The 10 most-cited articles in The Quarterly Journal of Economics, 2024

Abstract We study the staggered introduction of a generative AI–based conversational assistant using data from 5,172 customer-support agents. Access to AI assistance increases worker productivity, as measured by issues resolved per hour, by 15% on average, with substantial heterogeneity across workers. The effects vary significantly across different agents. Less experienced and lower-skilled workers improve both the speed and quality of their output, while the most experienced and highest-skilled workers see small gains in speed and small declines in quality. We also find evidence that AI assistance facilitates worker learning and improves English fluency, particularly among international agents. While AI systems improve with more training data, we find that the gains from AI adoption are largest for moderately rare problems, where human agents have less baseline experience but the system still has adequate training data. Finally, we provide evidence that AI assistance improves the experience of work along several dimensions: customers are more polite and less likely to ask to speak…

Abstract We answer three core questions about the hypothesized role of newly emerging job categories (“new work”) in counterbalancing the erosive effect of task-displacing automation on labor demand: what is the substantive content of new work, where does it come from, and what effect does it have on labor demand? We construct a novel database spanning eight decades of new job titles linked to U.S. Census microdata and to patent-based measures of occupations’ exposure to labor-augmenting and labor-automating innovations. The majority of current employment is in new job specialties introduced since 1940, but the locus of new-work creation has shifted from middle-paid production and clerical occupations over 1940–1980 to high-paid professional occupations and secondarily to low-paid services since 1980. New work emerges in response to technological innovations that complement the outputs of occupations and demand shocks that raise occupational demand. Innovations that automate tasks or reduce occupational demand slow new-work emergence. Although the flow of augmentation and automation…

Abstract For many decisions, we encounter relevant information over the course of days, months, or years. We consume such information in various forms, including stories (qualitative content about individual instances) and statistics (quantitative data about collections of observations). This article proposes that information type—story versus statistic—shapes selective memory. In controlled experiments, we document a pronounced story-statistic gap in memory: the average impact of statistics on beliefs fades by 73% over the course of a day, but the impact of a story fades by only 32%. Guided by a model of selective memory, we disentangle different mechanisms and document that similarity relationships drive this gap. Recall of a story increases when its qualitative content is more similar to a memory prompt. Irrelevant information in memory that is similar to the prompt, on the other hand, competes for retrieval with relevant information, impeding successful recall.

Abstract While hypothesis testing is a highly formalized activity, hypothesis generation remains largely informal. We propose a systematic procedure to generate novel hypotheses about human behavior, which uses the capacity of machine learning algorithms to notice patterns people might not. We illustrate the procedure with a concrete application: judge decisions about whom to jail. We begin with a striking fact: the defendant’s face alone matters greatly for the judge’s jailing decision. In fact, an algorithm given only the pixels in the defendant’s mug shot accounts for up to half of the predictable variation. We develop a procedure that allows human subjects to interact with this black-box algorithm to produce hypotheses about what in the face influences judge decisions. The procedure generates hypotheses that are both interpretable and novel: they are not explained by demographics (e.g., race) or existing psychology research, nor are they already known (even if tacitly) to people or experts. Though these results are specific, our procedure is general. It provides a way to produce…

Abstract Standard labor market models assume that workers hold accurate beliefs about the external wage distribution, and hence their outside options with other employers. We test this assumption by comparing German workers’ beliefs about outside options with objective benchmarks. First, we find that workers wrongly anchor their beliefs about outside options on their current wage: workers that would experience a 10% wage change if switching to their outside option only expect a 1% change. Second, workers in low-paying firms underestimate wages elsewhere. Third, in response to information about the wages of similar workers, respondents correct their beliefs about their outside options and change their job search and wage negotiation intentions. Finally, we analyze the consequences of anchoring in a simple equilibrium model. In the model, anchored beliefs keep overly pessimistic workers stuck in low-wage jobs, which gives rise to monopsony power and labor market segmentation.

