Answer in brief
Zurich and Tokyo occupy UBS’s high-risk band in the new 2026 report. The comparison highlights valuation imbalances, with an August data cutoff and no timetable for a correction.
A new report changes the leading cities
UBS housing risk reporting has a new reference point for October. The bank released its Global Real Estate Bubble Index 2026 on 22 September, placing Zurich and Tokyo in its highest risk band. Their scores are 1.69 and 1.54 respectively. The report covers 23 selected cities, with Lisbon and Seoul entering the sample, rather than a census of the world’s property markets.
The ranking directs attention to the relationship between valuations and their economic support. It does not establish that a bubble has already burst or that the highest-scoring city must fall first. Readers comparing this edition with older headlines should use the current report’s score table and its revised framework.
Compare the scores within their stated bands
The selected-city table below reproduces UBS’s reported numbers, with no VJOURNAL composite or alternative ranking. UBS defines high risk above 1.5, elevated risk between 1.0 and 1.5, moderate risk between 0.5 and 1.0, and low risk below 0.5. Scores are dimensionless standardized measures, so a change of 0.1 is not a ten-percentage-point change in a crash probability.
Collection for this edition ended on 26 August 2026. The report’s annual real-price comparisons generally refer to the second quarter, while individual input series have different timing and some recent observations are estimated. Publication in September must not be mistaken for live September or October transaction pricing.
| City | UBS 2026 score | UBS risk band |
|---|---|---|
| Zurich | 1.69 | High |
| Tokyo | 1.54 | High |
| Miami | 1.41 | Elevated |
| Dubai | 1.16 | Elevated |
| Seoul | 1.13 | Elevated |
| Lisbon | 1.04 | Elevated |
| London | 0.32 | Low |
| San Francisco | −0.02 | Low |
Five inputs connect valuations to the wider economy
UBS combines price-to-income and price-to-rent measures, changes in mortgage-to-GDP and construction-to-GDP ratios, and a city-to-country price comparison. For Dubai, Hong Kong and Singapore, an inflation-adjusted price index replaces the last component. The weighting uses factor analysis, with limits intended to avoid excessive emphasis on national variables.
This structure matters because a city’s score can move even if its latest transaction price changes little. Income, rents, credit and construction can alter the supporting picture. It also creates sensitivity to revisions, weight changes and data availability. The report explicitly cautions that revised valuation benchmarks are not directly comparable with previous editions.
A low score can coexist with expensive housing
London’s 0.32 and San Francisco’s minus 0.02 belong to the low band in this framework. Neither number means every household can afford a home there. A risk score evaluates imbalances relative to its inputs; a household budget depends on deposit, income, mortgage terms and ongoing costs. Those are separate calculations.
Likewise, an elevated classification does not disclose a particular building’s condition, legal risks or likely sale time. Citywide measures compress different neighbourhoods and housing types. Reading the index as a property-specific recommendation would discard the very differences its methodology and uncertainty notes require readers to retain.
Independent global data provides a different lens
The Bank for International Settlements’ 27 August analysis reports global residential prices down 1.2% annually in real terms in the first quarter of 2026, alongside a 1.7% nominal increase. Its aggregates use national series and economic weights. They cannot confirm UBS’s city score or be substituted for the selected-city comparison.
The two publications instead illustrate why inflation treatment, geography and dates must travel with housing statistics. The new UBS report establishes a dated assessment of imbalances. Whether, when and by how much any city corrects remains unresolved by the scores themselves.
Questions and answers
Does a score of 1.69 mean a 169% crash probability?
No. The UBS score combines standardized indicators of valuation and economic imbalances. It is not a calibrated percentage probability, an expected price loss or a countdown to a correction.
Does a low bubble score mean housing is affordable?
No. A city can have a low imbalance score while homes remain expensive relative to a household’s income or borrowing capacity. Affordability and the index’s bubble-risk classification answer different questions.
Are the scores live October prices?
No. UBS published the report on 22 September and states that data collection ended on 26 August 2026. Component series have their own observation dates, and estimates fill some missing recent data.