Calibration
Where the values each country starts the game with come from, and what the game does when no source has an answer.
In plain words
The game starts on 1 January 2026, with each country as the data show it in 2025 for its growth, its inflation, its unemployment, its budget, its debt, the rate of its central bank and its electoral calendar. The parties, their votes and the governments are those of 1 January 2026 (E19). Each figure has a first source and a fallback source, and when neither has an answer, the country takes the median of its group. The game records, country by country, each figure estimated in this way, and points it out to the player at the moment they choose their country. The population of each country is divided into groups using data from the World Bank, the United Nations and the International Labour Organization (Population and incomes). Tax rates come from OECD statistics and, outside the OECD, from open sources or estimates (E35).
Where each starting figure comes from
The model starts from World Bank and IMF data. The starting year is 2025, and the game begins on 1 January 2026.
| Quantity | First source | Then | Number of countries per source (213) |
|---|---|---|---|
| population, GDP in dollars | World Bank | IMF (Taiwan) | 212 + 1; 210 + 3 |
| potential growth g* | IMF, average projection for 2029-2030 | World Bank, median 2010-2019 | 189, 23, 1 through the fallback |
| equilibrium unemployment u** | IMF, projection for 2029-2030 | World Bank, median 2016-2025 | 102, 88, 1; 22 through the fallback |
| inflation target π̄ | IMF, projection for 2029-2030, kept between 1 and 10% | World Bank, median 2010-2019; otherwise 2% (advanced) or 5% | 189, 11; 13 by default |
| starting inflation | IMF, December on December 2024 | World Bank, then IMF, annual average | 194 + 1; 18 through the fallback |
| starting unemployment | World Bank, 2024 or the nearest known year | IMF | 184 + 7 |
| starting policy rate (E7) | BIS, December 2024 | the Taylor rule | 58 countries, euro area included; 155 through the rule |
| debt, revenue, government balances | IMF, general government | World Bank, central government | 191 to 195 from the IMF |
| effective rate on the debt | (primary balance − overall balance) / debt, between 0 and 25% | — | derived |
| group (advanced, emerging, low-income) | World Bank income group (high; middle; low) | Taiwan, by its GDP per capita | 84, 106, 23 |
| volatility of growth and inflation (E13) | World Bank, 2000-2019, robust standard deviation | IMF | 208 + 2 and 186 + 11; 3 and 16 through the fallback |
| population growth | World Bank, median 2016-2025 | IMF | to show purchasing power per capita |
| electoral calendar (E15) | Wikidata, CC0 | — | 223 countries on the map; 207 of the 213 simulated countries vote, the other 6 never do |
| debt limit (E16) | Ghosh et al. 2013, table 3 | the group median | 19 advanced countries |
The fallback: when no source gives a value, the country takes the median of its group. These are mostly small territories.
The starting bounds (choices, not yet argued): the calibration keeps each value it reads within a range, so that an aberrant figure does not start a country from a point the model cannot represent. Potential growth lies between −1 and 8%; equilibrium unemployment between 0.5 and 35%, starting unemployment between 0.1 and 50%; debt between 0 and 400% of GDP, revenue between 2 and 80%, primary spending between 2 and 90%; population growth between −3 and 6%; the effective rate on the debt between 0 and 25%, and 3% when debt is below 1% of GDP. These bounds come from the model. They have yet to be compared with the observed extremes, as those of the variables are (§3.1).
Why December-on-December inflation. It is the twelve-month rate that the statistics office publishes in January, and the pace at which prices are rising on that day. The annual average lags behind a rapid disinflation. Argentina had 220% on average in 2024, but 118% from December to December. And a December policy rate only matches December inflation. Since these are IMF data, they stay in the local cache, with the fallback on the World Bank annual average.
Why the IMF projections for 2029-2030. A five-year projection assumes, by construction, that the output gap has closed and that inflation is back at the target. This is exactly what potential growth, equilibrium unemployment and the target mean. The 2010-2019 median is a poorer substitute. It put France's equilibrium unemployment at 9.6%, still marked by the euro crisis, while the IMF projection puts it at 7.5%. The 2016-2025 median brings it back to 7.5.
The 38 countries outside the economy: territories without their own statistics (Åland, Antarctica, Anguilla, Gibraltar, the Pitcairn Islands, Niue, Vatican City…), disputed or unrecognised territories (Somaliland, Western Sahara, Northern Cyprus, the Siachen Glacier, the Spratly Islands), bases (Guantánamo, Akrotiri and Dhekelia, Baikonur), and North Korea and Eritrea, for which neither the World Bank nor the IMF publishes data. What becomes of them is still to be decided.
The country-specific values that come from a study are the 25 Okun coefficients (Ball et al. 2019) and the 19 debt limits (Ghosh et al. 2013). All other parameters are set by group (§4). The size of the shocks is specific to each country, but it is computed from the country's own data (E13).
Countries out of domain. Two countries are simulated and shown like the others, but their failures in the checks are reported separately, with the reason. Macao starts with a primary deficit of 37 points of GDP, paid for by its reserves, which the model does not see because it only knows gross debt. Zimbabwe starts from 687% inflation, beyond the hyperinflation bound. A listed country that stops failing makes the check fail, so a country cannot stay on the list once its reason has gone. Timor-Leste, once listed for the same reason as Macao, left the list this way. So did Sudan, on 23 September 2026. It was in de facto default at the start (arrears of 263% of GDP, revenue of 2.9% of GDP) and could not get out of its successive defaults. It has been getting out of them since monetary transmission was corrected (E1).