New paper on bank runs with Correia and Luck: "Bank Runs With and Without Bank Failure" Questions: - What are the determinants of runs? - When do bank runs result in bank failure? - Can runs trigger the failure of healthy banks and amplify small shocks into large crises? - Are runs themselves the initial cause of financial distress or are they a symptom of deeper fundamental solvency problems in the financial system? What we do: - Apply LLMs to historical newspapers to uncover over 4,000 runs on individual banks in the pre-FDIC US banking system from 1863 to 1934. Capture the most famous runs (Bank of the US - Merge data on runs and other bank-level events discussed in newspapers (suspensions, failures) to bank-level fundamentals (harder than it sounds!) What we find: (1) Runs are considerably more likely in weak banks, but can also occur in strong banks, especially in response to negative news about the real economy or the broader banking system. (2) However, runs typically only result in failure for banks with weak fundamentals [see figure below]. Strong banks survive runs through various mechanisms, including interbank cooperation, equity injections, public signals of strength, and suspension of convertibility (3) At the local level, poor fundamentals necessary for runs to translate into large declines in lending. Moreover, bank failures (with and without runs) translate into substantially larger declines in deposits and lending than runs without failures. Overall takeaways: - Poor fundamentals are key for whether runs pass through into failure and have severe consequences for the broader economy. - The findings temper the view that small shocks can result in large jumps to bad equilibria via runs on demandable debt. Full paper here. Comments welcome. Given the methodology and evolving AI tools, we expect to make refinements to the runs database over time. Any input is welcome. static1.squarespace.com/stat…

I think Ezra (inadvertently) proves Jenson’s point here. The reason financial firms behaved so irresponsibly pre-2008 is not because liability incentives failed to check them. It is because we diluted those incentives for decades, through growing implicit & explicit guarantees.

India produced 11% of the world’s tomatoes in 2022–24, but accounted for just 0.2% of processed tomato exports by volume. Across many crops, agricultural abundance has not translated into processing strength. What needs to change? 1/8

In case you're excited about labor economics and are considering a predoc: please come and work with my wonderful colleagues and me at the @Princeton @PrincetonEcon Industrial Relations Section. Please spread the word! 🙏👇 irs.princeton.edu/senior-res…

Excellent piece from Roland Fryer: "The failure isn’t that some families can choose a good school, but that too many can’t." wsj.com/opinion/the-educatio…

there is none. my gut feeling - and its not much more than that now - is that anthropic and openai will go belly up, and the current period will join tulips, south sea, 1929 and dotcom in the history books...

Even with 77% annual revenue growth, Open AI expects $ 278bn (!) of cumulative negative free cash flow from 26 to 30. Big part is compute and infrastructure spending. What is the path to profitability when these models are becoming commoditized? ft.com/content/6011d061-eee3…

Same

Agree... the Nikole Hannah-Jones piece is a bizarre combination of empirical naivete (bordering on arrogance) about how one family could improve an underperforming school, with a willingness to put social ideals ahead of your own child. More broadly, a striking example of the limits of a worldview that relies on individual self-sacrifice to solve a structural collective action problem.

Always show the treatment/control averages, not just the dynamic DD estimates! Example #104