Research note
What Makes a Good Country?
Evidence from 155 countries — what citizens say about their lives, what predicts it, and a ledger for writing your own answer
Abstract
What should a country be for? The standard yardstick — used throughout §1–5 — is the one measure where citizens grade their own countries: the Gallup World Poll asks people to place their lives on a ladder from 0 (worst possible) to 10 (best). This page sets those answers against twelve things that might explain them — money, health, freedom, trust, clean government, clean air. The catch is that those factors travel together (rich countries are also healthy and well-run), so ranking raw correlations mostly measures wealth over and over (§4); §5 splits the credit fairly instead. And because "a good country" is ultimately a philosophical choice, §6 hands it to you: weight what you value and see which countries — and which investments — fit. Methods and caveats in §9.
Key findings
- Ten factors together explain 84% of why some countries are happier than others (147 countries).
- Having someone to count on — social support — takes the biggest share of the credit (23.5%), and has ranked first every year since 2019, pandemic included.
- Money matters, but logarithmically: doubling income buys about the same happiness whether you start poor or rich (r = 0.80 on log income).
- Health, good government, and low corruption look strongly linked to happiness on their own — but most of that link is wealth wearing different costumes.
- Generosity, surprisingly, predicts nothing (0.3%); electoral democracy contributes little (5.3%) once governance quality is counted separately.
- The model misses in interesting places: Latin America is happier than its stats predict, and Botswana is 1.36 ladder points sadder.
1The world map
Every country, colored by whichever stat you pick — darker means more. Every stat can also travel through time: drag the slider (or press play) and watch up to two decades unfold on one fixed color scale.
Figure 1. Happiness (Cantril ladder (0–10)), 155 countries. Hatched countries have no data. Click a country to add it to the comparison below — click it again to remove it (outlined = currently selected).
How to read this
2Compare countries
Up to 5 countries side by side — every stat in the dataset, scanned across. Each number gets a hot/cold percentile rating against every other country: deep red = top of the world, deep blue = bottom.
Finland Europe & Central Asia | United States North America | |
|---|---|---|
| Happiness | 100 | 86 |
| Wealth | 85 | 94 |
| Inequality | 11 | 84 |
| Joblessness | 79 | 37 |
| Democracy | 94 | 74 |
| State Capacity | 98 | 85 |
| Integrity | 99 | 84 |
| Freedom | 96 | 25 |
| Support Network | 99 | 79 |
| Generosity | 41 | 81 |
| Longevity | 87 | 72 |
| Air Pollution | 4 | 12 |
| City Living | 66 | 74 |
Table 1. Country profiles, most recent values with the year each was measured. Bubble and bar rate the country's percentile among all reporting countries on one hot/cold scale — red = high for the world, blue = low, whatever the stat. The color says how high, not how good: hot Air Pollution is a lot of pollution.
3Happiness vs. each factor
One factor at a time: does more of it go with happier countries? The r number above the chart says how tightly the two move together — 1 is lockstep, 0 is no relationship at all.
r = 0.81 · n = 151 countries · log scale
Figure 2. Happiness vs. GDP per capita, most recent values. Every mark is a country, sized by population; the dashed line is the straight line that best fits the cloud. Click a mark to add that country to the comparison.
How to read this
4Everything correlates with everything
Here's the catch with §3: the factors aren't independent. Rich countries also tend to be healthy, well-governed, and less corrupt — so each of those looks strongly linked to happiness partly because all of them travel with wealth. This grid shows every pairwise relationship at once, and it's why the next section needs a fairer method.
| Happiness | Social support | Wealth (GDP) | Health (life exp.) | Gov. effectiveness | Freedom | Clean gov. (CPI) | Urbanization | Democracy | Air pollution | Unemployment | Inequality | Generosity | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Happiness | .82 | .81 | .75 | .73 | .66 | .63 | .63 | .54 | -.37 | -.28 | -.21 | -.11 | |
| Social support | .82 | .78 | .70 | .62 | .54 | .53 | .58 | .55 | -.44 | -.05 | -.18 | -.19 | |
| Wealth (GDP) | .81 | .78 | .86 | .83 | .45 | .70 | .65 | .47 | -.28 | -.08 | -.29 | -.16 | |
| Health (life exp.) | .75 | .70 | .86 | .76 | .43 | .66 | .57 | .47 | -.28 | -.15 | -.28 | -.15 | |
| Gov. effectiveness | .73 | .62 | .83 | .76 | .55 | .92 | .50 | .66 | -.37 | -.20 | -.28 | -.15 | |
| Freedom | .66 | .54 | .45 | .43 | .55 | .45 | .33 | .35 | -.16 | -.28 | -.04 | -.14 | |
| Clean gov. (CPI) | .63 | .53 | .70 | .66 | .92 | .45 | .48 | .67 | -.38 | -.12 | -.33 | -.08 | |
| Urbanization | .63 | .58 | .65 | .57 | .50 | .33 | .48 | .32 | -.14 | .06 | -.01 | -.13 | |
| Democracy | .54 | .55 | .47 | .47 | .66 | .35 | .67 | .32 | -.58 | -.04 | -.07 | -.14 | |
| Air pollution | -.37 | -.44 | -.28 | -.28 | -.37 | -.16 | -.38 | -.14 | -.58 | -.06 | .12 | .06 | |
| Unemployment | -.28 | -.05 | -.08 | -.15 | -.20 | -.28 | -.12 | .06 | -.04 | -.06 | .14 | -.19 | |
| Inequality | -.21 | -.18 | -.29 | -.28 | -.28 | -.04 | -.33 | -.01 | -.07 | .12 | .14 | -.14 | |
| Generosity | -.11 | -.19 | -.16 | -.15 | -.15 | -.14 | -.08 | -.13 | -.14 | .06 | -.19 | -.14 |
Figure 3. Correlation between every pair of stats, ordered by strength of association with happiness (top row). Blue = move together, red = move oppositely, deeper color = tighter link. Note the near-duplicate: the two clean-government measures agree at .91.
