Globalization and Its Discontents

In the AI Story

The book's core empirical content is a detailed reconstruction of the 1997 Asian financial crisis, the 1998 Russian financial crisis, and related episodes in Latin America and Africa. In each case, Stiglitz demonstrates that IMF policies — rapid capital-account liberalization, fiscal austerity during downturns, privatization of state enterprises — produced worse outcomes than alternative policies that the institution rejected on ideological grounds. The Asian crisis, in particular, was not a consequence of the policies that caused it but was exacerbated by the policies imposed in response: austerity during a contraction deepened the contraction, rapid capital-account opening had enabled the crisis in the first place, and the structural conditions attached to assistance pushed economic and political systems in directions the populations had not chosen.

The analytical framework generalizes beyond the specific cases. Markets function well under specific conditions — adequate competition, symmetric information, robust institutional enforcement, democratic accountability — and poorly when these conditions are absent. The IMF policies assumed the conditions; the reality of developing economies was that they were absent. Imposing market-based solutions without building the institutional infrastructure that makes markets work produced predictable pathologies: concentration of wealth, capture of newly privatized assets by those with political connections, destruction of social safety nets that had been maintaining stability, and political backlash that undermined the democratic institutions the policies were theoretically intended to support.

Applied to AI governance, the framework identifies a structurally analogous pattern. The policies being applied to AI — whether the EU AI Act's risk classifications, the American voluntary commitments, or the voluntary self-regulation of AI companies — are being designed by technocrats and industry representatives without substantive democratic input from the populations that bear the costs. The information asymmetries are severe: the populations most affected by AI deployment understand the technology least. The power concentrations are extreme: a dozen companies control the frontier, and they are represented in governance discussions in ways that affected populations are not. The distributional consequences are predictable: gains concentrate among the powerful, costs flow to the less-powerful, and the governance frameworks produced by the technocratic process preserve rather than correct the pattern.

The book's influence extends well beyond development economics. It established the template for critical engagement with technocratic governance that subsequent scholars and activists have applied to trade policy, financial regulation, climate policy, and now AI governance. Stiglitz's specific policy recommendations — greater democratic participation in international economic institutions, recognition that market-based policies require institutional prerequisites, transitional support for populations bearing the costs of policy changes — transfer directly to the AI context. The underlying argument, that legitimate governance requires affected populations to have meaningful voice in the decisions that shape their lives, applies to AI with particular force because the technology reshapes those lives at unprecedented speed and scale.

Origin

Stiglitz wrote Globalization and Its Discontents in the years after leaving the World Bank in 2000, publishing it in 2002. The book emerged from direct institutional experience: Stiglitz had chaired the Council of Economic Advisers under Clinton (1995–1997) before moving to the World Bank, where he increasingly diverged from the institutional consensus on structural adjustment. His public criticism, amplified by the book, made him one of the most prominent dissenters within the institutions of global economic governance — a position he has maintained through subsequent work on globalization, inequality, and now AI.

Key Ideas

Technocratic governance without legitimacy produces backlash. Policies imposed on populations without their input generate resistance that undermines both the specific policies and the institutional credibility of the imposing bodies.

Markets require institutional prerequisites. Competition, information symmetry, enforcement, and democratic accountability are not optional features of functioning markets but conditions of their function.

One-size-fits-all policies fit no one. Developing economies vary in institutional structure, historical development, and political capacity in ways that uniform policies cannot accommodate.

Concentrated power captures reform. Policies designed to open markets often enable the capture of newly privatized assets by those with political connections, deepening rather than correcting concentration.

The AI parallel. The same technocratic-governance-without-democratic-input pattern is currently being applied to AI policy, with predictable distributional consequences.

Debates & Critiques

The book generated enormous controversy within the international financial institutions and mainstream economics. Critics — including Stanley Fischer, the IMF's then deputy managing director — argued that Stiglitz misrepresented specific cases and ignored the constraints under which the institutions operated. Stiglitz's response acknowledged specific disagreements while maintaining that the central analytical framework was correct: the institutions had applied ideologically driven one-size-fits-all policies with insufficient attention to local conditions and inadequate democratic legitimation, and the consequences were predictable.