labour market permanently. If that reading is right, a return to zero would need an outright recession.
What would change it: A global recession, or Japanese core inflation falling below 1% for two consecutive quarters, would make a retreat to zero far more likely and this forecast wrong for the right reason.

80% confidence

South Korea's total fertility rate does not exceed 1.0 in any year through 2028.

through 2028 stated Aug 2026 Japan and Korea

Resolves on: Statistics Korea annual total fertility rate releases for 2026, 2027 and 2028.
Why this number: The rate bottomed at 0.72 in 2023 and rose to about 0.80 by 2025. The Japan and Korea page argues that rise is mostly a large early-1990s cohort passing through its thirties plus a post-pandemic marriage catch-up, not a response to the roughly $250bn of pro-natal spending. Echo effects of that size fade within a few years.
What would change it: A further rise above 0.9 in 2026 combined with continued growth in first marriages would suggest something structural is happening and that this reading is wrong.

92% confidence

China does not launch a full amphibious invasion of Taiwan before 2030.

before 2030 stated Aug 2026 Taiwan and the Silicon Chokepoint

Resolves on: Absence of a large-scale amphibious landing on Taiwan proper, per contemporaneous reporting and any official confirmation.
Why this number: The Taiwan page treats the status quo as the base case at roughly 60-75% over a decade, with blockade or quarantine as the more likely coercive option than invasion. Amphibious invasion across that strait is among the hardest military operations there is, and the cost of failure to the Chinese leadership is regime-level.
What would change it: Sustained large-scale amphibious exercise activity combined with a political signal that Beijing has abandoned the long game - or a Taiwanese formal independence declaration - would move this substantially.

88% confidence

The US dollar's share of allocated global foreign-exchange reserves is still above 45% at the end of 2030.

by end-2030 stated Aug 2026 The Dollar and the Monetary Order

Resolves on: IMF COFER quarterly release covering Q4 2030.
Why this number: The share has fallen from about 71% in 2000 to about 58% now - a slow drift of roughly half a point a year. The Dollar page argues no alternative has the combination of size, liquidity, and rule-of-law depth to absorb large reserve flows, and that the erosion is gradual rather than a displacement.
What would change it: A rapid move by several large reserve holders out of dollars, or a US default or capital-control episode, would break the gradualism this rests on.

78% confidence

Taiwan still produces more than 70% of the world's most advanced logic chips at the end of 2030.

by end-2030 stated Aug 2026 Taiwan and the Silicon Chokepoint

Resolves on: Leading-edge foundry capacity share per TrendForce or equivalent industry tracker, for the most advanced node in volume production.
Why this number: The Taiwan page puts TSMC's leading-edge share near 90% and argues that overseas fabs in Arizona, Japan and Germany are smaller, behind schedule, and will not catch the Taiwanese facilities for at least a decade. Process knowledge does not relocate on a policy timescale.
What would change it: Faster-than-expected yield parity at TSMC Arizona, or a serious Intel or Samsung recovery at the leading edge, would make this too confident.

82% confidence

Fossil fuels still supply more than 70% of world primary energy in 2030.

by 2030 stated Aug 2026 Energy Reality

Resolves on: Energy Institute Statistical Review of World Energy, 2031 edition, covering 2030.
Why this number: The Energy page notes the share fell from 94% in 1965 to about 81% today - roughly a fifth of a point a year - and that the constraint on the transition is replacement cycles for existing generation and grid assets, not preference or cost.
What would change it: A far steeper build-out of solar plus storage than current installation rates imply, or a demand collapse in a major economy, would beat this.

85% confidence

The United States does not experience a debt crisis - defined as a failed Treasury auction or a rate spike forcing emergency fiscal action - before 2032.

before 2032 stated Aug 2026 US Federal Debt

Resolves on: No failed US Treasury auction and no emergency fiscal legislation passed in response to borrowing costs, per Treasury records and contemporaneous reporting.
Why this number: The US Debt page argues the binding constraint is the gap between interest rates and growth, not the debt level, and that a country borrowing in its own currency from deep domestic and foreign demand does not face the mechanism people fear. Japan has run roughly double the ratio for fifteen years without it.
What would change it: Sustained real rates well above trend growth, or a durable loss of foreign appetite for Treasuries, would make this look complacent.

70% confidence

US measured labour productivity growth does not average above 2.5% a year over 2026-2031.

over 2026-2031 stated Aug 2026 Productivity and Growth

Resolves on: US Bureau of Labor Statistics nonfarm business sector output per hour, average annual growth across the six years.
Why this number: The Growth page sets out the optimistic case from task-level studies against Acemoglu's estimate of roughly half a percent of total TFP effect over a decade, and notes that general-purpose technologies historically take twenty to thirty years to show up in aggregate statistics because firms have to reorganise around them.
What would change it: Two consecutive years above 3% with the gain concentrated in AI-exposed sectors would be strong evidence the optimists are right and this is wrong.

88% confidence

No member state formally leaves the European Union before 2032.

before 2032 stated Aug 2026 Europe at a Crossroads

Resolves on: No completed Article 50 withdrawal beyond the United Kingdom's, per EU records.
Why this number: The Europe page argues that the EU's binding problem is slow decision-making and relative economic decline rather than dissolution, and that the observed cost of the British exit substantially raised the perceived price of leaving for everyone else.
What would change it: A eurozone crisis forcing a member out, or a governing party in a large member state winning an explicit exit mandate, would change this.

75% confidence

Affective polarisation continues to diverge across rich democracies rather than rising everywhere - at least three of the twelve countries in the Boxell-Gentzkow-Shapiro sample show flat or falling polarisation in the next comparable measurement.

next comparable measurement stated Aug 2026 Group Identity

Resolves on: A published update to Boxell, Gentzkow and Shapiro's cross-country affective polarisation series, or an equivalent multi-country panel using the same measure.
Why this number: Their finding was that six of twelve countries fell over four decades and that the drivers look US-specific. The Group Identity and Governance pages were corrected to say so. If the causes really are American, divergence should persist.
What would change it: A synchronised rise across almost all twelve would indicate a genuinely global driver and would mean the correction made to those pages understated the universality.

83% confidence

Japan's population is below 120 million in 2035.

by 2035 stated Aug 2026 Japan and Korea

Resolves on: Japan Statistics Bureau population estimate for 2035.
Why this number: Population peaked around 128 million in 2010 and is falling by roughly half a million a year, with the rate accelerating as the large post-war cohorts age. Only a migration change larger than anything currently contemplated would offset it.
What would change it: A migration regime shift on the scale of several hundred thousand net arrivals a year, sustained, would put this in doubt.

76% confidence

World total fertility falls below 2.1 - global replacement level - before 2035.

before 2035 stated Aug 2026 Demographics

Resolves on: World Bank world total fertility rate series, first annual reading below 2.1.
Why this number: The Demographics page notes the world figure is already close to replacement at about 2.19 and falling, and that fertility has declined faster than UN projections in most recent revisions rather than slower.
What would change it: A stall in Sub-Saharan African fertility decline would push this out; that region now carries most of the remaining global variation.

Open