Crisis validation
Did the score deteriorate when the economy did?
A risk index is only useful if it moves when real trouble arrives. Each documented crisis below is a fixed window taken from official and academic chronologies. The score for each year is rebuilt from published data alone using the same weights as the live index — no crisis information enters the calculation — and then compared across the three years before the crisis, the crisis itself and the two years after.
Crisis episodes tested
14
Country-specific windows across banking, sovereign-debt, commodity and pandemic shocks
Episodes with enough data to score
6
An episode is only scored when both the pre-crisis and crisis years have published data
Score rose during the crisis
6 / 6
Counted when the crisis-window score exceeds the pre-crisis average by more than 1 point
Episode by episode
| Crisis | Country | Window | Pre-crisis | During | After | Model reaction |
|---|---|---|---|---|---|---|
| Asian financial crisisInvestopedia — Financial Crisis → | 🇹🇭 Thailand | 1997-07 → 1998-12 | 22.8 | 35.0 | 17.2 | Risk rose +12.2 |
| Asian financial crisisInvestopedia — Financial Crisis → | 🇮🇩 Indonesia | 1997-08 → 1999-12 | 36.9 | 58.0 | 39.3 | Risk rose +21.1 |
| Asian financial crisisInvestopedia — Financial Crisis → | 🇰🇷 South Korea | 1997-11 → 1998-12 | 13.2 | 31.4 | 13.6 | Risk rose +18.1 |
| Russian default and rouble crisisInvestopedia — Financial Crisis → | 🇷🇺 Russia | 1998-08 → 1999-12 | 73.1 | 77.5 | 61.7 | Risk rose +4.4 |
| Argentine convertibility crisisInvestopedia — Financial Crisis → | 🇦🇷 Argentina | 2001-01 → 2002-12 | 56.5 | 87.4 | 43.3 | Risk rose +30.9 |
| Greek sovereign-debt crisisInvestopedia — Financial Crisis → | 🇬🇷 Greece | 2010-01 → 2016-12 | 23.5 | 48.4 | — | Risk rose +24.9 |
| Conflict and financial crisisInvestopedia — Financial Crisis → | 🇺🇦 Ukraine | 2014-01 → 2015-12 | 46.7 | — | — | Data unavailable |
| Oil-price and tenge devaluation shockInvestopedia — Financial Crisis → | 🇰🇿 Kazakhstan | 2014-07 → 2016-12 | 24.6 | — | — | Data unavailable |
| Brazilian recessionInvestopedia — Financial Crisis → | 🇧🇷 Brazil | 2015-01 → 2016-12 | — | — | — | Data unavailable |
| Oil-price recessionInvestopedia — Financial Crisis → | 🇳🇬 Nigeria | 2016-01 → 2017-12 | — | — | — | Data unavailable |
| Currency float and inflation crisisInvestopedia — Financial Crisis → | 🇪🇬 Egypt | 2016-01 → 2017-12 | — | — | — | Data unavailable |
| Turkish currency and debt crisisInvestopedia — Financial Crisis → | 🇹🇷 Turkey | 2018-01 → 2019-12 | — | — | — | Data unavailable |
| Balance-of-payments crisisInvestopedia — Financial Crisis → | 🇵🇰 Pakistan | 2022-01 → 2023-12 | — | — | — | Data unavailable |
| Full-scale war shockInvestopedia — Financial Crisis → | 🇺🇦 Ukraine | 2022-02 → 2022-12 | — | — | — | Data unavailable |
How to read this honestly
- This is a coincidence test, not a prediction test. A rise in the score during a crisis shows the index registers macro strain; it does not show the index would have warned in advance.
- The score is built from annual published data, which is revised and released with a lag. A crisis that started late in a year can therefore show up mostly in the following year.
- Unemployment and growth react fast, GDP per capita barely moves, and policy rates can fall in a crisis — which lowers the interest-rate sub-score even as the economy weakens. That is a real, deliberate property of the weighting, not a bug.
- Episodes where the score did not rise are shown, not hidden. Selective reporting of hits is exactly what makes most risk indices unfalsifiable.