Navigating the tax system often requires information and judgment that taxpayers may lack. Although taxpayers sometimes rely on human experts, AI could provide comparable assistance at substantially ...

More than 2,700 researchers, hailing from 39 countries, traveled to Cambridge, MA for the 49th annual NBER Summer Institute in mid-July. The Summer Institute included 49 distinct workshops arranged by ...

We use a sample of administrative payroll data covering a large and nationally representative share of U.S. workers to study how wages adjusted during the recent inflation period. Most firms apply a ...

We examine how prenatal exposure to air pollution interacts with gender-biased parental investments to shape long-run cognitive performance. Using nationally representative survey data from China, we ...

We document that firms with high price-earnings ratios tend to see increases in their subsequent earnings relative to sales, which we interpret as rents from ideas (innovation). We construct an ...

Generative AI has been promoted as the technology that could transform education by providing every student a personal tutor. We provide some of the first large-scale experimental evidence, from a two ...

What ultimately supports the market value of U.S. government debt? We measure valuation frameworks: the perceived channels that back debt value. We survey bond investors, registered voters, and ...

Ten states implemented Supplemental Nutrition Assistance Program (SNAP) Food Restriction Waivers to restrict purchases of sugary beverages with SNAP funds for the first time in program history in 2026 ...

The existence of aggregate welfare gains from trade is one of the classic insights from international economics. However, these conventional welfare gains are static, in the sense that they take ...

Macroeconomics tends to view exchange rate movements as transitory. However, persistent exchange rate realignments and the associated capital flows can have long-run implications for structural change ...

Large language models (LLMs) can make tutoring more scalable, but only if students use them to reason through mistakes rather than avoid effort. We study this question in a randomized field experiment ...

We use newly digitized records for 1.2 million students at 60 co-educational U.S. colleges, 1915–1943, linked to enrollment records for their children through 1963, to study how colleges shape marital ...