For two centuries, a powerful assumption has governed thinking about political development: that as societies become wealthier, healthier, and better educated, they inevitably become freer. This is the core prediction of mmodernisation theory, from Lipset (1959) through Fukuyama (1992).
The data no longer support it.
Using the Human Capabilities Index (HCI) — a 15-indicator composite measuring health, education, material welfare, and agency across 91 countries — matched to Political Topology Liberty scores over 225 years, we document a secular decline in the correlation between capability, and freedom. The relationship that was once amongst the strongest in social science (r = 0.79) has weakened to r = 0.57 and continues to fall.
By 2023, the number of Capable Autocracies (HCI ≥ 70, Liberty < 60) equalled the number of Free and Capable states for the first time in recorded history. The autocrat's bargain now works in purely material terms.
| Period | Correlation (r) | Key Dynamic |
|---|---|---|
| Pre-1900 | 0.79 | Only free nations could mobilize the institutions needed for development |
| 1900–1945 | 0.74 | Colonial extraction begins building infrastructure without freedom |
| 1945–1990 | 0.61 | Soviet model demonstrates industrial capability under totalitarianism |
| 1990–2023 | 0.57 | China, Gulf states, and Asian tigers prove autocratic mmodernisation at scale |
The decline is not gradual noise. It reflects a structural transformation in the global political economy: authoritarian regimes have learned to deliver the material components of human welfare — healthcare, education, infrastructure, even economic growth — without political opening. They have decoupled the bundle that mmodernisation theory assumed was inseparable.
HCI ≥ 70, Liberty < 60. China, Russia, Gulf states, Central Asian states, Singapore pre-reform, etc. Mean GDP: $17,600. Mean life satisfaction: 5.4/10. Gender parity: 0.90. They deliver healthcare and infrastructure but at the cost of voice, satisfaction, and equality.
Not all trajectories are the same. The 225-year dataset reveals four archetypal paths through the Liberty-Capability space. Each tells a different story about the relationship between freedom and development.
For 220 years, the US traced a textbook mmodernisation trajectory: northeast through the Liberty-Capability space, rising from L=42, HCI=35 (1800) to L=94, HCI=92 (2015). It was the model case for the theory that freedom and capability reinforce each other.
Then it snapped. Between 2020 and 2025, the US experienced the fastest horizontal collapse in the dataset: Liberty fell 46 points while HCI barely moved. The American trajectory now looks like an inverted "J" — centuries of progress followed by a sudden leftward break. The decoupling that was supposed to happen only in autocracies happened in the world's oldest democracy.
2025 position: L=48, HCI=92. The US is now a Capable Autocracy by the PTI measure.
China's trajectory is almost purely vertical. HCI rose from 19 (1949) to 86 (2025) — a gain of 67 points. Liberty moved from 4 to 5 — a gain of 1 point. No other country in history has achieved this scale of human development with so little political change.
China proves, at scale, that a state can build world-class infrastructure, achieve near-universal literacy, extend life expectancy to 78 years, and lift 800 million people out of poverty — all without political lliberalisation. The mmodernisation hypothesis fails most spectacularly here.
2025 position: L=5, HCI=86. Maximum decoupling.
South Korea represents the trajectory mmodernisation theory predicts. From Japanese colony (L=5, HCI=25) through autocratic mmodernisation under Park Chung-hee (HCI rising, L stagnant) to democratic transition in 1987 (L jumps from 35 to 72), Korea shows that capability can create pressure for freedom — but only when institutional conditions align.
2025 position: L=83, HCI=93. Diagonal success.
Venezuela demonstrates that decoupling works in reverse, too. Liberty collapsed from 72 to 8 (−64 points from peak), while HCI initially held steady and then began to decline. Once institutional capture reaches advanced stages, even capability degrades — but with a lag of 5–10 years.
2025 position: L=8, HCI=68. Decoupling downward.
Not entirely — but it needs radical revision. The South Korean path still exists. But the Chinese path exists too, and it's proving more popular. Mmodernisation theory's error was treating the bundle (freedom + capability) as indivisible. It is not. States can unbundle it, and 39 of them have.
Aid agencies have long assumed that building schools and hospitals would eventually create democratic citizens. The data suggest this is not automatic. Capability without institutional safeguards can empower autocrats as easily as it empowers citizens. Development policy must explicitly address institutional architecture, not just material outcomes.
The evidence is mixed. China's debt-to-GDP trajectory suggests authoritarian growth models hit diminishing returns. But Singapore (L=48, HCI=96) has sustained high capability for 60 years without ddemocratising. The "middle-income trap" may apply to some autocracies but not others.
The data show what Capable Autocracies don't deliver: life satisfaction (−1.1 points), gender equality (−0.09 parity gap), GDP per capita (−$18,100), and political voice (zero). The case for democracy shifts from material to intrinsic: dignity, agency, accountability, and the right to shape one's own governance. These are not measurable in GDP, but they are measurable in how people report their lives.
By the PTI measure (L=48, HCI=92), the US has already crossed into that quadrant. Even at more generous liberty estimates (L=57–70), the US is moving leftward on the scatter — towards the Capable Autocracy cluster. Whether this is a temporary perturbation or a structural shift depends on whether institutional resilience can arrest the decline before it becomes self-reinforcing.
The Great Decoupling finding was rated "Supported" by the independent thesis audit — one of only four claims to receive full confirmation. The correlation decline (0.79 to 0.57) reproduces exactly. The identification of 39 Capable Autocracies is robust to alternative HCI specifications. This is amongst the project's strongest empirical contributions.