Rethinking My Economics (2024 IMF essay)
Rethinking My Economics is the remarkable 2024 essay in which Angus Deaton, a Nobel laureate then nearly sixty years into his career, published in the IMF's Finance & Development what amounted to a public reckoning with mainstream economics. The essay opened with the directness of a scientist reporting a negative result: 'I have changed my mind.' What Deaton had changed his mind about was the capacity of the discipline he had practiced and advanced for half a century to predict and account for the distributional consequences of the policies it had championed. Globalization had produced aggregate gains while devastating specific communities. Market liberalization had increased efficiency while concentrating power. Technological change had raised productivity while hollowing out livelihoods.
The Capture Hypothesis — Contrarian ^ Opus
There is a parallel reading of Deaton's reconsideration that begins not with his analytical courage but with the institutional mechanics that produce such conversions. The IMF, having overseen decades of structural adjustment programs that immiserated millions while enriching connected elites, now publishes essays acknowledging that power matters. This is less intellectual evolution than institutional adaptation—the same organizations that enforced the Washington Consensus now rebrand themselves as critics of inequality precisely as popular resistance to their prescriptions reaches critical mass. Deaton's essay, in this reading, represents not a breakthrough but a managed retreat, acknowledging just enough of power's role to preserve the fundamental legitimacy of economic expertise while continuing to shape policy from above.
The essay's treatment of AI reveals this dynamic most clearly. By framing artificial intelligence as an intensification of existing concentration rather than a qualitatively different form of control, Deaton maintains the fiction that traditional economic tools—properly adjusted for power—can still guide us through the transition. But AI's capacity for behavioral prediction and manipulation, its integration with surveillance infrastructure, and its role in automating not just production but governance itself represents something beyond market concentration. It represents the emergence of what might be called algorithmic feudalism—a system where a handful of firms don't just dominate markets but architect the cognitive and social environments within which human agency itself operates. The call for unions to be 'at the table' assumes there will still be a table, that the forms of collective bargaining developed for industrial capitalism can constrain entities whose power operates through the manipulation of attention, desire, and possibility itself.
In the AI Story
The essay identified a specific analytical failure: mainstream economics had underestimated the role of power in shaping who benefits from economic change. The models assumed competition meaningful enough that workers could move to better jobs and consumers could choose among alternatives. The reality, which decades of evidence had made undeniable, was that power had concentrated in ways that rendered the competitive assumptions false. A small number of firms dominated entire industries. The political process, supposed to correct market failures, had itself been captured by the interests it was meant to regulate.
The essay contained a brief but analytically consequential passage on artificial intelligence. Deaton observed that 'the continuing rapid development of artificial intelligence means that this technological transition will endure,' and placed the observation in the context of 'pervasive economies of scale more powerful than older industries,' in which a small number of firms wield significant market power and pinpointing value creation has become 'next to impossible.' The passage was not a prediction about AI's capabilities. It was a diagnosis of AI's distributional structure.
The essay endorsed the argument of Daron Acemoglu and Simon Johnson in Power and Progress that 'the direction of technical change has always depended on who has the power to decide' and that 'unions need to be at the table for decisions about artificial intelligence.' The endorsement was significant because Deaton had spent much of his career in a professional milieu skeptical of both claims. The reconsideration represented a convergence between Deaton's empirical commitments and a political-economy framework his discipline had historically marginalized.
The essay is the intellectual hinge on which the current Deaton volume turns. It provides the author's own words on AI, his own framing of the distributional question, and his own acknowledgment that the analytical tools he had spent a career developing required augmentation to address the transition underway. The essay is short — a few thousand words — but it is the most direct application of Deaton's mature thinking to the AI moment, and it is the text the current book most consistently returns to.
Origin
The essay was published in the March 2024 issue of the IMF's Finance & Development magazine, a publication aimed at policymakers and development practitioners. It followed Deaton's 2023 book Economics in America, which had developed similar themes in greater length and with more empirical detail.
Key Ideas
Economics has underestimated power. The competitive assumptions of mainstream models fail to account for the concentration of corporate and political power that shapes distributional outcomes.
AI will intensify these dynamics. The economies of scale in AI development are more powerful than in older industries, producing more extreme concentration.
Unions must be at the table. The direction of technical change depends on who has the power to decide, and countervailing institutions are necessary for broadly shared outcomes.
Pinpointing value creation is next to impossible. The structure of AI-driven production makes traditional tools for attributing value to specific contributions inadequate.
The transition will endure. AI is not a cyclical phenomenon but a structural transformation requiring institutional responses of matching magnitude.
Appears in the Orange Pill Cycle
Institutional Evolution Versus Capture — Arbitrator ^ Opus
The tension between viewing Deaton's reconsideration as genuine intellectual growth versus institutional adaptation depends fundamentally on which temporal frame we adopt. Looking at the immediate question of whether mainstream economics can self-correct, the entry's optimistic reading holds considerable weight (70/30)—Deaton's willingness to publicly revise decades of work represents real epistemic humility rarely seen in Nobel laureates. His specific acknowledgment that power shapes distributional outcomes marks a substantive shift in how economic analysis might proceed. The profession's response to his essay, with younger economists increasingly incorporating political economy frameworks, suggests this isn't merely symbolic.
Yet when we shift to examining the structural dynamics of AI deployment, the contrarian's darker reading gains force (20/80). The scale and speed of AI-driven transformation does appear to exceed the adaptive capacity of existing institutions, including reformed economics departments and strengthened unions. The contrarian correctly identifies that AI's power operates through cognitive and behavioral channels that twentieth-century countervailing institutions weren't designed to address. Market concentration is one thing; the ability to shape preference formation itself is another. Here, Deaton's framework, even adjusted for power, may indeed prove inadequate.
The synthetic insight emerges when we recognize that both institutional learning and institutional capture can occur simultaneously—and that AI intensifies both dynamics. Deaton's reconsideration represents genuine progress within economics as a discipline (this much is indisputable) while also serving as a managed retreat that preserves expert authority (equally true). The key question isn't whether his evolution is authentic but whether any evolution within existing institutional frameworks—however sincere—can match the velocity of AI-driven transformation. The answer likely depends on whether we can develop new forms of collective intelligence that operate at the same scale and speed as the systems they seek to govern.
Further reading
- Angus Deaton, 'Rethinking My Economics,' IMF Finance & Development (March 2024).
- Angus Deaton, Economics in America: An Immigrant Economist Explores the Land of Inequality (Princeton University Press, 2023).
- Daron Acemoglu and Simon Johnson, Power and Progress (PublicAffairs, 2023).