These seminars are sponsored by the Mitsui Life Financial Research Center.
If you would like to be added to the email distribution list, please contact Gabriella Ring at [email protected].
mMrch 20
Sean Higgins, Northwestern
Title: Mental Accounting and the Marginal Propensity to Borrow: Evidence from a Large-Scale Credit Limit Experiment
Abstract: Why do unconstrained consumers respond to credit limit increases? Existing experiments find evidence for various mechanisms, including precautionary motives and interpreting the limit increase as a signal about future income. Mental accounting, or the idea that consumers separate financial decisions based on how money is categorized, may play a role, but this hypothesis has not been tested experimentally. We conduct a randomized controlled trial (RCT) with Cashea, a Venezuelan buy now, pay later (BNPL) provider with two distinct credit lines: one for food and drugstore purchases, and one for clothing, electronics, furniture, and other retail goods. We randomized 0%, 50%, or 100% increases on one or both credit lines for 210,000 active users. As predicted by a model of mental accounting, there is a large category-specific marginal propensity to borrow out of credit line increases of 32--42%, with limited or no spillovers across categories even for the least constrained users.
Time: 10:30-11:50 a.m.
Location: R0230
April 3
Vincent Bogousslavsky, Boston College
Title: Retail Trading and News: How News Days Drives Retail Losses
Abstract: We study how news events shape retail trading and performance in the era of trading apps and zero-commission brokers. Using trader-level data, we show that retail stock trading doubles on news days. This news-driven trading explains all retail stock trading losses in our sample. On news days, retail investors lose an additional 13 basis points per trade relative to non-news days. The evidence is consistent with retail investors overreacting to good news and underreacting to bad news, to the benefit of better-informed traders. Retail losses increase with short-selling activity, but this relation is concentrated on news days. Losses are particularly severe in equity options and disappear in index ETFs on macro news days, when private information is limited. We find no evidence that retail investors learn to avoid this behavior. Overall, we identify news-driven trading as a major source of retail underperformance in today’s zero-commission, media-saturated trading environment.
Time: 10:30-11:50 a.m.
Location: B1590 Corner Commons
April 10
Song Ma, Yale
Title: Putting Marginal Back in 𝑞
Abstract: How to explain corporate investment is a central question in economics and finance. Theory provides an elegant answer: marginal 𝑞, the shadow price of capital, is a sufficient statistic for the investment decision. But marginal 𝑞 is unobservable. For fifty years, empirical work has substituted average 𝑄, a proxy constructed from standard accounting data, and valid only under constant returns to scale, perfect competition, and homogeneous adjustment costs. These assumptions usually fail, and the average𝑄 is a weak predictor of investment. It remains unclear whether the problem is the theory or the measurement. How about we identify what the marginal project of a firm is and construct marginal 𝑞 accordingly? We take a bottom-up approach, leveraging text and modern AI tools to put marginal back in 𝑞. Firms produce vast text—filings, earnings calls, patents, news—encoding forward-looking information about investment opportunities that accounting numbers miss. For each firm-year, we use retrieval-augmented generation (RAG) to assemble the most relevant textual evidence under a best-effort no-look-ahead constraint, and prompt a large language model to identify the firm’s most plausible next projects along with their market values and costs. The resulting measure, 𝑞𝐴𝐼, substantially outperforms average q in predicting investment. Because our method identifies the marginal project, not just its value 𝑞, it opens questions that existing 𝑞 measures cannot ask. The investment-𝑞 elasticity is larger when the marginal project is compatible with existing operations and smaller for acquisition-driven expansion, consistent with reallocation frictions compressing extensive-margin investment. Decomposing the wedge between marginal 𝑞 and average 𝑄, we show that measurement error, capitalized rents, and omitted intangibles each contribute, with relative magnitudes varying across industries.
Time: 10:30-11:50 a.m.
Location: R0230
April 17
Richard Townsend, UC San Diego
Title: Interest Rate Predictability Confusion
Abstract: When the Fed signals that rate hikes are planned, many borrowers rush to get long-term loans before rates “go up.” We show that this widespread intuition is incorrect and can weaken the effectiveness of predictable monetary policy. Long-term rates are forward-looking, so information about future short-term policy rates should be priced immediately. Empirically, long rates do not positively comove with expected changes in policy rates. Yet professional forecasters predict that long-term rates will track the expected path of the policy rate, generating large forecast errors. Strikingly, forecasters mistakenly predict that the path of the long rate over the next four quarters will match the shape of the expected policy-rate path, including reversals and curvature. House-holds exhibit the same confusion, most strongly among those with high education and income. This misconception distorts real behavior. When the policy rate is expected to increase, households and firms rush to lock in long-term debt before rates rise further, undermining the contractionary goals of monetary tightening. Conversely, when the policy rate is expected to fall, borrowers delay long-term borrowing to wait for lower rates, undermining the expansionary goals of monetary loosening.
Time: 10:30-11:50 a.m.
Location: R0230
April 22
Cecilia Parlatore, NYU Stern
Title: Information Span in Credit Market Competition
Abstract: Recent technological change in lending converts previously subjective assessments into structured, easily accessible data. We study this transformation in a credit market competition model that distinguishes between information span (breadth) and signal precision (quality). Borrower quality depends on multidimensional fundamentals, assessed through hard or soft signals. Two banks observe private hard signals, but only the specialized bank receives a soft signal. Expanding the span of hard information allows the non-specialized bank to evaluate characteristics previously only available to the specialist, and reducing its winners curse. By contrast, greater precision of hard signals strengthens the specialized banks informational advantage.
Time: 10:30-11:50 a.m.
Location: B3560