We present the Armenian translation of the article The world’s most influential economist is oddly unconvincing from The Economist.
When he was called from Stockholm, it seemed that the call was delayed. In 2024, Daron Acemoglu, a researcher at the Massachusetts Institute of Technology, shared the Nobel Prize in Economics with Simon Johnson and James Robinson for explaining how institutions shape prosperity. Acemoglu is 58 years old but has been considered a giant in his field for some time. He has written seven books, including a new one about the economic impact of democracy and artificial intelligence, not to mention several well-known textbooks. His hundreds of articles, more than ten of which were published this year, are often filled with complex mathematics. He competes with Harvard University's Andrei Shleifer for the top spot in the IDEAS/rePEc database of economists' citations. His opinion carries significant weight among other experts and media.
However, if economists drink a few glasses, some of them will dare to express their true opinion about this giant. "Most of his theoretical work is useful, but he uses his models to justify previously tested and failed populist policies," criticizes one well-known economist. Some commentators don't even need alcohol. "I've literally been shouting about Acemoglu for a decade," says economist-blogger Noah Smith in response to the wave of online criticism of Mr. Acemoglu's work.
No one accuses the prolific economist of poor research, let alone academic negligence. "He is obviously a genius," says another, also noting that he is generous to doctoral students and junior colleagues. The question is whether Mr. Acemoglu's position at the pinnacle of the economics profession (and the accompanying intellectual influence) is justified by his scientific work.
Let's start with his empirical methods. The basis for his Nobel Prize is an article co-authored with Johnson and Robinson published in 2001, which has been cited more than 23,000 times. It attempts to explain why some countries are rich while others are poor. In places where European colonizers died en masse, possibly due to diseases, the colonizers built "extractive" institutions, concentrating power and resources in the hands of a few elite groups to extract as many resources as possible from the region. However, in more favorable climatic conditions, colonizers survived and built more inclusive institutions with proper roads, schools, and medical services. These institutions persisted, resulting in areas with low mortality rates among colonizers during the colonial period being richer today.
Photo: REUTERS
Economists now take these results seriously, but not literally. The article encouraged experts to think more deeply about institutions, but they subjected it to thorough analysis. In the 2024 report, the Nobel Committee acknowledged that mortality data are "sometimes inaccurate." In 2012, David Albouin from the University of Illinois at Urbana-Champaign showed that mortality rates in some countries were borrowed from other countries. He found that if they were adjusted, the conclusions of the article would become unreliable. Last year, Martin Buchner and his colleagues from the RWI research institute in Essen found that experts participating in their survey tended to support Mr. Albouin.
Mr. Acemoglu says that "these discussions and some of the criticism are very valuable." However, he counters that Mr. Albouin reaches his conclusion by ignoring half of the data from the original sample, including important countries like America, Canada, and Australia. It is this combination, as well as some selective statistics, that leads to unreliability, he says.
Acemoglu adds that if he were to revisit his work today, he would make "a number of changes, including in some estimates," but not in mortality data.
Recently, researchers examined a work by Mr. Acemoglu and several co-authors published in June, which stated that low birth rates are associated with faster GDP growth per working-age adult. This is a stunning finding that suggests that areas losing population, such as Japan, need not worry. However, other economists, including Jesús Fernández-Villaverde from the University of Pennsylvania, question whether historical connections are useful for understanding the present when many countries' birth rates are declining. This is a perfectly reasonable viewpoint, and Mr. Acemoglu and his co-authors briefly mention this in their conclusions. However, this mention may get lost and seem insignificant compared to the vast amount of mathematical calculations by Acemoglu.
Careful scrutiny can allow a determined critic to find shortcomings in any research work, and Mr. Acemoglu is the biggest target in the field of economics. However, the thorough examination associated with his fame cannot explain the second problem: that Mr. Acemoglu's ideas may not seem so impressive for a titan of social sciences. His public comments, especially regarding American politics, are cautious. For example, his "Trump's Unified Theory" claims that the American president "strengthens executive power and destroys limiting institutions and norms." Well, that's obvious.
Institutions and the "curve"
More troubling is that his grand thesis about institutions reveals little. Nations prosper when institutions are good and stagnate when they are bad. This is true. But what is an institution, really? Rules, norms, application, culture — everything. Where do institutions come from? "Critical junctures" and "institutional deviations," whatever that means.
Daron AcemogluIn reviewing one of Acemoglu's books in 2011, Tyler Cowen from George Mason University noted that the institutional changes described in it stem only from other institutional changes, which makes the main argument an eternal regression, like a curve.
Duncan Green from the London School of Economics argues that this scientific viewpoint mainly works retrospectively. Francis Fukuyama from Stanford University says that the book ignores the example of China, which has experienced rapid economic growth with institutions that can be characterized as "extractive."
Mr. Acemoglu counters that two chapters of the book are partially or largely devoted to China. He also notes that "to understand institutional changes from an institutional perspective, one must consider the specific institutions of the past and the historical events that influenced them...". "I don't think this is a tautology," he says.
Daron AcemogluAcemoglu diverges from accepted views to defend his bleak views on technology. The book "Power and Progress," which he co-authored with Johnson in 2023, mainly views a thousand years of innovation as a history of power grabs by elites and unintended consequences. The cotton-cleaning machine was a servant of slavery, and the Haber-Bosch technology enabled weapon production. Ordinary people apparently benefited only when they could compel capitalists to consider their interests. The book underestimates a much more compelling argument: that technological progress itself has enriched ordinary people many times more than in pre-industrial times.
Mr. Acemoglu's arguments about AI are so pessimistic that they are even unconvincing. In a 2024 article, he calculates that AI will raise American productivity only slightly over the next ten years. This article comes to this conclusion by ignoring the transformative impact of new AI products on the market. "In this analysis, he completely excludes the possibility that AI will lead to faster scientific progress, faster socio-scientific progress, or better decision-making," says Lawrence Summers, the honorary president of Harvard University.
"This is a valid criticism," says Mr. Acemoglu. "I did not predict agent-based AI, and it was not included in my forecasts." He also notes that AI models may be more useful in scientific research than he had envisioned. However, Acemoglu insists that his methods for assessing the impact of AI on productivity are correct and "offer some realism in claims about huge productivity improvements."
These academic disputes may seem trivial, but they could ultimately be significant. Mr. Acemoglu does not believe that AI will sharply increase productivity but is concerned about its impact on democracy and jobs. His new book discusses the potential social harms of technology (deepening atomization, spreading political falsifications), rather than its possible benefits. His proposal to limit these harms — "labor-intensive AI," which would increase the value of human labor rather than replace it, is excellent... and obvious.
Mr. Acemoglu's influence in the AI debate is clearly seen in a recent statement signed by dozens of prominent economists, in which they argue that "we must act right now" and "AI should be directed to complement human activity and be beneficial to society." Who is really "we"? The Trump administration? And who should decide which type of AI complements people and which does not? Even the economists who signed the statement are not sure. Mr. Acemoglu's star is so bright that it can sometimes blind critical thinking.
