Lars Lochstoer, University of California-Los Angeles
Stig Roar Lundeby, BI Norwegian Business School
Zhaneta Tancheva, BI Norwegian Business School
Abstract: Recent evidence in the psychology literature suggests that individuals’ degree of present bias is time-varying and increases under stress. We first document, using survey data on individuals’ expected and realized consumption, predictability in their own consumption forecast errors consistent with this notion. Next, we consider an asset-pricing model where a subset of investors has a time-varying degree of present bias. Their presence causes substantial priced discount-rate risk that has first-order effects on the level and time variation of asset risk premia. The mechanism is distinct from models with time-varying preference parameters and from models with biased expectations about aggregate outcomes.
Discussant: Andrea Buffa, University of Colorado-Boulder
Abstract: This paper resolves a long-standing zero-beta rate puzzle—the empirical finding that estimated zero-beta rates remain persistently high across factor models. I show that this apparent robustness may arise from pervasive model misspecification rather than reflecting a genuinely high risk-free rate. When a factor model fails to perfectly price assets, the corresponding zero-beta rate is no longer uniquely identified, and conventional estimators, based on the minimum-variance zero-beta portfolio, tend to bias the estimate upward toward the mean return of the global minimum-variance portfolio. To quantify this mechanism, I introduce a new investment-based measure of model misspecification: the maximum Sharpe ratio attainable by zero-investment, zero-beta portfolios. This measure captures the economic magnitude of pricing errors and links model misspecification to empirically observable investment opportunities. Studying a comprehensive set of classical and modern factor models, I find substantial misspecification, explaining why all models yield similarly elevated zero-beta rates. Simulation analyses confirm that realistic degrees of misspecification can fully reproduce the empirical magnitude of the puzzle even when the true risk-free rate is low.
Discussant: Ricardo Delao, University of Southern California
Abstract: Financial event studies, ubiquitous in finance research, typically use linear factor models with known factors to estimate abnormal returns and identify causal effects of information events. This paper demonstrates that when factor models are misspecified—an almost certain reality—traditional event study estimators can produce inconsistent estimates of treatment effects. The bias is particularly severe during volatile periods, over long horizons, and when event timing correlates with market conditions. We derive precise conditions for identification and analytic expressions for asymptotic bias. As an alternative estimation approach, we propose synthetic control methods that construct replicating portfolios from control securities without imposing specific factor structures. Revisiting four empirical applications, we show that some established findings — the Geithner Treasury Secretary announcement effect on banks’ stock prices (Acemoglu et al.,2016), pre-inclusion drift for index inclusion, and M&A acquirer effects — may reflect model misspecification rather than true treatment effects. While traditional methods remain reliable for studies with random event timing, our results suggest caution when interpreting event studies where the events themselves may be correlated with market performance.
Discussant: Sofonias Alemu Korsaye, Johns Hopkins University
