I document a new empirical puzzle in initial public offerings (IPOs): trading volume over the 600 business days following an IPO exhibits a pronounced U-shaped pattern. After the well-known post-IPO drop in turnover, trading activity rises steadily by about 22% per year. This increase cannot be explained by standard drivers of high turnover and is robust to aggregate market trends and to firm size at issuance. I further show that the U-shape is closely linked to lockup agreements, but not through their direct consequences such as insider sales or changes in free float. Instead, the results suggest that lockups shape long-run liquidity dynamics, pointing to persistent effects from investor anticipation, underwriter reputation, or shifts in trading clientele. These findings link IPO contract design to secondary market liquidity and open new questions for asset pricing and market microstructure.

This paper studies whether probability weighting can jointly account for three features of household retirement behavior: low wealth accumulation, limited risky-asset exposure, and low annuity demand. I embed rank-dependent utility in a calibrated life-cycle model with laborincome, equity, and mortality risk and an annuitization choice at retirement. Probability weighting of survival risk shifts consumption toward earlier ages and approximately halves retirement wealth. Probability weighting of labor-income and equity risk reduces risky-asset exposure at the benchmark calibration, while survival weighting works in the opposite direction through the composition of human and financial wealth. The largest effect is on annuitization: among households reaching retirement, demand for an actuarially fair annuity falls from 98% to 5%. These effects are robust across alternative calibrations. LISS survival beliefs display the same qualitative pattern as probability-weighted survival probabilities, helping reconcile longevity optimism with low annuity demand.

Financial losses can misidentify which generations bear the largest economic costs of disasters. I study this distinction in a life-cycle portfolio-choice model in which disasters affect wealth, labor income, and risk aversion. Under CRRA preferences, financial losses are concentrated at middle and older ages, whereas remaining-lifetime welfare losses peak for disasters occurring early in life at approximately 13-14%. Preference changes matter mainly through the behavioral responses they induce and their interaction with financial shocks. Separating risk aversion from intertemporal substitution using Epstein-Zin preferences shifts the largest welfare losses toward retirement, because changes in risk aversion no longer simultaneously alter intertemporal substitution. The results show that financial and welfare losses can imply sharply different intergenerational rankings.

Do preference measures become more informative when elicited in the same context as the behavior they are meant to explain? Using a representative Dutch household panel, we compare risk-and time-preference tasks framed as lotteries, investments, work, or health and link them to independently observed field behavior. Context matters sharply for risk but not for time preferences. Domain-matched risk aversion predicts risky investment, self-employment, and insurance choices, whereas lottery risk aversion does not. Lottery-context impatience, however, already predicts wealth, financial resilience, and exercise. In pooled tests, matching strengthens the risk-preference slope by 0.21 outcome standard deviations more than the time-preference slope (p = 0.006). More broadly, our results show that the portability of preference measures across domains depends on the preference being measured.

Risk preference elicitation methods are widely used to measure behavior under risk, but typically interpret choices through the lens of expected utility. When actual preferences depart from expected utility, the resulting estimates may therefore be distorted. We characterize the resulting measurement error. In the Holt-Laury multiple price list, probability weighting can either raise or lower estimated risk aversion, with the sign determined jointly by underlying risk aversion and task design. The error persists with a finer probability grid and appears in the fixed-probability Eckel-Grossman task. We then show how task design can mitigate this error under rank-dependent utility. Setting the probability to 37% eliminates the probability-weighting component under one-parameter Prelec weighting and reduces it under Tversky-Kahneman weighting.