Kuznets Waves

In the AI Story

The first Kuznets wave, associated with the British Industrial Revolution, produced severe inequality through the early nineteenth century before institutional responses — factory acts, public education, the beginnings of unionization — began moderating the distribution in the second half of the century. Full moderation took nearly a hundred years. The second wave, associated with electrification and mass production in the late nineteenth and early twentieth centuries, produced another rise in inequality that culminated in the Gilded Age and the concentration that preceded the Great Depression. The New Deal and post-war welfare-state construction eventually moderated it, producing the mid-twentieth-century compression in inequality that came to seem, retrospectively, like the natural order of capitalist economies.

The third wave, associated with globalization and information technology from roughly 1980 onward, produced the elephant curve distribution. Institutional responses have been partial and inadequate: trade adjustment assistance was chronically underfunded, labor law weakened rather than strengthened, progressive taxation eroded in most developed economies, and international coordination remained largely aspirational. The wave has not been moderated; it has reached a plateau in some dimensions while continuing to rise in others.

The AI transition constitutes a fourth wave arriving before the third has been institutionally addressed. This is structurally new. Previous waves had space between them — decades during which the institutional responses to one wave could stabilize before the next wave arrived. The AI transition began while globalization-era inequality was still rising, compounding the distributional pressure rather than replacing it. The institutional capacity built during the third wave must now address the fourth before it has finished addressing the third.

The speed compression is the defining feature. The first Kuznets wave's rise phase took six decades. Globalization's rise phase took two decades. The AI transition's rise phase is being measured in years. The institutional response time — the interval between distributional damage becoming visible and political action becoming possible — has not compressed correspondingly. Democratic legislatures operate at their historical pace; tax reform requires years; international coordination takes longer still. The gap between transition speed and response speed is wider than it has been in any previous wave.

Origin

Simon Kuznets developed the original inverted-U hypothesis in his 1955 American Economic Association presidential address, based on limited data showing rising then falling inequality during industrialization in the United States and several European economies. The hypothesis was influential for decades but came under empirical pressure as developing economies showed varied patterns that did not conform to the clean inverted-U shape.

Milanovic's extension reframes the original hypothesis as a useful idealization of a single wave while identifying the empirical pattern as a sequence of waves. The extension preserves Kuznets's institutional mechanism — that rising inequality triggers political responses that eventually moderate it — while explaining why the single inverted-U never fully completes before the next rise begins.

Key Ideas

Inequality moves in waves, not toward a steady state. Each major technological transition produces a rise; institutional responses eventually moderate it; the next wave arrives before moderation is complete.

The AI wave overlaps the globalization wave. Unlike previous transitions, AI inequality is compounding globalization inequality rather than replacing it. The institutional burden is double.

Compression of rise phases. First wave: sixty years. Globalization: two decades. AI: years. The rise is compressing faster than response capacity can compress.

Institutional moderation is not automatic. Each wave's moderation required specific political construction — labor movements, progressive taxation, welfare states — not a natural return to equilibrium. The same is true for AI.

Historical perspective as corrective. Seeing AI as the latest wave, rather than an unprecedented disruption, recenters the question on the known mechanisms — institutional construction — rather than speculation about technological exceptionalism.

Debates & Critiques

Critics argue the multi-wave framework is too generic to generate specific predictions, that each technological transition has features so distinct that averaging them obscures more than it reveals. Milanovic's response is that the framework's power lies not in the specifics of any single wave but in the identification of the recurring structural logic: technology produces rising inequality, institutions eventually respond, the response is always belated and always struggled-over. The generic pattern matters precisely because it redirects attention from the features of the current wave that seem unprecedented toward the institutional work that every wave has required.