Does hosting the World Cup change a city's wellbeing?

We estimate the effect of hosting 2026 World Cup matches on resident wellbeing, reading wellbeing from the words people search on Google in each US media market (DMA), and comparing every host city to a synthetic control — a data-built "digital twin" of that city assembled from non-host markets.

1 · Reading wellbeing from search — the keyword list

Following the GNH.today methodology of Greyling & Rossouw (2025), certain search terms track happiness measures. We build a composite wellbeing index from affect words (positive terms add, negative terms subtract, each standardized on the pre-tournament baseline).

We also track expected-confound words — terms like “game” or “win” that rise mechanically during a tournament regardless of mood. They are excluded from the wellbeing index and form a separate confound index that doubles as a positive control: it should spike when matches are played. Every word is collected so you can inspect any of them on the Results page.

2 · The digital twin — synthetic control + CausalImpact

For each host DMA we build a synthetic control: a weighted blend of non-host DMAs chosen so that, before the tournament, the blend tracks the host's wellbeing index as closely as possible (minimizing pre-period error, MAPE). That blend is the host's counterfactual twin — our best estimate of “what wellbeing would have done with no World Cup.”

Once matches are played, the gap between the host's observed index and its twin's predicted index is the estimated hosting effect (a modified CausalImpact, the same family as ad-lift geo-experiments). Because the US co-hosts, national-team excitement lifts every market — differencing each host against its twin removes that nationwide effect and isolates the hosting-specific signal.

Explore each host city's synthetic control

Pick a host market to see the donor DMAs that compose its twin. Line and dot size are proportional to each donor's weight.