How to read this
5What actually matters
The headline result. A model with all ten factors explains 84% of why some countries are happier than others — and this chart splits that credit fairly between the factors, testing every possible ordering so wealth doesn't get to claim everything it's merely correlated with. If you were building a country from scratch, this is the priority list.
R² = 0.84 · n = 147 countries
Figure 4. Each factor's fair share of the explained variance, with whiskers showing the plausible range (resampling countries 1,000 times). † marks stats that come from the Happiness Report's own survey — the softest numbers here, which is why the second tab re-runs everything on fully independent data. Inequality is left out: 39 countries don't report it.
How to read this
6The Founder's Ledger
Everything above assumed happiness is the goal. But if you were founding a country from scratch, what would you value? Write your constitution below — weight the four ends, or start from a philosophy — and the ledger re-ranks the world by YOUR definition of a good country, then re-runs the fair-credit analysis to show which levers best predict it.
a balanced constitution — happiness weighted, nothing ignored
What matters in your country?
102 countries can't be ranked under this constitution — they don't report every end you've weighted (Equality is the usual gap).
Where your country already exists
Rise under your constitution
Fall under your constitution
Punching above their wealth
delivering your values beyond what their income predicts
To build it, invest in
fair share of what predicts your objective · R² = 0.87 · n = 114
What this can and can't tell you
7Where the model misses
No model explains everything. These are the countries whose reported happiness sits furthest from what their stats predict — the gap is what the ten factors can't see: culture, conflict, faith, or quirks of how the survey lands.
Positive residuals — above prediction
Negative residuals — below prediction
Figure 5. Reported minus predicted happiness, in ladder points (147 countries in the model). Latin America consistently beats its predictors; Botswana falls 1.36 points short. Click a country to add it to the comparison.
8Does the answer hold up over time?
A finding that only appears in one year's survey is an accident. Re-running §5 on each year separately shows the ranking barely moves — through COVID included.
| ’19 | ’20 | ’21 | ’22 | ’23 | ’24 | ’25 | |
|---|---|---|---|---|---|---|---|
| Social support | 18 | 17 | 20 | 22 | 23 | 24 | 24 |
| Wealth (GDP) | 17 | 17 | 17 | 19 | 18 | 18 | 17 |
| Freedom | 10 | 12 | 12 | 13 | 13 | 13 | 13 |
| Health (life exp.) | 13 | 14 | 14 | 10 | 12 | 12 | 13 |
| Clean governance | 14 | 13 | 13 | 13 | 11 | 10 | 11 |
| Urbanization | 13 | 12 | 10 | 10 | 11 | 11 | 10 |
| Democracy | 8 | 8 | 8 | 7 | 6 | 6 | 5 |
| Unemployment | 5 | 3 | 3 | 2 | 3 | 3 | 3 |
| Air pollution | 3 | 3 | 3 | 3 | 3 | 3 | 3 |
| Generosity | 1 | 0 | 1 | 0 | 0 | 0 | 0 |
Figure 6. Each factor's share of explained variance, year by year (2019–2025). Social support ranks first in every single year.
9Methods, sources & limitations
The fine print — in full, for anyone who wants to check the work.
Data
Happiness is the national mean Cantril-ladder score (Gallup World Poll), via Our World in Data; social support, freedom to make life choices, and generosity are the World Happiness Report's decomposition terms (Data for Figure 2.1, 2011–2025). Democracy is the V-Dem electoral democracy index (via OWID); perceived corruption is Transparency International's CPI (via OWID). GDP per capita (PPP, 2021 intl-$), unemployment, life expectancy, PM2.5 exposure, urbanization, population, and WGI government effectiveness are from the World Bank API. Snapshot values take each country's most recent figure within six years of the data frontier.
Model
Ten factors, z-scored, income as log₁₀; CPI and WGI government effectiveness are averaged into one governance composite (they correlate at r = 0.91 and would split importance credit). Relative importance is the exact Shapley value of each factor's contribution to R² over all 2¹⁰ subsets (equivalently LMG); confidence intervals are percentile bootstrap over countries (B = 1,000). Complete-case estimation, n = 147.
The Founder's Ledger (§6)
The custom objective is a weighted sum of standardized end scores (equality = inverted Gini), computed only for countries reporting every weighted end. The "invest in" chart re-runs the same exact-Shapley decomposition as §5 in the browser, with the four ends removed from the predictor set and unemployment and PM2.5 sign-flipped so every mean points toward its favorable direction. No bootstrap intervals are shown for custom objectives.
Limitations
All estimates are observational and cross-sectional: they identify covariation, not causation, and reverse causation is plausible for several factors. Three factors (†) are derived from the ladder regression itself and inherit part of its structure — hence the independent-sources re-estimate in §5. Country-level means say nothing about individuals (ecological inference). Coverage varies by indicator; income inequality is omitted from the model for this reason. Regional groupings follow the World Bank's 2025 classification.
Data retrieved 2026-10-04 · analysis computed 2026-10-04 · pipeline rebuilds nightly