This article explores the use of social signaling as a policy tool to sustainably affect childhood immunization. In a 26-month field experiment with public clinics in Sierra Leone, I introduce a verifiable signal-in the form of color-coded bracelets-given to children upon timely completion of the first four or all five required vaccinations. Signals increase parents' belief in the visibility of their actions and knowledge of other children's vaccine status. The impact of signals varies significantly with the cost and perceived benefits of the action. There are no discernible effects on timely and complete immunization when the signal is linked to an easier-to-complete vaccine with low perceived benefits, and large positive effects when the signal is linked to a costlier-to-achieve vaccine with high perceived benefits. Parents adjust their behavior nine months before realizing the social image benefit, demonstrating the motivational strength of signaling incentives. Of substantive policy importance, bracelets increase full immunization at one year of age by 9 percentage points, with…

Abstract The gender ask gap measures the extent to which women ask for lower salaries than comparable men. This article studies its role in generating wage inequality, using novel data from an online recruitment platform for full-time engineering jobs: Hired.com. To use the platform, job candidates must post an ask salary, stating how much they want to make in their next job. Firms then apply to candidates by offering them a bid salary, solely based on the candidate’s résumé and ask salary. If the candidate is hired, a final salary is recorded. After adjusting for résumé characteristics, the ask gap is 2.9%, the bid gap is 2.2%, and the final offer gap is 1.4%. Further controlling for the ask salary explains the entirety of the residual gender gaps in bid and final salaries. To further provide evidence of the causal effect of the ask salary on the bid salary, I exploit an unanticipated change in how candidates were prompted to provide their ask. For some candidates in mid-2018, the answer box used to solicit the ask salary was changed from an empty field to an entry prefilled with…

Abstract Satoshi Nakamoto (2008) invented a new kind of economic system that does not need the support of government or rule of law. Trust and security instead arise from a combination of cryptography and economic incentives, all in a completely anonymous and decentralized system. This article shows that Nakamoto’s novel form of trust, while undeniably ingenious, is deeply economically limited. The core argument is three equations. A zero-profit condition on the quantity of honest blockchain “trust support” (work, stake, etc.) and an incentive-compatibility condition on the system’s security against majority attack (the Achilles heel of all forms of permissionless consensus) together imply an equilibrium constraint, which says that the “flow” cost of blockchain trust has to be large at all times relative to the benefits of attacking the system. This is extremely expensive relative to traditional forms of trust and scales linearly with the value of attack. In scenarios that represent Nakamoto trust becoming a more significant part of the global financial system, the cost of trust…

Abstract Frictions that impede the mobility of workers across occupations and space are a prominent feature of developing countries. We disentangle the role of insecure property rights from other labor-mobility frictions for the reallocation of labor from agriculture to nonagriculture and from rural to urban areas. We combine rich household and individual-level panel data from China and an equilibrium quantitative framework featuring sorting of workers across locations and occupations. We explicitly model the farming household and the endogenous decisions of who operates the family farm and who potentially migrates, capturing an additional channel of selection in the household. We find that land insecurity has substantial negative effects on agricultural productivity and structural change, raising the share of rural households operating farms by over 40 percentage points and depressing agricultural productivity by more than 20%. Comparatively, these quantitative effects are as large as those from all residual labor-mobility frictions. We measure a sharp reduction in overall…

Abstract Many consumers care about climate change and other externalities associated with their purchases. We analyze the behavior and market effects of such “socially responsible consumers” in three parts. First, we develop a flexible theoretical framework to study competitive equilibria with rational consequentialist consumers. In violation of price taking, equilibrium feedback nontrivially dampens the impact of a person’s consumption on aggregate consumption, undermining the motive to mitigate. This leads to a new type of market failure, where even consumers who fully “internalize the externality” overconsume externality-generating goods. At the same time, socially responsible consumers change the relative effectiveness of taxes, caps, and other policies in lowering the externality. Second, since consumer beliefs about and preferences over their market impacts play a crucial role in our framework, we investigate them empirically via a tailored survey. Consistent with our model, consumers are often consequentialist, and many believe that they have a dampened impact on aggregate…