Research
Work in Progress
The Bank Relationship Channel in Germany’s Great Depression
Banking failures are central to accounts of Weimar Germany’s economic collapse, yet the proposition that the destruction of specific bank relationships transmitted distress to firms has never been tested at the firm level. This paper constructs the first such test, linking approximately 2,900 industrial firms to their pre-crisis banks through paying-agent relationships recorded in the Handbuch der Deutschen Aktien-Gesellschaften and extracted via large language models. The fraud at Nordwolle, Danat’s largest single debtor, triggered a run that Danat’s thin capital base could not absorb. Clients of the failed bank faced distress rates 5 percentage points above comparable firms, a 50 percent increase on the 10 percent baseline. Firms with greater short-term funding exposure were disproportionately affected, consistent with credit disruption rather than selection. Clients of Dresdner Bank, which was recapitalised rather than closed, show no elevated distress. The relationship-specific credit channel was real and, for the firms it reached, economically significant, but because Danat served only a fraction of German industry, its aggregate contribution to corporate distress was small.
Questionable Research Ideas:
Here is a list of bad economics research ideas I have from brain-fuzz and drawing from daily inspirations from random encounters of papers: they are often challenging, undesirable, and difficult to research but nonetheless interesting. If you are an economics or economic history researcher: you would be more than welcome to take look!
If you spot potential in any of these ideas - perhaps you know of relevant datasets, can suggest improved empirical strategies, or see ways to make them more feasible - I'd love to hear from you. I'm also open to potential collaboration. Feel free to reach out at sihao.feng@stcatz.ox.ac.uk
Institutional Narratives in the Art Market
Claude’s Research Proposal.
How does qualitative, socially produced information enter prices in markets characterized by extreme heterogeneity, illiquidity, and limited arbitrage? This very preliminary and speculative project approaches the question empirically by constructing original time-series indices that track the evolving endorsement of art categories by institutional arbiters of the art world such as museums, exhibition organizers, and scholars over the past century in the North American and European context. The central exercise investigates whether observable shifts in this institutional attention precede movements in category-level art prices, consistent with slow information diffusion in a market where fundamental value is contested and processing costs are high. The paper documents preliminary patterns that may speak to broader questions about price formation under substantial uncertainty, although I remain agnostic about the specific equilibrium mechanism at work.
St. Catherine’s College, University of Oxford
The Barbican